| Presentation | 2003/8/22 Multi Document Summary Analysis with Faceted Genre Yohei SEKI, Koji EGUCHI, Noriko KANDO, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | We assume that summaries vary according to the reader's information needs/intentions for information retrieval. We set several summary types like fact-based summary, prospective summary, numeric-data focused summary, or advice (opinion-oriented) summary etc. In our research, we define the viewpoints by combination of topics and summary types. The purpose of this study is to build a multi-document summarizer to produce summaries according to viewpoints. As an exploratory stage of investigation, we examined effectiveness of source documents genre to produce different types of summaries. We estimated R^2 (coefficeint of determination) improvement rates for sentence extraction with multiple linear regression analysis. The criterion variables is the sentence weights in source documents. The weights are computed based on the similarity of the human-produced summaries and sentences in the source documents. The explatory variables are sentence weitghting parameters. If we add 4 genre features into the explatory variables, from 4 to 12 summaries have a significant improvement for R^2 by each. We revealed that the augumentative feature (the third genre feature) relates to the commentary summary type. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Multi-Document Summarization / Viewpoint Summary / Genre Classification / Faceted Classification / Multiple Linear Regression Analysis / Opinion Summarization |
| Paper # | NLC2003-23 |
| Date of Issue |
| Conference Information | |
| Committee | NLC |
|---|---|
| Conference Date | 2003/8/22(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Natural Language Understanding and Models of Communication (NLC) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Multi Document Summary Analysis with Faceted Genre |
| Sub Title (in English) | |
| Keyword(1) | Multi-Document Summarization |
| Keyword(2) | Viewpoint Summary |
| Keyword(3) | Genre Classification |
| Keyword(4) | Faceted Classification |
| Keyword(5) | Multiple Linear Regression Analysis |
| Keyword(6) | Opinion Summarization |
| 1st Author's Name | Yohei SEKI |
| 1st Author's Affiliation | Department of Informatics, The Graduate University for Advanced Studies() |
| 2nd Author's Name | Koji EGUCHI |
| 2nd Author's Affiliation | Department of Informatics, The Graduate University for Advanced Studies:National Institute of Informatics |
| 3rd Author's Name | Noriko KANDO |
| 3rd Author's Affiliation | Department of Informatics, The Graduate University for Advanced Studies:National Institute of Informatics |
| Date | 2003/8/22 |
| Paper # | NLC2003-23 |
| Volume (vol) | vol.103 |
| Number (no) | 280 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |