Did you ever wonder how the MySQL fulltext indexing works? I mean the mathematics behind MySQL's fulltext index feature. At the Epsilon-Delta weblog there is an article by Ted Dziuba entitled "Dissecting MySQL Fulltext Indexing" that explains the mathematics behind Fulltext searching and what it can and cannot do.
The method Ted Dziuba described in the article is a linear-algebraic way of thinking about fulltext indexing. While it is not the exact method described in the MySQL documentation, it produces the same results, and makes any student of linear algebra think that yes, all of those matrix-inversions done by hand were actually good for something.
Link:
The full article: http://epsilondelta.net/2006/02/08/dissecting-mysql-fulltext-indexing/
Fulltext Indexing is a method by which MySQL quantifies the relevance of textual entries to an input query. The concept is similar to the way Google quantifies the importance of pages on the web with PageRank. However, in the context of a database, the requirements of the method are a little different.
The method Ted Dziuba described in the article is a linear-algebraic way of thinking about fulltext indexing. While it is not the exact method described in the MySQL documentation, it produces the same results, and makes any student of linear algebra think that yes, all of those matrix-inversions done by hand were actually good for something.
Link:
The full article: http://epsilondelta.net/2006/02/08/dissecting-mysql-fulltext-indexing/
16 Mar 2007 14:57:12
The article is deleted from wordpress blog :(
16 Mar 2007 15:31:29
Thanks Iliyan,
I found out that the blog is moved. The url is corrected now.