IDF and Vector Space Models

I’m back from SIDIM XXIV. It was a great conference in honor of Professor Oscar Moreno, from the Gauss Research Laboratory and NIC.PR (responsible for the .pr Internet domains.).

Dr. Moreno is a legend in the area of pure and applied mathematics. I have the privilege of meeting with him.

The conference plenary speakers were equally three legends:

Elwyn Berlekamp, University of California at Berkeley
Solomon W. Golomb, University of Southern California
Guang Gong, University of Waterloo

The event was a success, although some speakers read straight from their notes. As an interdisciplinary conference on pure and applied mathematics, all kind of topics were covered.

I got the chance to present research work on a new global weight algorithm we are testing called scaled inverse document frequency (SIDF), a variant of the well-known IDF scheme.

For those unfamiliar with IDF and its implementation with ranking algorithms, Dr. Deepak Khemani from the Artificial Intelligence & Database Research Group at Indian Institute of Technology Madras has published a very useful tutorial presentation on Vector Space Models.

The tutorial is based on our series of articles on the subject and provides a better understanding of the theory. We could not have done it better.