The 640 MEG Shareware Studio 2

home *** CD-ROM | disk | FTP | other *** search

/ The 640 MEG Shareware Studio 2 / The 640 Meg Shareware Studio CD-ROM Volume II (Data Express)(1993).ISO / prog / aitools.zip / LL1READ.ME < prev next >

Wrap

Text File | 1986-10-28 | 110KB | 2,829 lines

____________________________ | | | | | LOGIC-LINE SERIES 1 | | | | OPERATING MANUAL | | | | SHORTENED VERSION | | FOR MAXELL | | GIFT OF INTELLIGENCE | | | |____________________________| Copyright (c) 1985, 1986 Expansion Programs International, Inc. Post Office Box 839 Chesterland, Ohio 44026 USA (216) 449-6104 and Arnold Kochman ALL RIGHTS RESERVED è COPYRIGHT NOTICE Copyright(c) 1985, 1986 by Expansion Programs International, Inc. and Arnold Kochman. All Rights Reserved Worldwide. This publication has been provided pursuant to an agreement containing restrictions on its use. The pub- lication also is protected by federal copy- right law. No part of this publication may be copied or distributed, transmitted, trans- cribed, stored in a retrieval system, or translated into any human or computer language, in any form or by any means, elec- tronic, mechanical, magnetic, manual, or otherwise, or disclosed to third parties without the express written permission of Expansion Programs International, Inc., PO Box 839, Chesterland, Ohio 44026 U.S.A. LEXSET was customized for the Thunderstone LOGIC-LINE SERIES 1 program by Arnold Kochman. TRADEMARKS Thunderstone, Comprehension, Assimilation, Accelerated Information Training System (AITS), Basic Data Assimilation Training Course (BDATC), Logic-Line, and the Logic-Line Series are registered trademarks of Expansion Programs International, Inc. IBM PC is a trademark of IBM. NEC APC III and NEC APC are trademarks of NEC Information Systems, Inc. MS is a trademark of Microsoft Corporation. DISCLAIMER Expansion Programs International, Inc/Thunder- stone makes no representation or warranties with respect to the contents hereof and specifically disclaims any implied warranties or merchantability or fitness for any particu- lar purpose. Further, Expansion Programs Int'l, Inc./Thunderstone reserves the right to revise this publication and to make changes from time to time in the content hereof without obliga- tion of Expansion Programs Int'l, Inc./ Thunderstone to notify any person or organiza- tion of such revision or changes. WARNING Please make sure you read the LOGIC-LINE oper- ating manual and use it in conjunction with the LOGIC-LINE program. Do not expect the search- ing mechanisms to operate properly without reference to the instructions and correct pro- cedures. This is a whole new generation of technology and the user interface must be learned. è TABLE OF CONTENTS I. INSTALLATION TIPS AND GETTING SET-UP II. TRYING OUT LOGIC-LINE 1 FOR THE FIRST TIME III. LOGIC-LINE BACKGROUND, THEORY AND PURPOSE IV. HITS AND MISSES V. TYPES OF SEARCHES AND THEIR BASIC OPERATION VI. LOGIC-LINE 1 COMMAND SUMMARY VII. USING THE LEXNOTE PROGRAM VIII. USING THE OPERATING SYSTEM INTERFACE PROGRAM IX. BASIC PROCEDURES X. NOTES ON CROSSFILES XI. NOTES ON HASHFILES XII. WANT TO KNOW MORE? è I. INSTALLATION TIPS AND GETTING SET-UP In the process of operating Logic-Line various operational files are created, which will be explained. There is insufficient space on the floppy disk you have received in your Maxell box for this to occur. Thus we suggest that you copy the appropriate files onto either a separate floppy, or into your hard disk before trying to operate the program. Only the LEXSET.EXE file must be copied, and the --.doc, --.x, and --.hsh files if you want to use them. This will be explained further on. If for any reason you should have trouble getting the program to run on your system, or the program seems to run somewhat but the screen is also showing a lot of extraneous symbols or other "garbage", do the following: 1. Boot up your operating system disk. 2. Type in the command: TYPE ANSI.SYS [Rtn] If your screen message is "no file found" you need to obtain a copy of that file from an MS-DOS or PC-DOS operating system and put it into your system. 3. Type in the command: TYPE CONFIG.SYS [Rtn] If such a file exists as part of your operating system, the contents of that file will be displayed on the screen. 4. Check to see if the line "DEVICE=ANSI.SYS" is somewhere in the file. If it is not (or the file doesn't exist at all), this is probably the source of your difficulty. 5. If no CONFIG.SYS file exists you must create one, and insert the line DEVICE=ANSI.SYS. If there is a CONFIG.SYS file without the line DEVICE=ANSI.SYS you must add that line to the existing file. Use any word processor which enters the characters in flat ASCII format. Most any word processor (in non-document mode) will serve, or you can use EDLIN to do this, which is usually available as part of your operating system. è 6. Once you have fixed up the CONFIG.SYS file, you must re-boot the system before trying the program out again. Take the disk out, turn your system off and back on again, re-boot the system disk, and then try the program again. It should now work. USING EDLIN TO MAKE A CONFIG.SYS FILE EDLIN is not the most friendly word processor around, though many programmers feel it suits their needs. However, if you have no other word processor available that enters flat ASCII format files, EDLIN is fine to use, and should be available on your MS-DOS or PC-DOS operating system. The whole idea is to use a word processor or editor to create a file named "CONFIG.SYS" which contains the line "DEVICE=ANSI.SYS" inside it. If you aren't used to using EDLIN, don't try to use it to edit an existing CONFIG.SYS file; it will be easier simply to erase the existing CONFIG.SYS file and start over. (Just make sure to re-enter whatever was in the file before, as it may be set up a particular way for other reasons affecting your system.) If you need to use EDLIN to create a CONFIG.SYS file, just follow the instructions below exactly, and it will work fine. INSTRUCTIONS: 1. With your operating system disk in,once you have an operating system prompt, type in the command: EDLIN CONFIG.SYS [Rtn]. 2. Type in 'i' (then Return) for insert mode when the screen looks as follows: New file *i [Rtn] 3. Type in 'DEVICE=ANSI.SYS' (then Return) when the screen looks as follows: New file *i 1:*DEVICE=ANSI.SYS [Rtn] è 4. Type in 'Z' while holding down the Ctrl (Control) key (then Return) when the screen looks as follows: New file *i 1:*DEVICE=ANSI.SYS 2:*^Z [Rtn] 5. Type in 'e' (then Return) to exit the EDLIN program, when the screen looks as follows. This will return you to the operating system. New file *i 1:*DEVICE=ANSI.SYS 2:*^Z *e [Rtn] A> 6. Your CONFIG.SYS file is ready for use. Re-boot your system disk, and that file will be employed. GETTING STARTED: LOGIC-LINE SERIES 1 operates with MS-DOS versions 2.0 and greater. If you are running earlier versions of MS-DOS (i.e., less than 2.0) you may possibly have no difficulty, but we cannot guarantee it. MS-DOS versions below 2.0 may be okay, but we suggest you update your system to something more current to be sure. LOGIC-LINE 1 also operates with PC-DOS 2.0 and greater. DISK CONTENT: The files on your Maxell disk which pertain to LOGIC-LINE SERIES 1 are the following: 1) LL1READ.ME 2) LEXSET.EXE 3) LIBERTY.DOC 4) LIBERTY.HSH 5) FREEDOM.X 6) FREEDOM.HSH 7) ORDERFRM.LL èThe LL1READ.ME file (as you have discovered) explains the most basic ground-rules of how to use the Logic-Line 1 program. For a more full explanation including screen by screen model searches, filtering remedies that tie all the processes together, information on using Logic-Line as a tool to assist in creating expert systems, technical background on how the program works with bibliography and lots more, for $25 (+ shipping) you can order the full manual and documentation from: Thunderstone PO Box 839 Chesterland, OH 44026 (216) 449-6104 Ordering the full manual and documenation will register you as a Logic-Line user, and entitles you to program updates, defective disk replacement, phone hotline service, and periodic info letters on our latest technical developments. Also available for $20 (+ shipping) is a supplementary on-disk tutorial which we call the Logic-Line User's Supplement. This was written by an outside author to cut down the learning curve on learning to use the Logic-Line program (whether Series 1, 2, or 3); we also provide you with some supplementary data files and drills to assist you in learning how to do the searches. If you wish to expand the program searching scope to encompass numerical and symbolic information, hypenations, and such into your correlations, Logic-Line 2 is available (including full manual and documentation) for $125. Logic-Line 3, for $150, allows you to customize the Logic-Line 2 searching tool to search datafiles in other phonetic languages such as French, Spanish, Italian, and so forth. Logic-Line 3 is a package of Logic-Line 2 and Logic-Line 3 together, so you do not need to order both of them. See the ORDERFRM.LL file for more information on products available from Thunderstone/EPI, Inc. You can print out this file by simply engaging your printer (sometimes with Ctrl/P) and using the DOS TYPE command [TYPE ORDERFRM.LL (Rtn)], or whatever command you usually use to print files. Fill out the form and send it in with your order for the products offered on that list. LEXSET.EXE is the execute file. It is not copy protected. You can install it in your harddisk by simply copying that file into the sub-directory (or main directory) you want it in. It is small enough that you can put it in several places if you wish. Logic-Line 1 will search and correlate any text file in ASCII format as long as these files are lined up in the same path with the LEXSET.EXE file. LIBERTY.DOC is a practice data file containing Patrick Henry's speech "Give me liberty or give me death" given to you for practice purposes. LIBERTY.HSH is the corresponding hashfile, to be explained later. èFREEDOM.X is a crossfile provided for practice (this will be explained later). FREEDOM.HSH is the corresponding hashfile. You must have COMMAND.COM and EDLIN.COM lined up in the same path with the LEXSET.EXE file for for general use. You get these files from your DOS operating system. For the system interface to work correctly, a copy of the command processor (COMMAND.COM) must be available in the current path. This will usually be the case; however, if not, you will have to check your MS-DOS manual for information on the "COMSPEC=" entry, on "COMMAND.COM", and on the path command. For a one diskette only system this means as a practical matter, that the "LEXSET.EXE" file, must be installed on your system disk. This can be done using the "COPY" command. For LOGIC-LINE 1 to run efficiently, you should provide for 10 files to be open concurrently. Depending upon your system's defaults you may wish to specify: FILES=10 and BUFFERS=10 in your "CONFIG.SYS" file. As mentioned before, you must be sure to include the line "DEVICE=ANSI.SYS" in your "CONFIG.SYS" file. This omission can prevent the program from running properly, or may seem to run the program, but with additional "garbage" on the screen. Also as mentioned before, make sure that you have a copy of ANSI.SYS on your operating system. If it isn't there, then the CONFIG.SYS instruction to use it will be to no avail. While in most cases this file will be part of your operating system, it does happen occasionally that it isn't included. If so, get a copy from your dealer, or from wherever you got your operating system, and copy it onto your operating system disk or onto your hard disk. Do not write-protect the software program, as this will prevent the searching algorithm from functioning properly. è II. TRYING OUT LOGIC-LINE 1 FOR THE FIRST TIME If you just got your disk, you will want to know what kind of program this is and how it works, but you won't be wanting to read all about it. So try this, just to get a general idea of the way Logic-Line handles datafiles; then read the rest of these LL1READ.ME instructions so you understand how to get the program to do what you want it to do. 1. Type in "LEXSET", and the Logic-Line 1 front screen will appear. 