12.07.2015 Views

Information Retrieval - ad-teaching.infor...

Information Retrieval - ad-teaching.infor...

Information Retrieval - ad-teaching.infor...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Latent Semantic Indexing 8/9• With the SVD, rank-k approximation becomes easy– For a given m x n matrix A, compute SVD A = U · ∑ · V T– Let U k = the first k columns of U– Let Σ k = the upper k x k part of ∑– Let V k = the first k columns of V– Then AA = U k · Σ k · V T k is the desired approximationthat is, that rank-k matrix A k which minimizes ǁ A – A k ǁ– Let's look at our example in Octave ...17

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!