home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Frozen Fish 2: PC
/
frozenfish_august_1995.bin
/
bbs
/
d03xx
/
d0366.lha
/
Makewords
/
Unjumble.doc
< prev
next >
Wrap
Text File
|
1990-08-12
|
4KB
|
92 lines
¢1;33;40m
U N J U M B L E v1.14
¢0;31;40m
Unjumble accepts groups of 3 to 8 letters and tries to make English words
from them.
Unjumble is public domain and may be freely distributed.
¢1;33;40m
------------------- H O W T O U S E U N J U M B L E -------------------
¢0;31;40m
To run Unjumble type "UNJUMBLE" at the CLI prompt, or double-click its
icon.
A brief set of instructions will be displayed, and then a prompt for you
to enter a group of letters. If you enter more than eight letters or
less than 3 letters an error message will be displayed. If you enter
anything that is not a letter (a-z; A-Z) then an appropriate message is
displayed.
After you enter the 3-8 letters a list of possible words will be
displayed. The list will be from 0 to 20 items long. The most probable
words, in Unjumble's opinion, will be near the top, although words will
often show up near the end. A number will appear before each item, which
is Unjumble's opinion of how likely the item is an English word.
To quit, press <RETURN> without any preceeding characters.
The amount of time required to generate the list of items depends on how
many letters are entered. For a stock Amiga 2000:
Number Number
of Seconds of
letters required Permutations
------- -------- ------------
3 0 6
4 0 24
5 0 120
6 1 720
7 3 5,040
8 18 40,320
You can't stop Unjumble once it starts working on a group of letters, but
the longest you will have to wait is about 18 seconds.
Do not RUN Unjumble, as in "RUN Unjumble", because it won't work.
Remember that when you enter a 8 letter word, the odds are one in
2,016 that the word would show up in the list of twenty by pure chance.
Unjumble will beat the odds by quite a bit.
¢1;33;40m
------------------ H O W U N J U M B L E W O R K S --------------------
¢0;31;40m
Unjumble analyzes each string of letters by looking at the trigrams in
it. A trigram is a group of three letters, such as AAA, AAB, ING, QQQ,
etc. The trigram ING is much more likely to occur in an English word
than QQQ, so any item that contains ING is more likely to be a word than
one containing QQQ.
Unjumble creates every possible arrangement (permutation) of the letters
in the string and looks at all of the trigrams in each arrangement. Each
trigram is assigned a weight based on (the log base 2 of) how often it
occurred in a dictionary of 38,500 words. The sum of the weights for
all of the trigrams in a string is the number listed along with each
item. A table of up to twenty arrangements is kept, with duplicates due
to multiple occurrences of a single letter eliminated (e.g., kEEp).
After all of the arrangements have been tested, the table of the ones
with the highest weights is displayed.
Unjumble knows the weights for about 4300 trigrams that occurred in the
dictionary file. With 26 letters in the alphabet there are 26*26*26 (or
17,576) possible trigrams.
¢1;33;40m
-------------------------- W H O D I D I T ----------------------------
¢0;31;40m
Ron Charlton
9002 Balcor Circle
Knoxville, TN 37923
Phone: (615)694-0800
Plink: R*CHARLTON
BITNET: charltr@utkvx1
05-Jul-90
¢1;33;40m
------------------------------ B A S I S --------------------------------
¢0;31;40m
The idea of using trigrams as the basis for an unjumbler came from an
article in BYTE magazine by Bob Keefer (July 1986, p. 113).