2. Type in 'f' for filename; you will be asked for the name of a file. (You can enter 10-12 files at a time, but just do this for now.) Enter the name LIBERTY.DOC (lower or upper case is okay), followed by the Return key. The filename will appear at the top of the menu. 3. Type in 'w' for word to look for; you will be asked for a word to add. (You can enter 10-12 words at a time, but just do this for now.) Enter the word BIND (lower or upper case is okay), followed by the Return key. This word will appear at the top of the menu. 4. Type in 'g' for go. This will execute a simple 'g' search. The program will look for aggregates from the textfile containing the word "bind" and will present these to you on the screen. 5. If you are in manual mode (the default if you didn't choose one) a menu appears on the screen after each aggregate is presented. You can say 'y' for yes, to save that aggregate and dump it to a savefile automatically being created; or you can 's' skip it and it won't be saved. You can also select 'L' for "Lexnote" (see later on for explanation) which among other things lets you see the context around that aggregate. (A 750 byte window is opened around the text so you can see it, by selecting 'v' for "view" under the Lexnote subroutine.) 'Y' or 'S' continues the search, while 'A' aborts the search and allows you to go on and do something else. è 6. When this search is complete, if you want to get a quick idea of how the program makes correlations, then continue and do the following. It's important that you just did the above search, so that certain results will have been dumped into what we call the "LOOK.AT" file. 7. Type '@' to "set associations". 8. Type '=' for "REPLACE" associative elements from file. 9. When asked to enter filename for addition of associative elements, type in "FREEDOM.X" as the crossfile name. 10. Type 'r' to return to main menu. 11. Type '*' for "use results". 12. Type 'c' to execute a cross-search. 13. Let the search run its course. The interaction and output will be similar to that of a 'g' search, but as you will see, you will get aggregates located from the textfile which are associated, but not on a literal basis. There's lots more to learn about how to locate possible associations; but this should give you a general idea of how the program operates and interacts with you, and where you can learn to take it. è III. LOGIC-LINE BACKGROUND, THEORY AND PURPOSE The purpose of LOGIC-LINE SERIES 1 is to serve as an intelligent aid to persons who wish to access and study as well as retrieve data from a larger body of information or from textual material on a frequent basis. Its multiple use from retrieval to content analysis serves both the traditional and AI communities as well. At a first level, LL1 is a tool to search, study, and retrieve information independent of how it has been input or what state of order it is in. This is by itself a valuable facility. At the next level LOGIC-LINE 1 can be used to help pervade a database and find what one needs through the intelligent cross- association algorithms within the programs. Essentially, Logic-Line connects text references to search requests by first discarding certain noise words (fuzzy set) contained in each (the unimportant bridge words like "what", "do", "is", "and", etc.) and then looking for patterns among the key words or specifics in the search request and the key words in the text files. The key words or specifics are defined by those words remaining after the noise words are tossed out. During a search, Logic-Line performs a mathematical computation regarding the content of the text, not a symbolic one. This is a form of content analysis which bypasses the requirement of the machine having to actually understand the meaning of words and grammar structures; it is only aware of pattern, and in this way the Logic-Line system is a model for machine thinking in a world of binary numbers. LOGIC-LINE 1 is built around a number of techniques which permit the searching of textual files much more rapidly than can be done by conventional methods. Units of information rather than being manipulated as pre-planned and pre-arranged groups of data and records as in many database management systems commercially available, are instead moved about by words, symbols, and sentences. Thus you can pinpoint and retrieve data, information and concept structures within a given textual configuration much more accurately, rapidly, and elegantly. You can use words, symbols, keywords, groups of words, and groups of keywords, to search a file, data base, or groups of files or databases. You can search a file or database of just about any size. Complex or simple file searches can be done on your micro computer which previously could be done only on larger scale equipment, and with a great speed advantage. èThe rationale behind this is related to our design of possibilistic software which parallels the sub-cognitive thought processes in an assistive manner, rather than concentrating on development of probabilistic and deterministic solutions presented without necessity for the user's final decision point. While machines themselves may or may not evolve to a point where they can demonstrate real judgment and thinking capability across a spectrum of subjects as an everyday affair, a human using information made available by the computer clearly has that capability already. Thus, by using the computer as a service facility for the human mind, the user can adjudicate when an item of information occasionally does not fit his data request, while taking advantage of the greatly increased speed that can be obtained when a computer is only expected to be "artificially" intelligent, rather than actually intelligent, by human definition. Artificial Intelligence is a subject currently receiving considerable discussion and is an emerging science. Whether a piece of hardware or software is actually intelligent is purely speculative at this time; but what is important now is whether the hardware or software can perform a valuable function. This approach creates a new view to the man/machine relationship, at least in the field of decision support science. The machine is a tool to support decisions that the man using it will make; but we keep in mind that the man is the one making those final decisions, while the computer is simply supporting him in that decision-making activity. You will find that the ability to cross-correlate information will significantly increase your information assimilation abilities in general, and may enhance the development of your own human cognitive functions. è IV. HITS AND MISSES When you use Logic-Line 1 to execute a search, your target is a cluster or aggregate of information which is relevant to what you are looking for, and accurately responding to the search question. When you find something like this we call it a "hit". You will occasionally get a retrieved excerption of information which is not what you are looking for; when this happens we call it a "miss". This program design provides for increased speed in retrieving information due to its not having to filter out all the misses before retrieving an aggregate of information. This design also keeps the memory requirement (size) of the program very small, which means you can use it on a micro computer leaving most of your disk space free for your data files, not for the searching facility. The search program locates and recognizes a numerical pattern, rather than a conceptual meaning as a human would do. Thus sometimes more than one word will sometimes have the same numerical pattern, or "meaning" to the computer program, which you the human will recognize as a "miss". By design, the program explores a lot of pathways, some of them blind alleys. The person sitting at the keyboard is the ultimate decision point on whether something is a blind alley or not; the computer's job is to discover as many pathways as it possibly can as fast as it can. We give up a bit of accuracy in favor of speed. è V. TYPES OF SEARCHES AND THEIR BASIC OPERATION In Logic-Line 1 you have available to you 3 types, or levels, of searches: 1) Straight search for words or symbols. 2) A priori cross search from a given set of words or symbols. 3) Associative element cross-correlative cross search. The element of recursion, which uses the results of the previous search or searches, does not come into play until the 3rd level of associative search. It is this level of search which brings the data retrieval activity into the realm of Artificial Intelligence. The above types of searches can be used separately or in combination. Where they are used in combination the most intelligent and effective searches can be done. In order to properly weave an intelligent search which uses all 3 search types, one must first understand how to use each separately. Get the full manual for a complete explanation of all these, and some training in how to put them all together. INTRODUCTION TO COMMANDS: The 3 types of searches are invoked most basically in connection with these commands: 1) The straight search 'w'/'g': A literal search which looks for words, whether lower or upper case, as they appear in the specified file or files. To do this you use the 'w' command to enter the word or words, and you use the 'g' command (for "go") to execute the straight search for the word or words you entered. 2) 'c': The a priori cross search uses a set of words or symbols, as separated by spaces and ended with a period, in a "crossfile" to search a specifed file or files. This should be differentiated from searching with a single word or words (one at a time, as with the 'w'/'g' function). è To do this you use the 'c' command alone, to implement a crossfile name which you specify. This is not really a literal search for the words in the crossfile, although at times it will not differ from a 'g' search which searches for a series of words in a specified file (or files). You are looking for patterns in the textfile which have possible similarity to patterns in the crossfile. 