Electronic Thesis and Dissertation Repository

Degree

Master of Science

Program

Computer Science

Supervisor

Mercer, Robert E.

Abstract

The traditional reference evaluation method treats all citations equally. However, a citation can serve various functions. It may reflect the citing paper author’s motivation as well as his/her true attitude towards the cited paper. Investigating such information can be achieved through citation content analysis.

This thesis develops an 8-category classification scheme on citation function and polarity to help understand what role a citation played in scientific papers. A biomedical citation corpus is annotated with this scheme and experimented with supervised machine learning methods. Several types of features that capture the characteristics of citation sentences are extracted by natural language processing techniques to serve as the inputs of automatic classifiers. The importance of cue phrases in citation classification is also addressed and discussed.

Share

COinS