I'll begin with the Mofat research paper. The indea is to include an internal index in each inverted list, and also use special compressions. I've to create the inverted list with the class TrecIndexing and then I'll build an other inverted index, using special compression and accumulators.

For the compression, the author recommands two kinds of compression : Elias gamma and Golomb, b. For the moment, I'll use the gamma compression, because it's already used in terrier. Thus, I want to use the two classes BitOutputStream and BitInputStream.

I need to build an iterator on the posting list for skiplist.

The author uses the cosine measure for weighting. I also have to inplement this structure with the iterator on skiplists.

last edited 2008-04-10 08:43:14 by CraigMacdonald