Thunderstone Support Forums

In running the three searches below, I've noticed a few differences in the results, which is fine, since the searches are distinct. However, I was wondering if you could give me a few hints as to why each search returns the count it does.

-----------------------------------------
select count(*) from mytable where myfield like '(tiller,maxey,glisan)'

(1200 hits returned)
-----------------------------------------
select count(*) from mytable where myfield like 'tiller,maxey,glisan'

(7 hits returned)
-----------------------------------------
select count(*) from mytable where myfield like 'tiller' OR myfield like 'maxey' OR myfield like 'glisan'

(1100 hits returned)
-----------------------------------------

I'm particularly interested in learing why there's a large discrepancy between the first and second searches, and how the parens play a role in grouping the terms together. Also, could you highlight in a general way the difference between commas in the search terms and OR connectors? Thanks!

Query 1 is doing OR logic: parenthetical (term1,term2) is a set of equivalents to Texis. In the query 2, it looks like it is configured to treat the comma as a separator, and do default AND. From info above not sure why the 3 is different from 1, it may need more detail (open tech support ticket). But 3 is less efficient... a better way to do that in Texis is .... myfield LIKE 'tiller maxey glisan @0' (with intersections enabled).

In query 2, since the list is not in parens, all terms are required, as opposed to any term in the other two.

The reason for query 1 and query 3 differences is that in 3, post-processing is being applied (even though technically not needed), and if there are certain punctuation marks appended to a word, the post-search does not consider it a hit, whereas the index-only search would (in query 1).

Removing ' from wordc would make queries 1 and 3 more consistent.

http://thunderstone.master.com/texis/ma ... ml?q=wordc