3) '*' 'c': In the associative element cross-correla tive cross search, you are still searching a specified file or files by means of a specified crossfile (constructed from one or many separate words and/or symbols). What we call for short a "star-c" search, invokes recursion by use of the asterisk '*' (or "star") prior to the 'c' command. You could say that the a priori 'c' search is a 2-way direct association; i.e., crossfile to textfile. The difference in the star-c search is that you are not just directly searching the textfile with the contents of the crossfile; you are adding a third associative element with the addition of a third file, called the "LOOK.AT" file. You could say that the crossfile must first "look at" this second file before it searches the contents of the specified textfile (or files). Thus you are creating a 3-way (rather than 2-way: crossfile to textfile) matrix of association: that is, crossfile to look.at file, to textfile. This is, of course, actually a mathematical process. FURTHER EXPLANATION OF FUNCTIONS: For each of the 3 types of searches covered above, the searching process actually goes on in specially encoded files called hashfiles. Each file, whether textfile, crossfile, or look.at file, has a corresponding hashfile which is a fraction of the size of the textfile. (See a computer language dictionary for further data and description of hashing technology.) A hashfile is automatically created for a textfile when you search it. èWhen you use a crossfile to search a textfile you must hash the crossfile first, and you do this with the 'h' (hash/rehash) facility. In each of the 3 search types you are looking for some sort of a match. The first type is in mathematical terms a straight match; the second type is still looking for a match, although a more complicated match of pattern. In the third type, you are in essence looking for associations of patterns of meaning via the following components: the contents of the crossfile, the contents of the look.at file, and the contents of the textfile. As this process is done in the hashfiles rather than the textfiles, the process is entirely mathematical rather than related to meaning as a human would understand it. A search algorithm looks for potentially similar patterns between what you are searching with (the contents of the crossfile and the look.at file) and what you are searching for (selected contents in the textfile). An association matrix deep inside the program determines whether a match is or is not possible between the two patterns, based on whether it is within certain adjudicated mathematical limits. If it is a possible match, then that mathematical location is transposed back to the textfile and marked with a file pointer. The program then parses out a chunk surrounding that location and presents it for viewing as a pattern of inference. The user supplies the missing factor of meaning, and makes the final human decision as to whether the parsed aggregate of text is or is not relevant to what he was looking for. OUTPUT AND ITS RELATION TO SEARCHING: Regardless of what type of search you are using, once a point of match or association is found in the file, it is dealt with in the same manner. Within certain limits, the program will try to go back to the last period and forward to the next, ignoring exclamation points and question marks. This parsed out aggregate of information is presented in full on the screen. The user (if in 'm' manual mode) then has a choice of whether to keep it by selecting 'Y' for "Yes, this is relevant," or to reject it by select ing 'S' for "Skip it." èThose aggregates which the user decides to keep are extracted and dumped into a separate file along with all the other extracted aggregates of information. This file is called the "LOOK.AT" file, the same file referenced above. This file will accumulate all the results of your searches until you clear it and start over. While each of the 3 search types dumps extracted information into the look.at file, it is only the third type which uses those search results to accomplish its search. Thus a 'g' search and a 'c' search can be done independently of each other and independently of the content in the look.at file. However, by invoking the '*' (which means "use results") before the 'c' for a "star-c" search, the contents of the look.at file be come an integral part of the parameters of your next search. Additionally, if there is no content in the look.at file, then you can't begin a star-c search because there is no third associative element to complete the association matrix. Thus you will see that the first step in a star-c search must always be an action to put something specific into the look.at file. A 'g' search is the best to use as you can most tightly control the results you are put ting into the look.at file. In other words, by: (1) Selectively locating an extract from a textfile which represents a concept you are trying to draw inference from (with a 'g' search) and dumping it into the look.at file, and then: (2) Creating a separate crossfile which contains a concept you wish to cross- correlate against the contents in the look.at file; you can then (3) Execute a star-c search which uses both those points as reference and draws patterns of inference between the crossfile, the look.at file, and the textfile. This whole process comprises the associative element search, and is the search which approximates intelligence, even though it is the user who supplies the final meaning. If you wish to create a new matrix of association, you can use another and another and another crossfile, to pit against the look.at file and to search the textfile. Your look.at file is building as you do this, which means the search process is "learning" as it goes. èEventually, you will wish to start an entirely new search; to do this you will have to remove the contents of the look.at file. As this is a cumulatively built file, you will have to first extract the content that is in there, before you put new content in. To do this you use the command 'b' for "begin extract". It may be that in the course of a search you wish to segment off the results of part of the search from another part of the search. In this case, even though the look.at file is being built up cumulatively, you can selectively name separate files and archive the selective search results into these separately named files (as well as into the look.at file). These savefiles, as they are called, are created with the 'n', for "name", and 'r', for "archive result" functions of the program. All these files, that is, the look.at file and any savefiles you name along the way, are noted on your directory, are in standard ASCII format, and are directly accessible to you on your directory. They will respond to a word processor and can be edited and printed for report generating purposes. They can also be hashed and selectively used as crossfiles for subseqeunt searches. Or they can be organized and used in the generation of expert systems. THE LEXNOTE SUBROUTINE: Regardless of which type of search you are using, your target is an aggregate of information which is presented as a possible match to what you are looking for, as covered previously. As information is retrieved, you have the option of further inspecting that aggregate of data by use of the LEXNOTE subroutine. This sub-program allows you to inspect and tabulate the aggregate in detail, as well as to look at it in context of the database by opening windows around it. LEXNOTE can be used on any hit which is presented by any type of search. If you are unsure of the relevance of a certain aggregate, use the 'v' function in LEXNOTE to "view" the surrounding area and establish by context if this is or is not what you are looking for. Very often context will reveal an extremely intelligent association which was not clear when only the parsed aggregate was viewed. You can use 'd' to "display" the line (aggregate) again, and 'c' to "continue" the search. Another less obvious use of LEXNOTE is to use 't' to "tabulate" the results, or in other words, to inspect the distribution of what the program is considering to be "noise words" and what the program is considering to be "specifics". èThose words the program is considering to be noise words (such as "a", "an", "the", "but", "and", etc.) are filtered out from the coding in the hashfile. Only the specifics are used for the search. The LOGIC-LINE program you are using is programmed for modern English. If you are searching texts of Old English or of some other language, the basic mathematical algorithm will still work, but the noise word files would have to be edited. LOGIC-LINE 3 allows you to edit these noise word files and customize LOGIC-LINE for use with just about any phonetic-based language. (For more information please contact Thunderstone.) STRAIGHT WORD SEARCH: The straight word search and data retrieval process consists of specifying files to search, and words you wish to search them with. The selected words will retrieve textual material in the specified file or files, and present it to you exactly as it appears in those files. All references in the file or files which contain your selected search words will be given to you, one at a time, in the sequence retrieved by the program. LOGIC-LINE 1 offers a way to accomplish this literal search of the text stream; you can look for words as words, which disregards upper and lower case ('w'/'g'). Use this straight search option to dump specific entries into your look.at file before you go into an associative element search. You will see in the later procedures that a 'g' search is required as the first step before a star-c search is executed with a crossfile. Sometimes, as in the Filtering Processes outlined in Appendix A (contained in the full Logic-Line 1 manual), you will want to set up the parameters of a star-c search before you do anything else. Still, you must actually execute a 'g' search to dump information into the look.at file, before you actually execute the star-c search. A PRIORI CROSS-SEARCH The a priori (or given word set) cross-search is designed to permit a user to configure a natural language query, or sentence, or list of keywords, to operate as a unit request upon a data base. Having this facility tends to bypass some of the complexity inherent in figuring out what the parameters of an associative element cross-search could be. It is a good place to start searching for what you want to know, by simply listing out the spectrum of keywords around which you want to search, or formulating a query to "ask" the data base. èOf course as there is no thinking person inside the computer, you will not get a "thought out" answer to your query. But you will get a curiously accurate pattern of response, which is a "match" to the pattern created by the string of words in your list or your query. You can often get startlingly intelligent responses using this type of search mechanism. Usually it will dig out some concept of what is in the database so that you can then further interrogate the data base with other methods of search, such as an associative element star-c search. You must keep in mind one characteristic of this kind of search. If you formulate several queries, such as "What do you think about __________?" with the only difference being in the "__________", your search response will probably bring up aggregates which are the same for much of the search. Realize that the search is first responding to the "What do you think about" portion of the query, so the line of logic followed will begin the same with each such query. Be patient though, and you will usually find that even with similar queries such as these that there will be a few different aggregates brought up at the end of each search. For creative