AutomaticSummarization

Automatic Summarization

Automatically create a summary extracting the most significant sentences in a text.

Summaries may be produced from a single document or multiple documents.

A very good survey on this theme is the work done at the Carnegie Mellon University.

Single Document Summary

The following techniques (both focused on technical documents) are currently implemented:

• H. P. Luhn (April 1958). The Automatic Creation of Literature Abstracts. IBM Journal: 159â€"165.</li>
• Oliveira, P.C.F., Ahmad, K., & Gillam, L. (2002), A financial news summarization system based on lexical cohesion, Proc. TKE2002, Nancy, France.

Multiple Document Summary

• Lex Rank: http://citeseerx.ist.psu.edu/viewdoc/download;jsessionid=18FD35E27F54A0B76359DA12E389BADC?doi=10.1.1.76.9855&rep=rep1&type=pdf
• MEAD: http://clair.si.umich.edu/~radev/papers/centroid.pdf