Evaluating Summaries for Multiple Documents in an Interactive Environment

Abstract

While most people have a clear idea of what a single document summary should look like, this is not immediately obvious for a multi-document summary. There are many new questions to answer concerning the amount of documents to be summarized, the type of documents, the kind of summary that should be generated, the way the summary gets presented to the user, etc. The many approaches possible to multi-document summarization makes evaluation especially difficult. In this paper we will describe an approach to multi-document summarization and report work on an evaluation method for this particular system.

Resources

Details