optimization of this searching method, we suggest that you practice varying your formats and query styles so as to elicit varied patterns of response from the data base. This is a skill which professionals in different walks of life naturally learn in order to elicit varied responses from people, and is not an unnatural way of obtaining data, whether from a person or from a machine. ASSOCIATIVE ELEMENT SEARCH: The associative element star-c search is used in conjunction with a sub-menu called up with the '@' command, which we refer to as the "set associations" menu. This offers 3 kinds of associations, which we will herein explain. Each of these associative element search methods automatically locate possibilities of logical unions of information and inference inside a larger universe of textual data. When humans do this, the activity is called sub-cognition, associative logic, or concept building. In order for a human to do this with a body of information, he must read all or much of the data before he can string his own associations and inferences. èThe action of making contact with all the information is what takes the most time for a person. The associative element search acts as a tool to the person by searching through the database for that selected information the person wants, and stringing associative links through the material in a manner consistent with the user's comfort level and personal thought patterns. The final deductions, conclusions, and decisions about the information retrieved are left up to the searcher. When you make the decision to do a star-c ('*' 'c') search, you are electing to use recursion by instructing the search algorithm to "use the results" of what it has already found as it executes the cross search. The way in which it uses these results is modified by the set associations menu, which provides 3 options, as covered below. When we view these options we are talking about the way in which the associative links are set up inside the program. THE SET-ASSOCIATIONS OPTIONS: When you set the parameters of a star-c search, you must: 1) First choose the type of associative link you want to use (+), (=), or (a), 2) Enter the crossfile you will be associat ing (a crossfile name), and 3) Instruct the program to use the results (*) of the foregoing search as contained in the look.at file, and then 4) Execute the cross (c) search. It is possible to execute 'g' or 'c' searches to dump things into the look.at file both before step 1 and/or after step 2. Decisions as to sequence will come with experience, and as you learn the basic procedures and those in Appendix A. Nevertheless, the important thing to understand is that the above 4 steps are required at some point, and you must understand the 3 associative link options in order to decide which to use when. There are 3 methods of setting the associative links: '+' ("accumulate" associative elements from the file), '=' ("replace associative elements from the file), or 'a' ("allow" any and all associations). èTo understand how these options relate, it is easiest to first understand the difference between the '+' and the '=' association options. CONVERGING (+) VS. DIVERGING (=): The '+' option is cumulative in that it expands the avenues of association in accordance with the additional crossfile elements introduced; i.e., the crossfile you enter with the '+' command will be associatively linked up with everything that has gone into the search up to this point. This could be viewed as a convergent search, as it associates this plus this plus this, and so on, along a discrete path converging on a basic theme. The '=' option is substitutive in that it redirects the channels of association in accordance with the additional crossfile elements introduced; i.e., the crossfile you enter with '=' will replace the associative links you had in place up to this point with the associative links you now set. This could be viewed as a divergent search, as it associates this or that or that, and so on, along a path branching outward, while maintaining a basic logical line. To be more precise, in an accumulative (+), or convergent association, the accumulation is referring to the crossfiles you enter. It does not refer to the textfile or to the look.at file. When you select '+', you are adding the crossfile you are about to enter to whatever associations are already in place. In a substitutive, or replacive (=) or divergent association, the substitution is referring to the crossfiles you enter. It does not refer to the textfile or to the look.at file. When you select '=', you are replacing all past assocations that were in place with ONLY the crossfile you are about to enter. Thus if you select an '=' association with a certain crossfile and execute a star-c search, the search will be done with associative links between the contents of the look.at file, the contents of the crossfile entered, and the contents of the textfile. If you then wish to expand the avenues of association to include associations with the crossfile you just entered, then select '+' before entering the next crossfile. If you wish to replace the next association rather than add to the last one, then select '=' before entering the next crossfile. For example, let us say you are searching a textfile called "FORESTS". You do a 'g' search and extract a sentence concerning "planting seed-bearing trees", dumping it into the LOOK.AT file. You then do a cross-search against the textfile and the look.at file with the concept "moisture" as represented by a file containing the word "water". You execute the search and come up with various hits. èNow you want to expand the avenues of that association, to include the "water" crossfile, so you select '+' and enter a new crossfile containing the word "storms". As long as you select '+' before a new crossfile, you will include each new crossfile in the association. Now if you want to replace those crossfiles with a new concept, then select '=', and then enter your new crossfile containing, say, "fire". Your association links will be replaced by this crossfile alone, as cross-correlated against the contents of the LOOK.AT file and the textfile "FORESTS". ALL-ASSOCIATIVE: Essentially "all associations" (a) means, literally, that the search facility will try to search out all associations of the nature requested. When you select 'a', you are really selecting an expressed default association. When you select '=' or '+', you are setting up a converging or diverging associative link in connection with a crossfile you are about to enter. If you select 'a', there is no instruction to enter a crossfile, as, by definition, "Allow any and all associations" means that the association would not be limited by an entered crossfile. The type of associative link which is in place for a routine 'g' or 'c' search is in fact any and all associations, in a 2-way, rather than a 3-way association matrix. In an all-associative element star-c search, you are making any and all associations between the content of the look.at file and the entire content of the textfile (excluding noise words). As the possibilities are quite extensive, this creates a very broad search. If you could not find what you wanted on either a '+' or '=' element search, then the all-associative search will pick up any and all possibilities which were left. This type of search can be used as a facility for viewing much of the text in easy to experience chunks along the line of a particular text stream or point of view. The all-associative search is the most exhaustive process. It will literally give you almost all the content to view, but in a sequence which will be more likely to have what you're looking for closer to the beginning than toward the end, with a diminishing percentage of possible association. èWHERE TO START: There is a peculiarity to the interaction of the various associative linkings that must be understood so as to make an overall search do what you want it to do. The default association for any type of search, including 'g' searches and 'c' searches, is all-associative, or "allow any and all associations". This means that when you use a 'g' or 'c' to dump information into the LOOK.AT file, you have in place an all-associative pattern as the default. Now, when you go the next step into a star-c search, if you select '+' as your first type of associative link, realize that you are ADDING the crossfile you are about to enter to whatever associations have already been made. Therefore, you have a fairly unlimited association to start with, as the crossfile is only expanding upon the all-association which has been made thus far. For this reason, if you select '+' for your first search, it could happen that you get into a seemingly long or nonsensical loop, which will eventually complete itself, but then again if the file is very long, it could seem to be rather endless. All that has occurred, is that by starting the star-c '@' parameters with a cumulative (+) associative link, you have linked the entered crossfile to the last associative links, which in this case most recently was "all-associative". The remedy is to REPLACE (=) the most recent all-associative links with your next crossfile by selecting a (=) divergent search before you select a (+) convergent search. Thus as a firm rule, when you set the first parameters for your star-c search, always select '=' before entering the first crossfile. Then you can follow up with '+', '=', or 'a', as you deem appropriate, to accomplish the goals of your search. ADVICE: The mechanics of human intelligence is generally understood to be a rather complicated affair. Thus any attempt to approximate this quality in a machine is bound to have its complexities. This section has attempted to lay out the basic workings of the LOGIC-LINE program design, and explain the different levels of search operations and their relation to each other as well as their relation to the goal the human has in using it. The full manual (available from Thunderstone) contains model searches as well as appendices which tie together the various searches. We recommend this for a full optimization of the LL1 program. è VI. LOGIC-LINE 1 COMMAND SUMMARY Following are explanations for each of the basic commands available on the "LEXSET" main menu. w = WORD: Enter a word, or add a word to the list of words to be searched for. As many as 10-12 words may be entered, although just a few is much more likely. They may be upper or lower case; it makes no difference. f = FILE: Enter a file name, or add a file name to the list of files to be searched. As many as 10 to 12 files may be entered, depending upon size of file. g = GO: Execute a search based upon parameters established by way of the 'w', 'f', 'm', or 'a', and 'n' commands. This will generally be your starting point. This command will accomplish a simple, linear, and literal search of the file, files, or data base. m = MANUAL: Select manual mode for search ('g' or 'c' command). Manual mode allows the user to view each data cluster as it is retrieved. Each time an aggregate appears on the screen it stops for you to view it as different from automatic, where the aggregates are presented very quickly, automatically, and at the speed that the search is accomplished by the program without regard to the user. In manual mode, the user has 3 options after each data