[Scholar]

I have made a utility in c++ which solves the jumbled words.This is done by comparing all the permutatons of the word given with the words stored in the database.The database consists of around 80000 words .The database is in the form of 26 datafiles one for each alpahbet(a-z).
My problem is that the program takes too much time as it compares each permutation of the word given with the all the words in the concerned datafile.Say, for "euque " ,it will compare it with datafile containing words starting with letter "e". So as the length of the word increases ,the time taken to solve that also increases.
For my core-2-duo 2.0 Ghz processor time taken to solve a 7-letter word is around 8 seconds and for an 8-letter word it is about 1 minute.
Kindly help me in improving the speed of my program so that it can solve at least 11 letter word in optimum time.
For more details of program , you can conatct me at
< address and phone number deleted >

[JosAH]

I don't really understand your question; given a bunch of characters do you want
to see if a permutation of them makes up a word in your database?

kind regards,

Jos

ps. for your own privacy I removed your address details

[arnaudk]

The way you are doing it is probably the most inefficient way, when you look at "euque", you don't need to search through 80000 words in your head to figure out the correct spelling is queue. There are many efficient text matching algorithms out there, googling a bit should yield something much more efficient.

As you pointed out, the first letter can change in the course of finding the correct solution. The length of the word, however, does not. How about sorting your dictionary according to word length instead of starting letter? This has the advantage that, although longer words generate more permutations, there probably are fewer long words in the English language so that will definitely speed things up. Plus you will have fewer files and you will only need to open one file for each new jumbled word.

[JosAH]

I think the OP should work on his database instead. For all the words in the db
make a new db as follows: take a word from your old database and order the
characters in the word alphabetically. That new 'word' is the key; the original
word is a value that is associated with that key. Add the tuple to your new db
and repeat for all the words in your old db.

You end up with something like this (just a tiny example):

aaabnn - banana
aadt - data
adn - and dna
cdku - duck
eequu - queue
...
etc.etc.

Now when you have a permutation, permute it to a key of your database and voila.

kind regards,

Jos

[arnaudk]

Indeed, that would be extremely fast for all word lengths.

[JosAH]

Originally Posted by arnaudk

Indeed, that would be extremely fast for all word lengths.

yup, an 80,000 words isn't that much (*); everything can be done in core speeding
everything up much more again. An STL map could do the dirty work ;-)

kind regards,

Jos

(*) even though 80,000 words isn't much, the entire King James bible uses less
that 32,000 words (including conjugations etc.) and 32,000 tells me: 15 bits ;-)