Table of contents:

ABBYY RETRIEVAL & MORPHOLOGY 4.0 ENGINE

ABBYY Retrieval & Morphology (ARM) Engine is a Software Development Kit for integrating advanced ABBYY full text index, retrieval and linguistic technologies into local or client-server applications via 32-bit COM-style API. The Engine supports 34 natural languages.

Where to use

Text Retrieval API of ARM Engine allows developers to integrate quick and efficient means of retrieving needed documents from heaps of text information stored in such electronic repositories as corporate information systems, CD archives or WEB servers into their applications. 

The Engine makes complex task of retrieval easy by creating compact index and using efficient search algorithms and morphologic knowledge bases of natural languages.

Morphology API of ARM Engine allows developers to integrate functions of morphology analysis (spell checking, lemmatization, paradigm generation) into various systems of automatic processing of electronic text information: text search and retrieval applications with morphology analysis, classifiers of text information (e.g. ôspamö protection), etc.

Proven technology

The ARM Engine is a proven tool for application development. ABBYY uses its perfect full text retrieval and morphology technologies in the following ABBYY end-user products:

• ABBYY Lingvo English-Russian-English electronic dictionary system was the first commercial product that used ARM technologies. Thanks to these technologies tens of thousands of Lingvo users all over the world can translate any word they need in a fraction of a second. 
• ABBYY FineReader, new generation of intellectual OCR systems and unrivalled winner of international comparative tests, successfully uses the ARM technologies for the context analysis and spell-checking. 
• ABBYY Internet Index Server application based on ARM technologies allows efficient multithreaded search in Internet and now performs successfully at ABBYY's web-site.

Success Stories

The quality of ARM technologies is highly reputed on the world market. Among the companies successfully integrated the technologies into their products are:

• Siemens Nixdorf uses ARM technologies as component of ARCIS electronic archive û complete system of archiving, information processing and record maintenance. 
• ôNezavisimayaö Newspaper, one of the most readable Russian newspapers (total circulation including appendices makes hundreds of thousands) uses ABBYY Retrieval & Morphology Engine in its annual publication CD archive. 
• ½Optima╗ integrated the modern and intellectual ARM technologies of text processing into its OPTiMA-WorkFlow system, considerably enlarging the systemÆs capabilities of electronic document archives processing.
• ½ALESTAö integrated ARM Engine into its products Alesta Search for BackOffice, Alesta Library and Alesta Search Intranet, providing these products with tools of full-text search in Russian and Ukrainian languages. 
• ôLANITö uses ABBYYÆs full-text search module as a part of its document management system LanDocs.

Functionality

Please note: ARM Engine doesnÆt have any elements of user interface.

Full Text Retrieval API features:

• Full Text Indexing. The system creates compact and efficient full text index of specified multi-language documents. The index stores the words and their locations within 
  an information repository (file servers, web servers, databases). In order to create a compact index and provide the fastest search possible, the system uses morphology analysis, lemmati-zation and stop-words dictionaries. An example: lemmatization for Russian language allows the search system to index not the entire 4 000 000 wordforms, but only 150 000 word stems. Stop-words dictionaries contain the set words insignificant for searching (prepositions, conjunctions etc) that are not included in the index so it becomes shorter and more efficient. As a matter of fact, there are many parameters that affect the speed of indexing and relative index size, just to name a few - computer configuration, structure initial information base, format and languages of documents etc. The following is an average estimate for indexing of ôplain textsö in TXT format and in one language:

 

	Speed of indexing	The ratio of index size to information source
Without lemmatization	20 MB/minute	40%
With lemmatization	10 MB/minute	30%

• Full Text Search. This API allows one to enter a word in any form and get all the word entries in all its forms in all indexed documents. ARM Engine performs both simple and advanced search. In the simple search a single operation (AND, OR or NOT) with the words to be searched is specified. As the index contains information about coordinates of each word in the set of indexed documents, one can specify the distance between words to be searched and their relative position. Also the search for word combination is supported. An advanced search uses a special query written in special ½query language╗, which allows one to specify very precisely the search parameters using arbitrary combination of logical operators. Both queries, as they are based on a compact index, Boolean logic, wordform and stop-word dictionaries, perform extremely efficient search even in multilanguage documents û the time in which the user gets results is just some fractions of second, so the user may not even notice that some time has actually elapsed.
• Fuzzy Search. ARM Engine performs fuzzy search, i.e. search for words which differ from the ones provided in the query in one or several characters, otherwise called ôsearch for like wordsö. This feature is valuable if one either searches documents with a certain number of misprints, or is not sure of a wordÆs spelling or if one just wants to find words similar to the one provided in the query.