aggre gate is presented for viewing. These options are presented on the screen as follows: respond: Y to include or S to skip or A to abort or L enter LEXNOTE Y: means YES, you want to include that aggregate in the data acquisition file being created by the search. è S: means you want to SKIP that aggregate and continue with the search. You would usually choose this when the aggregate being shown is a repeat of one already retrieved, or when an aggregate is retrieved which is incorrectly included in this search, as a false indication. This will be frequent toward the end of most all-associative searches. A: means you want to ABORT that particular search before it is complete, and continue back on the LEXSET program. L: See section on Using the LEXNOTE program. a = AUTO MODE: Select automatic mode for search ('g' or 'c' command). Automatic mode automatically retrieves data clusters and puts them into a separate file for later study. This mode can be used for a fast viewing facility, to be studied later. The automatic mode does not permit any user interaction once it gets going. It's probably best used for large gross searches. It is a data retrieval milling machine, and the file it creates can be edited at a later time. n = NAMES: Select names for a SAVEFILE and NOTEFILE. This facility permits you to set up special files which generally correspond to an area of inquiry, and allows you to pursue several lines of research and retrieval without getting the results confused with each other. To understand this more clearly, here are definitions for 3 kinds of files: LOOKFILE: File used to save sentences or data clusters extracted during a search operation until the next search operation is done. The contents of this file are preserved when LEXSET is exited, so the material can be accessed with an editor or most word processors for whatever you want to do with it. This file appears on the directory as "LOOK.AT". è This file has technical significance and use to the searching facility and the inner workings of the LEXSET program, and makes possible associative searches. It is better to use the SAVEFILE facility on the LEXSET menu for the purposes of tabulating and recording information searched. SAVEFILE: This file works the same as the LOOKFILE except that the results of a search do not go into the savefile unless you specifically name (n) the file before executing the search, and archive the results into it (r) after the search. The SAVEFILE appears at the top of the LOGIC- LINE main menu, as: $AVE-1T.TMP. This name remains until you use the 'n' command to rename this file. The name you give the file will appear at the top of the screen in place of $AVE-1T.TMP. (See section on "Naming Files" for information on allowed names.) NOTEFILE: The NOTEFILE contains references to the mater ial in the form of pointers, but does not contain the actual retrieved information. The LEXNOTE subroutine interacts with the NOTEFILE. The NOTEFILE appears at the top of the LEXSET main menu, as: $AVE-1T.---. This name remains until you use the 'n' command to rename this file. The name you give the file will appear at the top of the screen in place of $AVE-1T.---. (See section on "Naming Files" for information on allowed names.) If you have no need to save a specific notefile, choose not to name it by typing the 'Rtn' key when asked for a notefile name (after you give a name to the savefile). è USING THE 'n' COMMAND: You use the 'n' command when setting up the parameters of the search, before using the 'g', 'c', or '*' 'c' commands to execute the search. As above, simply designate a name for the SAVEFILE, and if you wish, for the NOTE FILE, before starting the search. Then when the search is complete, use 'r' to archive the result, and the information retrieved during the search will be put into a permanent file with the names designated. * = USE RESULTS: Sets up the routine to use results from a previous search while doing the next search. This is required to do any associative element search. The search facility will be drawing on data it has compiled in the LOOKFILE (that is, the "look.at" file). By using this command, you are selecting recursion in the data processing by using the results of the previous search ('g' or 'c') for this search ('c' or 'g'). The fact of "using results" is perhaps of more interest technically than practically. What is important to remember about this command (*) is that it is used majorly for the following 2 situations: 1. If you have just completed a search, having created a temporary file which holds those results but you have not yet archived them to a permanent file, you can search or cross search those results by pressing '*' then 'g'. In doing so you will notice that "LOOK.AT" will appear under the "Files to Look In" Section on the LEXSET main menu. This is because you are now searching over the results of the last search (i.e., the data clusters retrieved in that search, only) rather than the whole of the files you first searched. è This type of '*' 'g' search can find intersections of sets of information from large data bases quickly, and can be assistive in locating literal common denominators in small or large bodies of information. It could be especially useful in conjunction with automatic mode, if you had little time available but wanted to do a precise search for later review. 2. When you want to do a convergent (+) or divergent (=) associative element cross search, in order to properly activate this facility you must press '*' and then 'c'. If you neglect to press the '*' before pressing 'c', even if you are searching with a crossfile, you will activate an all-associative search, but not a recursive convergent or divergent search. c = CROSS SEARCH: Execute a crossfile search. This is the couterpart of the 'g' command. Where 'g' executes a literal search based on whatever search parameters you have set, 'c' executes an associative search. This generally would follow the '*' command, for an associative element search. However, a 'c' search can also be done without the '*', as in an a priori 'c' search. In either case, the 'c' command operates in conjunction with a pre-configured word list which will be built by you in a separate file. This file is called a crossfile. In this separate cross word file, or cross search file, words must be separated by blank spaces, and ended with a period. These crossfile, cross-referencing word lists or sentences are built by accessing the operating system commands (see Section on Operating System Interface.) The 'c' command is the signal to start searching, using that crossfile. In relation to this, you will notice 2 categories on the main menu screen: è CROSSFILE WAS: This shows you while searching what the most current crossfile is or was which was used to do a search. ASSOCIATIONS: This shows you which kind of associative link is (or just was) in use. (See next section "Set Associations" for more information.) @ = SET ASSOCIATIONS: This '@' accesses a menu which permits you to select the type of associative linkages you want to use in the forthcoming star-c search; that is, one of the following choices: + - ACCUMULATE associative elements from file (this is the convergent association search; associates this plus this plus this along a discrete path converging on a basic theme). = - REPLACE associative elements from file (this is the divergent asso ciation search; associates this and/or that, along a branching outward path while maintaining a logical line). a - ALLOW any and all associations (this is the all-associative search) r - RETURN to LEXSET (after setting association links) Further instructions when using '+' or '=': To understand the relationship of these types of linkages please read the section: Types of Searches and Basic Operation, which covers the subject fully. r = ARCHIVE RESULT: This 'r' copies the results of the last search from a temporary file (LOOKFILE) to a SAVEFILE. This can only be accurately done if you used the 'n' command before the search began, and assigned a name to the SAVEFILE. (See 'n' command information in this section.) è s = START OVER: This 's' allows you to clear the workstation in terms of files searched and words used to search, without having to exit the LEXSET program first. When you use this 's' command, you simply START OVER with empty word and file lists. You would also use this if you made an error in using the 'f' (file) or 'w' (word) commands. q = QUIT THE PROGRAM: This 'q' closes all existing files and returns you to the operating system. $ = OPERATING SYSTEM: This '$' presents a secondary menu from which you can select an option via the operating system interface. This is the menu that puts you into the DOS EDLIN program so you can create cross word files to search with, where you need them for associative element searches. The menu looks like this: 1 - CREATE a new text file 2 - EDIT an existing text file 3 - DELETE a text file 4 - DIRECTORY display 5 - RETURN to LEXSET These options are covered more fully in the Operating System Interface Program section. L = EXECUTE LEXNOTE PROGRAM: 'L' accesses a natural language breakdown program, allowing you to study a retrieved cluster of information linguistically. The command relates to lines of text "remembered" from previous searches, accessed via a NOTEFILE, the name of which can be set by the 'n' command from the main menu. è The program has the most application where you are customizing LOGIC-LINE to run in a lan guage base other than Modern English, which is the domain of LOGIC-LINE 3. The use of the LEXNOTE program for routine LOGIC-LINE use is covered in the next chapter, Using the LEXNOTE Program. b = BEGIN EXTRACT: Extracts (clears) the content from the LOOK.AT and those recursive files that are involved in an associative element search. 'b' begins the extract, and the message "done" appears when the action is complete. This can be done from inside the LEXSET program; thus there is no need to exit with 'q' to begin a new search from scratch. Execute this function when you are beginning an entirely new search and you don't want anything left from the last search(es) to influence your next search. h = HASH/REHASH: Use 'h' to hash (create a hashfile) a file automatically. Use it also to rehash a file which has been edited or tampered with. è VII. USING THE LEXNOTE PROGRAM The 'L' command accesses a natural language breakdown program, called the LEXNOTE program. It allows you to study a retrieved cluster of information from a linguistic viewpoint. The command relates to lines of text "remembered" from previous searches. This data is accessed via a NOTEFILE, the name of which is set by the 'n' command from the main menu. When the LEXNOTE program is accessed, it starts with the first data cluster previously retrieved and recorded in the NOTEFILE, and gives you a menu of various options with which to address that aggregate of acquired information. As you will see, you can exercise an option to tabulate, view, or re-display the retrieved aggregate; or you can choose to continue on to the next aggregate recorded in the NOTEFILE and exercise those same options; or you can return to the LEXSET main menu. To maintain the option to use LEXNOTE on the results of specific past searches, be sure to use the 'n' (name) function before you do the search, to give a permanent name to the NOTEFILE. As covered in the LOGIC-LINE Command Summary, the NOTEFILE will be shown at the top of the LEXSET main menu as $AVE-1T.