November 16, 2005
Updating our Search Algorithm
Google recently updated it's search algorithm. I happened to be adding a new field to our search indexes (document copyright) and noticed a quick win for helping better ranking of search results in TUSK. So yes, we too have updated our search algorithm.
What was the quick win? People have been complaining for a long time that the search results from our content database are not ranked very well. Primarily that we rely on the full text indexing of DBIx::FullTextSearch which doesn't allow us to adjust document ranking based on other criteria. In many instances, content authors will take time to manually create a set of keywords that summarize the document. This list is stored elsewhere in our database and not indexed. My change pushes those keywords into the index and makes sure that if a search matches a keyword on a piece of content that the content is placed before one that was matched on full-text parsing alone.
This could potentially be a big win for a very small amount of work, people have been complaining for a long time now and we've even started researching new search tools that would give us more flexibility.
Posted by mike at November 16, 2005 11:28 AM