Discourse and citation analysis with concept-matching
Agnes Sandor, Aaron Kaplan, Gilbert Rondeau
We present here two natural language processing systems for highlighting passages in scientific texts in order to help researchers to rapidly access relevant knowledge. The first system detects sentences containing expressions fulfilling discourse functions in scientific argumentation like background knowledge, summary sentence, contrast with past findings, etc. The second system detects sentences containing bibliographical references and characterizes the relationship that the authors describe between their work and the work they refer to. The systems are implemented in the Xerox Incremental Parser
International Symposium : Discourse and document (ISDD), Caen, France, June 15-16.
result.pdf (32.26 kB)