---, until you give it another name with 'n'. When you press 'L' from the LEXSET main menu, the data cluster you are currently viewing in manual mode, or the first data cluster from the NOTEFILE most closely connected to the files you are searching or last searched, will appear on the screen. Below this will be a menu which contains this information: T - tabulate word groups V - view surrounding text C - to continue D - to display the line R - to return to main menu These LEXNOTE commands are explained in detail as follows: T = TABULATE WORD GROUPS: Pressing 'T' will tabulate the cluster into word groups. This creates a schematic representation of the retrieved information (best used with a color monitor), which with a little practice will become quite useful for scanning extracted material when you're not sure exactly what you are looking for. è It will be found that viewing the data cluster through the viewpoint of this LEXNOTE facility permits new possibilities of inference to be drawn. It filters out the "noise words", and provides a list of keywords delineated as "specifics" which can be used to search the data base from different angles. It is also through this facility that you can fine-tune noise word files when dealing with something other than Modern English. (Refer LL3.) V = VIEW SURROUNDING TEXT: This displays an expanded window around the retrieved text. This window extends for 750 bytes up and down. This facility allows you to put the retrieved data cluster back into the context it came from, to expand your understanding of the information without cluttering up your retrieved information file with too much information. C = CONTINUE: This allows you to continue, or move from one retrieved data cluster to the next, while in the LEXNOTE program, in the sequence in which they were retrieved, and in which they are listed by pointers in the NOTEFILE. D = DISPLAY THE LINE: This 'D' allows you to reDisplay, or put back the retrieved data cluster you have been studying on the screen for viewing. If it is used following 'T', the tabulation is cleared from the screen, and the original data cluster put back. If it is used following 'V', the window on the screen is closed, and the original retrieved data cluster is put back. è R = RETURN TO MAIN MENU: Pressing 'R' takes you out of the LEXNOTE program and puts you back into LEXSET, at the LEXSET main menu. THE NOTEFILE: When you do a search, as explained earlier, a file pointer appears prior to the textual aggregate. Where the textual aggregates are dumped into the look.at file and into any savefiles you name, the hexidecimal file pointers are separately dumped into a default notefile, and into any notefiles you specifically name. The name of the default notefile is: "$AVE-1T.---" and this is the file that appears at the top of the LOGIC-LINE menu. If you give a notefile a specific name, that name will replace the default name. If you are not paying particular attention to the notefile, then you may not notice that the default notefile is a continuing record of all the file pointers you have encountered in all your searches; in other words, it is cumulative, and the contents of the notefile are not extracted when you use 'b' to extract the content of the look.at file. This means that the size of that default file gets larger and larger the more you use the LOGIC-LINE program. Thus, you may want to erase it from time to time if it gets too large, or if you want to be sure to save it, conceivably you might want to copy it onto another disk. Of course it is more likely that if you are concerned with saving the notefile results permanently that you would be naming that file with the 'n' function; nevertheless, this is mentioned by way of explanation of program mechanics. MAKING USE OF THE NOTEFILE: Among other things, the notefile is a way of storing information in a compact way. Let's say that you were doing a search of the subject of biology, so as you were searching you named the notefile "biology", and named different savefiles as you went to keep track of the text. The "biology" notefile will contain a record of all the file pointers from your whole search of biology. If later you want to review the search results from the notefile, then you could do the following: è 1. Enter the LEXSET program. 2. Type 'n' and change the default notefile name to "biology". 3. Type 'L' for LEXNOTE. 4. The first textual aggregate connected with the first file pointer contained in that notefile will appear, with the LEXNOTE menu. 5. If you wish to just run through all the text, then select 'c' for continue from the LEXNOTE menu and it will take you from aggregate to aggregate, providing a review of the entire run of searches which you archived into this notefile ("biology"). Similarly if you want to review the results in the default notefile, simply enter LEXSET, type 'L', and you will be into the $AVE-1T.--- file, as it is just that, the default notefile. è VIII. USING THE OPERATING SYSTEM INTERFACE PROGRAM When using LOGIC-LINE, there will be times you need to use certain operating system functions before you are ready to exit the LEXSET search program. When you use 'q' to exit LEXSET, you immediately exit back to the operating system; it also closes the existing search files. There are many times that you need operating system functions, but you aren't ready to close off your current search. Any time you need certain operating functions but don't wish to exit the LEXSET routine, you can use the specially designed LEXSET operating system subroutine. This is the operating system interface program, and is accessed on the LEXSET main menu by pressing '$'. We have given you a facility which permits you to use a specially designed subroutine to interact with the operating system and its editing subroutine EDLIN, to create and edit various kinds of files. The most common use of this subroutine will be to build cross-search files for associative element convergent or divergent searches. When you want to access the menu of functions in this operating system subroutine, press '$'. The following menu will appear: 1 - CREATE a new text file 2 - EDIT an existing text file 3 - DELETE a text file 4 - DIRECTORY display R - RETURN to LEXSET To make use of '1' and '2', you must know or learn how to use the EDLIN program. Instructions for this should be located in your DOS operating manual. The '3', '4', and 'R' functions are self-explanatory. è IX. BASIC PROCEDURES PREPARING THE FILES TO SEARCH: There is no particular organizational or systematic procedure required to prepare the files or data base you wish to search. LOGIC-LINE goes straight to the information in the files, using natural language to process it. Any information you have available which is in standard data base form, which was entered in non-document mode, is ready to search immediately. Where you wish to search files which have been entered in a specific document mode, such as belonging to some word processing programs, you may run into difficulty as you are trying to search through an encryption which was entered with the data as it was word processed. In this case you will have to strip off that encryption so that the files can be read by this (or any) searching program. There are programs which do this, as a technical procedure, and they can usually be obtained from the company who sells the word processing program. The above, it should be noted, is purely a technical procedure, and does not require a person who is trained in the content of the database to reorganize it or re-comprehend it. In other words, if you are working in a law office and need to ready the files, you can probably get a computer technician to handle it rather than a legal assistant. If there is any difficulty with this and you do not know what to do to get going, please call the hotline number on your software registration card, and we will assist you or advise you as best we can. When files are entered where it is known that these are files which will be searched by exterior programs, it is generally understood that they must be entered in non-document mode. You are likely to encounter difficulty only when searching files which weren't originally intended for searching. The beauty of the LOGIC-LINE program however, is that you can still search these files without first preparing or systemizing them, as soon as you have stripped off whatever encryption was inadvertently entered with the information. èSEARCHING IN-HOUSE RECORDS Interesting and useful applications for LOGIC-LINE 1 are to use it to search your in-house records, whether these be patient reports in a doctor's office, legal depositions or testimonies in a law office, all the data dumps you got from a larger data base over the last year, articles or stories you have written, or your letters out from the last 6 months. LOGIC-LINE 2 lets you additionally search your spreadsheets, tax records, codes and formulae, real estate descriptions and measurements, and whatever else you might have around. The list of what would be useful to search is limited only by your imagination. When you undertake this, the first thing to consider is what form the information was entered in. If it is available on disk, in regular ASCII form, it can most likely be searched by this program. It may be, however, that the first thing you have to do is to strip off any special coding or encryptions which prevent the information from being searched. This is often the case as a matter of simple security in many companies. As mentioned above, this is a technical matter which those people who constructed the information will know about. Presuming that you have now got information which was entered in non-document mode ready to search, or that you had your files stripped so that they can now be searched, you are ready to go. Decide what kind of search you want to do and use these basic procedures to get going. Start with one file and work on that until you understand the procedure. Very often, as in the case of many different patient records, or letters, or depositions, you are confronted with a very large data base, but which is composed of many, many separate short 1-3 page files. Once you have gotten the procedures down so that you can better concentrate on the information you are searching for, you will probably find it unwieldy to continue to handle all these short files separately, within the framework of the LOGIC-LINE main menu. In this case, find out from the word processing manual that goes with your word processor, what procedures allow you to put several short files into a long file. It is usually very easy to do this. èOnce you have figured out how to do this, and what you want to compile, then make a long file called, for example, "LETTERS", or "DEPOS.84" or "PATIENT.RPT", or whatever you like, composed of all the short files. (See the chapter Notes on Hashfiles under the section Naming Files for rules on allowed names.) This will assist you in keeping a minimum of attention on the procedure, while being able to concentrate on the information you are trying to analyze, what you are trying to find out, and accomplishing the goal of your search. The following procedures begin once you have typed in "LEXSET" followed by the "Return" key, have entered the program, and are looking at the LOGIC-LINE main menu. STRAIGHT 'g' SEARCH: DIRECT KEY WORD/FILE SEARCH (SIMPLE WORD SEARCH) PROCEDURE: 1. Select 'm' for manual mode or 'a' for automatic mode. NOTE: The default setting is 'm' for manual, if no selection is made when the LEXSET program is first entered. This is important to note when doing a cross- search you wish to interact with, as if you start the search on automatic you will have to wait til the search completely runs its course, which in some instances can take quite awhile. That is why the default is manual mode, to prevent this from occurring. 