如何为 sphinx 查询设置更具体的规则？

Question

I have a database with list of movies and tv series. The search I built uses Sphinx.

I use SPH_MATCH_ANY mode, because I want that if a person search for "x-men 2" he will also find entries about "x-men 1" and "x-men 3".

My problem is, it also searches for entries with "2", for example "Cars 2", or "Supernatural season 2".

I want to tell it not to search entries that have no 'x-men" in them. In MYSQL I would write it like this:

MATCH (mname) AGAINST ('x-men') AND mname like '%x-men%' 

or something like that.

In Sphinx I tried this:

$s->Query(x-men 2 "x-men",'test1');

But It still searched for entries with the number "2" in them..

So how can I accomplish that?

Thank you.

Answer 1

Ok problem solved, a guy from Sphinx forum gave me this solution:

$s->setMatchMode(SPH_MATCH_EXTENDED2);
$s->Query("(x-men | 2) x-men",test1);
$s->setRankingMode(SPH_RANK_ANY);

=)

Answer 2

You need to think like a computer... for us is quite obvious that the word "x-men" is more relevant than "2", but for a computer they are both equal.

You could always tell sphinx to index "x-men 1","x-men 2" and "x-men 3" to the same index, using the wordforms

x-men 1 > xmen
x-men 2 > xmen
x-men 3 > xmen

so if someone searches for "x-men 2" he will see the results for the 3 movies. Or you could tell sphinx not to index single numbers, or set the min length for indexes .

But then the users might complain that they wanted ONLY to see the info of "x-men 2" and not the other movies.

In my opinion, when configuring a search engine you don't have to guess what the user wants as a result, because different users will expect different results, but you should always give him the tools to refine his search. For example, one person would expect that x-men 2 returns all x-men movies, another person would expect it to return info related only to x-men 2, and (why not?) someone will expect it to return all movies containing the words "x-men" and "2"... you can't make everybody happy =P

Good Luck