An Extensive Empirical Study of Feature Selection Metrics for Text Classification

Explore this paper's citation graph

Summary

An empirical comparison of twelve feature selection methods evaluated on a benchmark of 229 text classification problem instances, revealing that a new feature selection metric, called 'Bi-Normal Separation' (BNS), outperformed the others by a substantial margin in most situations and was the top single choice for all goals except precision.

Type
article
Published
2003-03-01
Cited by
3,005
References
15

Keywords

Feature selection, Computer science, Artificial intelligence, Selection (genetic algorithm), Feature (linguistics)

References

Cited by

Related papers