全文搜索未给出期望的结果

Question

Below are the commands I ran

create database fu;

create table table_name( name varchar(10));

insert into table_name values('karan');

insert into table_name values('nitin');    

insert into table_name values('orip');

insert into table_name values('karan orip');

insert into table_name values('karan nitin');

alter table table_name add fulltext(name); //fulltext

select * from products where match(name) against('karan');

Now, the above query returns me empty set. Why is that?

Also, is I do

select * from products where match(name) against('karan' in boolean mode);

The above statement gives me perfect result.

Answer 1

You seem to use the MyISAM storage engine. There's a limitation that words that are found in more than 50% of all rows will be treated as stopwords:

Your search word 'karan' is found in 3 of 5 rows, so it's over this mark.

MyISAM Limitation
For very small tables, word distribution does not adequately reflect their semantic value, and this model may sometimes produce bizarre results for search indexes on MyISAM tables. For example, although the word “MySQL” is present in every row of the articles table shown earlier, a search for the word in a MyISAM search index produces no results:

[...]

The search result is empty because the word “MySQL” is present in at least 50% of the rows, and so is effectively treated as a stopword. This filtering technique is more suitable for large data sets, where you might not want the result set to return every second row from a 1GB table, than for small data sets where it might cause poor results for popular terms.

You can get around this issue by using the InnoDB engine if you're on MySQL 5.6 or newer.

The 50% threshold can surprise you when you first try full-text searching to see how it works, and makes InnoDB tables more suited to experimentation with full-text searches.

from MySQL manual, Natural Language Full-Text Searches

Answer 2

Because of limitations and performance issues in MySQL built-in full-text search indexes I would suggest to use external full-text engine like Sphinx or Lucene/Solr. Both of them will give you much more speed and way better functionality and relevance. This is will be mandatory if you plan to search against big amount of data, in this case MySQL FT search can take seconds to complete while external systems which are based on inverted index could search gigs of data withing milliseconds.

Solr is written in Java and requires JVM so could be a good choice if you already use Java in your application. Sphinx is written in C++, work as a daemon and support MySQL protocol so could be a bit easier to work with. You can get a sense of how to use Sphinx here: http://astellar.com/2011/12/replacing-mysql-full-text-search-with-sphinx/ Sphinx also support snippets (results highlighting) which could be useful.

In any case when using external search engine you may still want to query MySQL to fetch metadata for found documents.