Take a look at the new version of this page: [ beta version ]. Tell us what you think.

ABSTRACT

This paper presents a novel approach to classifying library records by making use of what we call "author profile," a representation of an author's expertise along a library classification. Coupled with a string kernel classifier, the idea is shown to bring a significant improvement over a baseline.

REFERENCES

• 1: Eibe Frank , Gordon W. Paynter, Predicting library of congress classifications from library of congress subject headings, Journal of the American Society for Information Science and Technology, v.55 n.3, p.214-227, February 2004 [doi>10.1002/asi.10360]
• 2: E. Ishida. Text categorization using title and a table of contents. IPSJ SIG Notes, 2006. In Japanese.
• 3: A. Karatzoglou and A. Smola. kernlab -- An S4 Package for Kernel Methods in R. Journal of Statistical Software, 2004.

INDEX TERMS

Primary Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL
      H.3.6 Library Automation

General Terms:
Algorithms

Keywords:
SVM, author profiling, library classification, string kernel