Talk:Inverted index
Which Kind of Inverted Index?
Were we supposed to do an inverted file index or a full inverted index? I've implemented both in the Tcl solution even though that does make the code quite a bit more complex (and in production code, I'd use SQLite with its FTS module as that lets you do all this and more). –Donal Fellows 09:56, 21 April 2010 (UTC)
- Nice job. whatever you did can be a "bonus" task I guess...
- I am not sure I understand. What is a "full inverted index"?
- It's defined on the Wikipedia page you linked to. :-) Basically, it's keeping not just the file name/index, but also the index into the file of the word being indexed. What it means is that you can construct a metric over the words in the file so that you can say "this word is next to that one" in a relatively efficient fashion. –Donal Fellows 08:37, 22 April 2010 (UTC)
- Since the specification did not say anything about supporting phrase searching, I imagine that that should be a different task. --Rdm 14:04, 29 April 2010 (UTC)
- It's defined on the Wikipedia page you linked to. :-) Basically, it's keeping not just the file name/index, but also the index into the file of the word being indexed. What it means is that you can construct a metric over the words in the file so that you can say "this word is next to that one" in a relatively efficient fashion. –Donal Fellows 08:37, 22 April 2010 (UTC)
- Re “production code / fts module”, I just wanted to take away the mystery of search a little. Will try to do union-find soon. --Tinku99 19:27, 21 April 2010 (UTC)
- Good. Could do with more of these sorts of tasks. –Donal Fellows 08:37, 22 April 2010 (UTC)