2. Optional: If you wish to name the files you will be searching, then type 'n' to name savefile and notefile. 3. Type 'f' to input the first file you want to search. An 'f' will appear on the screen followed by an instruction to enter the file name. Enter the name, followed by 'Return', and that file name will appear at the top of the screen under the heading "FILES TO LOOK IN". Repeat that procedure, typing 'f' first, the filename and then 'Return' for each file you want to search. è 4. Type 'w' to input the first word or keyword you want to search the file or files with. A 'w' will appear on the screen followed by an instruction to enter "word to add". Enter the word or keyword, followed by 'Return', and that word will appear at the top of the screen under the heading "WORDS TO LOOK FOR". Repeat that procedure, typing 'w' first, the word, and then 'Return' for each word you want to search for. NOTE: It does not matter which sequence you use to enter words and files. The words can be entered before the files if you like. 5. Once you have the file or files selected and key words chosen, then type 'g' for "Go" and your straight search will begin. The cursor will flash on the screen from time to time while the program is searching; don't interrupt this action until the search is complete (as in auto matic mode) or until the first entry has come on the screen (in manual mode). You will know the search is complete when the LEXSET main menu reappears on the screen. The word that was searched for will also now be in capital letters if it was in small letters before. NOTE: Although you can enter words and files in lower or upper case letters, you may wish to use lower case as a general operating rule. This way you will have an additional memory cueing device to follow along your searches by noting if a word has changed from lower to upper case. If the word is now in upper case letters, then the search has been completed. 6. If you were searching in manual mode, the program will stop when the first data aggregate is presented, and a menu will be presented to you. Select one of these choices: Y for Yes, S for Skip, A for Abort, or L for Lexnote, to continue the search. è 7. If you want to save the results of the search, then type in 'r' (for "Result") to archive the results of the search. If you gave a name to the savefile and notefile before the search began, the results of the search will transfer to a file under the names you designated, and will now appear on the directory. These files are ready for looking at, editing, or printing, as you wish, and handle just like any other ASCII file. This step is optional, and you can continue with other searches without it. ASSOCIATIVE ELEMENT 'c' CROSS-SEARCH PROCEDURE: 1. Follow Steps 1-5 for a straight 'g' search, or direct key word/file search, and build a "LOOK.AT" file of a few data aggregates which constitute the basic concept or idea you wish to align your entire forthcoming associative search around. It is best to keep this search rather small; 1 short or long aggregate, or 2-3 short aggregates is best to keep the search precise. NOTE: You may have to experiment with a few 'g' searches first using no more than 1 word or so, to define as you want it the concept the search will be building from. If so, use the 's' command to clear the screen and start over until you get the point you want to start from just right (i.e., the 'g' search results that go into the "LOOK.AT" file). Once you know exactly what search parameters you want to use (in terms of words and files for a 'g' search), you may need to make sure nothing else from the previous (trial & error) searches is in the retrieved "LOOK.AT" file as a starting point. In this case, simply type 'b' to "begin extract", and the LOOK.AT file is automatically cleared. The message "done" appears when the action is complete. è Then repeat straight search steps 1-5 outlined above with your selected search parameters. 2. Type '@' to access the associative element menu and set the type of association that will be used in the search. 3. Choose '+', '=' or 'a', as covered earlier under Types of Searches (provided you choose '=' for the first associative element search you execute). 4. Select associative file. Enter the cross- file name that will be being used to search with. At this point in the proce- dure you cannot create new files, so you must use a crossfile that has already been created and put into hash form. 5. Type 'R', "Return to LEXSET", to return to the Logic-Line main menu. 6. Type '*' for "use results". This directs the searching mechanism to use the contents of the "LOOK.AT" file you set up in the first straight 'g' search. 7. Type 'c' for cross-search, and your search and retrieve process will be underway, using the crossfile already designated to search and retrieve information from the designated files. 8. You will know the search is complete in the same manner as is covered in the straight 'g' search, under Step 5. When the search is complete for a cross-search, the file will appear at the top of the screen under that category (Crossfile was:). The previous straight search Step 6 regarding the manual mode menu, and Step 7 regarding use of 'r' for "archive results", apply the same in a 'c' cross- search as in a straight 'g' search. NOTE: The '*' command when used with (just before) the 'c' command, is a powerful tool. Successive applications of *c ("star-c") commands, i.e., the same search over and over accumulating a larger and larger LOOK.AT file as you go, permits logical combinations to be discovered, and can be very helpful when studying a large data base. èA PRIORI 'c' CROSS-SEARCH PROCEDURE: 1. Do Steps 1-3 of the straight 'g' search; that is, selecting 'a' or 'm' mode, naming files if desired, and entering files you wish to search. 2. Type 'c' for cross-search. The letter 'c' will appear on the screen, followed by "enter cross-file name". 3. Enter the crossfile name you wish to search with (that is, WITHOUT entering '*' for use results first), and as soon as you type in return to enter the crossfile name, the search will be underway. 4. Step 8 under Associative Element 'c' Search applies the same for an A Priori 'c' search. NOTE: This a priori type of 'c' search differs from an associative 'c' search in that it is searching the file for those elements that respond directly to the pattern of meanings in the crossfile; rather than the more complicated associative element searches which are wider in scope, entertaining more possibilities of inference. Experience and familiarity with the system and the different types of searches is ultimately the best guarantee for successful and accurate data retrieval. Databases and writing styles vary from author to author and different search types lend themselves to different data bases. Experience will give you judgment as to when to do what and how. CLEARING THE LOOK.AT FILE As you search, information is put into various files which you can see on your directory. Textual and file pointer information (which marks where in the file the references were found) goes into various files such as the LOOK.AT file, the LOOK.LAS file, the $AVE-1T.TMP file, the $AVE-1T.--- file, and such. èYou can check the content of these files from time to time if you are interested by using the 'TYPE' command from your operating system when outside the LEXSET program. (See your operating system manual for more information on this.) For example, you can enter the command: "TYPE LOOK.AT [Ret]" and you will see what's in it. If you didn't use the 'n' ("name files") function from the main menu to name your files before a search, you can still capture the results in this way. What's most important about these files is that they are used in recursive 'c' searches by the program itself as it searches. In other words, the searching program learns as it goes, using what it has learned as it seeks out each new aggregate of information. This is the way in which it knows (most of the time) when it has found an aggregate of information that it already presented, which shows on the screen as "KNOWN". Often you will get a string of "KNOWNS" as the program looks carefully as hard as it can for any new entry. Occasionally it will show the same aggregate more than once, and if you are in manual mode you can simply "Skip" it with the 'S' command, which prevents it from being saved in the savefile. There will be times when you want to begin a search fresh, with no arbitrary information to direct the current parameters. In this case, you will want to clear these files before setting the search parameters of the new search. The 'b' key will clear out the content of those files which affect subsequent recursive searches; i.e. those associative element searches which make use of the look.at file to accomplish the search. As this is an internalized function in Logic-Line, the procedure is simple and doesn't require exiting the LEXSET routine. 1. Type 'b' for "begin extract". 2. You'll know the action is complete when the message "DONE" appears on the screen. SETTING UP A CROSSFILE PROCEDURE: You must have a crossfile ready to use before you hit '@', '*', or 'c'. (See Chapter 7: LEXSET Command Summary for more informa tion on those commands.) The best way to create a short crossfile to search with is: 1. Type '$' to access the operating system interface. è 2. Follow the instructions in Chapter 9 for how to make use of the EDLIN program to create a file. 3. Prepare the crossfile for use in the search by hashing it. To do this, simply type 'h' for "hash/rehash". If you edited an old crossfile, you can also use 'h', which will delete the old hasfile and rehash it. è X. NOTES ON CROSSFILES Where a 'g' search searches for the pattern created by one word, a 'c' search searches for the pattern created by a whole file, in relation to the whole text. Thus the content of the crossfile you use to do your search makes quite a bit of difference to the results of the search. As an associative element 'c' search is a different process from an a priori 'c' search, the ideal crossfile for each differs somewhat. We have found that the following information and rules are helpful in creating and naming crossfiles. ASSOCIATIVE ELEMENT CROSSFILES The crossfiles used for an associative element search are best configured as one word, or at most 1-3 word files. The idea here is to crystalize a concept that is being used for the cross-search. The facility really works best when only one word is put into the file, or similar words that really get across the same thing. As this is the case, we have found it convenient (though it isn't technically required for you to do it this way) to give a name to these cross word files