I’ve been working on the cwdb’s search engine capabilities recently. In particular, I aim to generalize the search system to include the filtering botch in the last revision.
Mode00: http://cwdb.azaphrael.org/search.php
The original search engine. Quite crappy. Referenced here for comparison.
Mode01: http://cwdb.b4k4.ath.cx/search
The first revision, used as a benchmark on the new system. Uses MATCH() AGAINST() fulltext searching.
Mode02: http://cwdb.b4k4.ath.cx/search2
The second revision, this method splits the authors and types off into arrays for filtering, and uses MATCH() AGAINST() IN BOOLEAN MODE to search normal keywords. Permits +/-/quoting phrases. Permits on-the-fly grouping.
2 Comments
The grouped results are easier to read. Still easier to read would be if you distinguished wrap-around lines from listing starts; there should be CSS dirty tricks to ‘indent’ the first line to the left.
See my comments about learning metadata elsewhere in this blog.
In general, I like the idea of using cwdb’s specialist/database knowledge to improve search results. Do you have heurestics for “was the search on an author” type stuff?
Per-list-item first-line indenting has been styled.
By ‘heurestics for “was the search on an author” type stuff’ do you mean something akin to search/_Quinn?