Puzzling search results

[aschecht]

A user searched for "tinea v*" (no quotes) and got the result

1 result found containing all search terms. 686 results found containing some search terms.

But neither the first result (which should have been the one with "all" search terms) nor any of the other results on the first page of results included both search terms even though there is one (and only one) page that has both tinea and a word starting with v (v*).

The user was looking for tinea versicolor. I recreated the search with the following search terms:
tinea ve* - similar results as above
tinea ver* - similar results as above
tinea vers* - correct results with tinea and vers* found on first page.

You can re-create this at:
www.icdmeister.com/subscriber/
using username: tester and password: tester (valid through 7/31/0

I'm curious why it says it's found both search terms but doesn't list this page as the first search result even though I know it's there.

Not a big issue but I'm just trying to understand the intricacies of the search process.

Any ideas?

Andrew

[Ray]

It doesn't look like you have enabled the "match all search words" option which appears to be what you are expecting.

This means it is currently searching with "match any search words", and thus a page is not required to have all the search terms to be included. This is why the results show many pages which only include a word starting with v (v*) and not tinea.

If I explicitly specify the "match all search words" option through this URL:
http://www.icdmeister.com/subscriber/search/search.php?zoom_query=tinea+v*&zoom_and=1

The results appear to be what you are expecting and that it is working fine.

You can also do this on your own site through the "Advanced search" you have made available, and changing the selection from "match any search words" to "match all search words".

[aschecht]

Thanks Ray. I was under the impression that if the search is done as "Match Any Search Words" then the search results are displayed first listing those that matched ALL search words followed by those that matched ANY search words. This makes intuitive sense as one would be casting a wide net but would have the most specific results (those matching all terms) showing at the top of the search results list.

In most searches I do, the ordering of results seem to follow this pattern (example: search for "acute bronchitis" (no quotes)). In the example specified in this thread, it does not follow this pattern. Is it something having to do with the wildcard?

Andrew

[Ray]

Yes, and yes. Words matching more search terms will appear higher in the results even when you have "match any search words" enabled.

However, when you use wildcards, this is no longer obvious. The number of search terms become indefinite because "v*" can potentially match many words, and each word found here is considered to be a different search term.

[aschecht]

This answers my question. Thank you.

With respect to the seach in question, the results would be more useful if, despite the wildcard in one of the search terms, the results ranking somehow preserved the ability to prioritize pages that had ALL search terms (the non-wildcard search term AND the wildcard search term) over those that had either/or. If feasible and you think this is worthwhile, perhaps it could be considered for a tweak in an upcoming version.

Andrew