which is a direct mental cue to what you are trying to do. Thus we generally give all such files a name which begins with the actual word (or concept) in the file with the extension [---.x]. For example, if you want to do an intelligent associative element cross-search on the concept "good", you would make a file with the total contents of [good.], and name the file "GOOD.X". If you wanted to do such a search for the concept of "children", you might make a file with the total contents of [children.], or even containing [children son daughter.], and you would name the file "CHILDREN.X". It is a good idea to prepare several such [---.x] files ahead of time, before you begin addressing a large data base which you want to dig into. Then you have your tools all ready to mix and match with during the search; the object being to find out whatever it is you want to know. A search should first be done locating 1 or 2 key aggregates which will make up your LOOK.AT file. The concepts in the LOOK.AT file constitute the foundation of meaning which becomes the basis which the program runs the [---.x] files against. You can set up [---.x] files which are dichotomous in meaning when compared to the contents of the LOOK.AT file, or which are parallel to the contents of the LOOK.AT file and modify and expand upon it. èThe process of the associative element search is somewhat analogous to the process of multiplication. You start by retrieving 1 or 2 aggregates of data using a key word. Then by setting up another file (the associative element [---.x] crossfile), you are multiplying the overall meaning-content pattern of the LOOK.AT file by the overall meaning-content pattern of the [---.x] file, producing various aggregates of information. In doing this you will be able to find inference strings or lines of logical reference which will possibly relate. You then select those that do relate, discarding those that don't relate. Experience in doing various associative element type searches will teach you what to expect as a rule. See the Model Searches Chapter in the full LOGIC-LINE 1 manual for reference to actual searches done using associative element crossfiles, and as an example of what actually occurs. A PRIORI CROSSFILES When you do a search in this manner, you are making use of a rapid cross-search facility which can be used to process natural language queries, as well as single 1-3 word associative element files, on up to using one text file to search another text file (though this would result in an all associative search more so than anything else). The most usual crossfile you would use would be a rather simple one line sentence, or question, directed as an interrogation to the database. For this reason, and to differentiate this type of crossfile from the [---.x] files described in the foregoing section, we give these types of files names with query numbers, like QUERY1, QUERY2, QUERY3, etc. For example, supposing you want to ask the following preliminary questions of a file: "What do you think about war?" "Give me the statistics on divorce." "How does the system operate?" You might want to use these questions as a straight a priori 'c' search before getting into more precise search parameters. Create a crossfile for each of the above. The content of the first example above would be: [What do you think about war.], and you can name it "QUERY1". The content of the second example above would be: [Give me the statistics on divorce.], and you can name it "QUERY2". And so on. Then you would of course put each QUERY-- file into hash form with the 'h' ("hash/rehash") command so that they are ready to be used in a straight a priori 'c' search. èThe above is the most common example of this type of crossfile and how to name it. Another use might be to search for a list of things, like a file whose content is: [red orange yellow green blue indigo violet.] Such a file you might name for its content, giving the name "RAINBOW". As this still differs in approach to a [---.x] file we would recommend keeping the name "RAINBOW", rather than "RAINBOW.X". How you name your crossfiles has more to do with your own mental cueing patterns than with the actual operation of the program. But we have found that an orderly system of creating and naming your crossfiles, such as is given here by way of example, is assistive in making the most of this searching program as a tool for decision support. SETTING UP CROSSFILES FOR USE The basic rule is that any crossfile before it can be used for a search must be hashed. If for any reason you edit or make any changes at all in a file, including changing the name of the file, you must: 1. Delete the old hashfile. 2. Re-hash the crossfile. Both of these functions (1) and (2) above are taken care of by the 'h' key, for "hash/rehash". However, you must remember to execute that function when needed. It's easy to forget this as you are heavy into a train of thought on a search; but if you find the program isn't doing what it ought to, suspect that a text file was tampered with or edited without re-hashing the corresponding hashfile, and re-hash it with 'h'. è XI. NOTES ON HASHFILES A HASHFILE is a specially encoded shorthand representation of the contents of a regular textual input data file on your disk. It is a hidden file in the sense that the user is not expected to ever access this file directly under any circumstances. In order for the LEXSET searching facility to work, the first step is to create a HASHFILE from the regular textual file. When this action has been done you will know it, because it will appear on your directory with the extension [.HSH]. For example, let's say you are working with a file called FORESTS.DOC, or simply FORESTS. When a HASHFILE has been created for this file, or when that file has been "hashed", the HASHFILE for FORESTS.DOC will appear on the directory as FORESTS.HSH. If you are working with many different files you'll find your directory getting quite long because for every file, which still exists in its original textual form on the disk, there is a counterpart [.HSH] file. The way you create a HASHFILE is by typing 'h' for "hash/rehash". The message "DONE" will appear when the action is complete. You don't always have to execute the 'h' hashing function to create a hashfile. Hashing is automatically the first step of a straight 'g' search. So where you are executing a straight search on a file for the first time, simply execute the search, and the hashfile will be created. At the same time, while it may appear that you can hash a file by using the 'c' command, it will not be correctly done and will cause errors in the search. So if a textfile has not yet been hashed, and your first search will be an a priori 'c' search, then hash the textfile first with 'h', then do the search. The first time a file is searched it will take some time, because this hashing action is being done. A relatively short file only takes a minute or so; but a reasonably large database or very long file (e.g., 100+ k) can take 5 minutes, perhaps longer. The second time a file is searched, only the search need be done; not the hashing plus the search. Thus, after the initial hashing, a search can be completed extremely quickly. RE-HASHING FILES: It is very important not to tamper in any way with the HASHFILE, as it will result in an incorrect search and may adversely affect the LEXSET search facility. èIt is not uncommon that data bases and files get added to, edited, or otherwise revised. However, realize that any time you make any change at all, however slight it might be in the original textual file, it will make the old .HSH file invalid. Thus keep as a rule that any time a textfile is edited in any way, the corresponding HASH file [.HSH] is erased and re-hashed before the LEXSET program is used again in conjunction in any way with that datafile. Use the 'h' "hash/rehash" facility to do this. Note that this rule applies to the short 1 to 3 word cross search files that you use with the 'c' command to execute an associative search, and the perhaps longer a priori crossfiles, as well as to the textfiles; if edited, you must remember to use the 'h' key to rehash that crossfile before it is again used for searching. NAMING FILES Generally speaking, when you assign a name to a file, you simply use the same rules that apply to naming ASCII files. See your operating system manual for more detailed information on this. In addition to the above, there are a few rules that apply specifically to use with the LEXSET program, and this is chiefly in regards to the 3 digit extension on the name. This is because every file in use has a corresponding hashfile, which has the same first 1-8 letter name, but differs only in the extension, which becomes ---.hsh, as covered earlier. Thus, you cannot create 2 files whose names differ only in the 3 digit extension (.---), as the search program will map into the same hashfiles and get stuck. Filenames must differ from each other somewhere within the 1-8 digit specification. There are also some special names that must not be used when naming files. They are: ----.HSH LOOK.AT LOOK.LAS Also, the SAVEFILE must have a different name from the NOTEFILE. The program will not prevent the user from doing any of these things, but it is a very bad idea, as the search mechanism will become confused and won't work properly after that. èINTERRUPTING A SEARCH It's unwise to interrupt an automatic mode search facility by taking the disk out or otherwise preventing the search facility from completing what it started. While it can be done, it has the potential liability of affecting the search mechanism. It is best to let the search complete itself if in auto mode. In manual mode simply wait for the next data aggregate to come on the screen, which will be followed by an option to abort if necessary. Using the abort facility in manual mode is perfectly fine, and was designed so that the user would have control over the direction of the search most of the time. Use auto mode when you know you have time to let the search complete. On a large file this can sometimes take 5 minutes or more, the first time the file is searched. Use manual mode when you want to be involved in the search as it is being done, and want to control, steer, and change direction as you're doing it. In manual mode you can abort any time you like. If you do have to interrupt an automatic search, the remedy so that that disk can be used again is as follows: 1. Clear the LOOK.AT file with 'b'. 2. Use 'h' to rehash all the text files that were in any way connected with the search you just interrupted. è XII. WANT TO KNOW MORE? Do you still have questions? Want to know more? Curious about the background technology? Want to see a bibliography? Would some drills be helpful? How about screen by screen model searches to learn each search type? Want to know how to use Logic-Line to assist in building expert systems? Curious what other products Thunderstone has to offer? Take a look at the file called ORDERFRM.LL (that stands for "Order Form Logic-Line", as different from the FORM.ESI file for "Order Form ESIE"), and you'll see what additional support aids and products are available from Thunderstone, for how much, and where to get them. Print out the ORDERFRM.LL and an order form is all set for you to send in to us, and tell us how we can help you.