Re: htdig: transliterations


Webstar Technical Support (eesa2@webstar.co.uk)
Thu, 19 Feb 1998 16:55:46 +0000


Hi Andrew,

>There are a couple of things you can do to solve your problem. They all
>involve the use of the fuzzy search algorithms that are part of
>ht://Dig:
>
>1) Use the synonym fuzzy algorithm to set up a mapping of the same
>words with different spelling.
>2) Use the soundex fuzzy algorithm for matching since it will eliminate
>all vowels from words. (The example you gave differed only in vowels,
>so soundex would work although it would probably screw up other
>words...)
>3) Write your own fuzzy algorithm to do exactly what you need to do.
>4) Use any combination of the above.
>
>I would start with #1 above as an immediate solution. After than I
>would investigate #3.

Thanks for that, I think that I will try to start with option 1. What I am still unsure with, going back to my example, if I put in a synonym like:

mohammed mohammad muhammed muhammad

Then if someone searches for "mohammad" they would also get all the documents with "mohammed" in them but would it work the other way round if they searched for "mohammed" would they get the documents with "mohammad"? Do you understand what I mean?

I think I might have a go later at writing my own rules but that won't be for a while as I have too many other things at the moment.

I will let you know how I get on though.

By the way, thanks for an excellent program. :)

Regards and best wishes,

Abdul-Wahid Paterson



This archive was generated by hypermail 2.0b3 on Sat Jan 02 1999 - 16:25:42 PST