fuzzy search

[Scott Runkel]

> I need to implement fuzzy searching option. I have database with
> course information. When a new one is being added, I need to check
> whether a course with a SIMILAR name is already there. Does anyone
> has experience with implementing this type of searches to give me
> tips or references?

This is probably a very difficult task. You could start with just searching for
a match between any single word in the course title entered matches any single
word in the existing course titles. For example, let's say there's already
"Anthropology of Urban Life in New York City 1900 to 1920". Then you come along
with "History of New York". Try to match "history" then "new" then "york" and it
will match. Same with any course that has "anthropology" in it. Etc.

Otherwise you would need to get into word meanings and synonyms and such for
every possible word, and that would be pretty tough.

Another option would be to have a separate field that lists key words or a
description of the course. Then the new course title could be compared to that
field as well, or the new course could have its own description, and the
descriptions' words could be searched against each other and a ranked result
could be returned showing the likelihood that the course is the same.

-SR

Scott Runkel
Director of Technology
Hypix Media, Inc.
503-722-2123