Search Engine?

modred11 · Post by **modred11** » January 17th, 2011, 7:48 pm

What does Librivox use to power it's catalogue search?

I'm curious cause it seems like it could use a few improvements, for example "G.A Henty" has 0 results whilst "G. A. Henty" has 17 (a space between the "G." and the "A." is the difference). It seems like both those searches and "GA Henty" should all have the same results. Google's search engine already does this (coming up with what appears to be the exact same results for all those searches), and it seems like the difficulties with authors having different names could be fixed by giving them "aliases" (removing, or at least greatly reducing, the need for the "Author Name Consistency" thread)

The problem with GA versus G. A./G.A. could be fixed the same way, but ideally it could be fixed by having the search engine try to recognize what the user is doing, recognizing in this case that GA/G.A./G. A. are initials composed of the letters G and then A (this would even allow it to recognize that George Alfred is a match for the initials).

Searching through the previous threads that mentioned search engines, I saw a suggestion for a 'listeners who listened to this also listened to: ...' type of system, and I wanted to point out that Gutenberg has actually done that with their database. It's not really that useful for finding more audiobooks given that Gutenberg has alot of books that Librivox doesn't have readings of, but it's interesting that they both have that system in place, and that they have some of Librivox's audiobooks listed in their database. I don't know what system Gutenberg is powered by, or if it's similar to what powers Librivox, but their search engine does seem to have the same flaw relating to authors with initials in their names.

Post by **annise** » January 17th, 2011, 10:16 pm

There are plans to improve the search function. Having said that , as someone who has done a lot of internet searching I would not think of searching for anything else but Henty as a starter - I would only go to initials or names if I got a large number of hits - but maybe I have just done too much genealogical searching

Anne

modred11 · Post by **modred11** » January 18th, 2011, 2:29 pm

I see so it's custom written PHP for Librivox? I kindof suspected that might be the case

Post by **hugh** » January 19th, 2011, 3:47 pm

yes it's a giant mass of beeswax, php, paperclips, ruby, and spit, love and some swearwords too. custom built.

the plans in the works are to improve the browsing features - which is a front-end thing, and won't be touching (as far as I know) the back-end search engine.

althought tis, who built it, may well suggest some changes.

modred11 · Post by **modred11** » January 19th, 2011, 9:56 pm

That's cool, I know a little php (played around with it on a site my school let me use; but I was literally playing & I didn't get much real experience with it), & I know CSS / HTML / a little Python / C++ &c. I'd be interested in helping out with whatever needs doing if I could actually be any help, though I don't know how many programming people you have or if you need any more?