• spell-checking and spelling suggestions for words written incorrectly;
• lemmatization - generation of the primary form of the word (the form in which the word is present in a dictionary) from any given form of this word;
• paradigm generation (i.e., generation of all wordforms) for any given form of word;
• inflection type detection for not in dictionary words;
• grammatical value detection for any wordform;
• creation of updateable user dictionaries.

Supported languages

ARM Engine supports 34 languages:

Indexing and search through multilingual documents (any combination of the above-indicated languages) is available.

Supported formats

The ARM Engine allows to index documents in following formats:

• TXT;
• TXT Unicode;
• HTML.

• Russian - Windows-1251, KOI8-R, Cyrillic ISO (ISO-28595), OEM-866;
• Czech, Polish, Hungarian, Slovak, Croatian, Romanian  û Windows-1250;
• Bulgarian, Ukrainian û Windows-1251;
• Greek û Windows-1253;
• Turkish û Windows-1254;
• Estonian, Latvian, Lithuanian û Windows-1257;
• Armenian û no ANSI-format supported;
• other supported languages - Windows-1252.

The ARM Engine was successfully tested for compatibility with the following programming platforms:

• Microsoft Visual C++ 4.x and above;
• Microsoft VB 5.0 and above.

• PC with Intel 486DX or higher processor;
• Windows 95, 98, ME, 2000 for client part and Windows NT 4.0, 2000 for server part;
• MS Internet Explorer 4.x and above for registration of COM-servers;
• 32 Mb of RA╠;
• from 20 to 105 Mb of free hard-disk space for installation (depends of the languages installed);
• from 10 Mb of free space to hold temporary files (depends on size of the index base).
  

* - The prices include 3 developerÆs licenses by terms of the End-User License Agreement for SDK.

- The prices include technical support of developerÆs engineers.
- The prices do not include shipment costs.

** - For receiving the evaluation version a Non Disclosure Agreement for Trial SDK must be signed.

We offer TIME-LIMITED FULL-FUNCTIONAL ABBYY Retrieval & Morphology 4.0 Engine for FREE evaluation to our prospective customers, so that they can try the software out in real conditions without any functionality limitation. This Trial version works for 60 days in full-function mode. After that, all product features shut down. The terms of the Trial version use are specified in the Non-Disclosure Agreement (NDA) for Trial SDK. To obtain your evaluation copy from our FTP-site please download the NDA, fill it in, print out, sign and fax at one of the following numbers: +1 888 467-4151, +49 89 921-85173. After you do it, please, let us know to engine@abbyy.com, and we will reply you by e-mail with the coordinates of the Trial version.

Terms of use and Runtime licensing

DeveloperÆs License of ARM Engine grants a customer of the SDK the right to use ARM Engine for developing only or internal use of the developed applications only. If the customer wants to distribute his applications with ARM functions included he must sign an additional Runtime License Agreement with ABBYY. The Agreement specifies royalty payment for each distributed copy of developerÆs application with ARM Engine libraries inside, as well as other distribution terms. Please contact ABBYY for such Agreement.

To a developer expecting to distribute less than 500 per user licenses of ARM-enhanced application per year, the following Runtime licenses packs are available*:

*If your prospective sales volume is over 500 per user licenses per year or your application has per server or another licensing model the Runtime license price is to be negotiated. To quote the price, please contact ABBYY.

You can order ARM Engine DeveloperÆs and Runtime licenses by contacting ABBYY.

ABBYY Software House (headquarters in Russia): engine@abbyy.com (Worldwide)
ABBYY USA: sales@abbyyusa.com (North America)
ABBYY Europe: engine@abbyyeu.com (Western Europe)
ABBYY Ukraine: developer@abbyy.com.ua (Eastern Europe)

Updated: