Page 1 of 1

wordc and langc

Posted: Sat Sep 18, 2004 10:40 pm
by james120
If I want to include numbers in my definition of words, is this what I would do:

wordc = [\alnum\']
langc = [\alnum\' \-]

And, what is your experience with how much of a performance hit this might cause? I am indexing technical literature, and it would be somewhat common for various things to be named with a combination of letters and numbers.

Finally, are these properties (and the rank knob properties, which I am not going through) set in the same manner as the Search and Optimization parameters (that is, with a SQL query of "SET property = value;")? If yes, can they be overridden on a per-query basis? I am mainly interested in knowing if users can choose the likepproximity and likepleadbias which they feel gives them the best results, without affecting the settings for anyone else.

wordc and langc

Posted: Sat Sep 18, 2004 11:08 pm
by John
Wordc and langc are used at query time, and the index expressions via addexp at index time. By default numbers are indexed. The basic effects of adding numbers to wordc and langc is that word form processing such as suffixes will be applied to search terms with numbers, and in case a linear search is needed it will look for entire words rather than substrings for search terms with a digit.

Yes, they are all set with the SQL SET command, and like all those settings apply for the duration of that session, e.g. while tsql keeps running, or a single web request. They can not affect other users.

wordc and langc

Posted: Sat Sep 18, 2004 11:35 pm
by james120
So if I am not maintaining state (perhaps with the vortex EXPORT directive, which I am not using), I should set all these options with every SQL query, correct?

wordc and langc

Posted: Mon Sep 20, 2004 11:00 am
by John
They should be set with every invocation of texis or tsql, not necessarily every SQL query if you are doing multiple SQLs in a single invocation.