Burst Events Detection in Text Streams by Using Keyphrases
碩士 === 國立中正大學 === 資訊管理學系暨研究所 === 99 === Mining text streams for hot topics and events has attracted extensive attention in the world because of its broad applications. Since keyphrases have more expressive power than single term and keyphrases can be utilized to represent documents more semantically...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2011
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/44508835581874558353 |
| Summary: | 碩士 === 國立中正大學 === 資訊管理學系暨研究所 === 99 === Mining text streams for hot topics and events has attracted extensive attention in the world because of its broad applications. Since keyphrases have more expressive power than single term and keyphrases can be utilized to represent documents more semantically. In this research, we try to detecting burst events by using keyphrases.
We give a formal definition to the above problem and present the frameworks with five steps to solve the problem: (1) use KP-Miner to extract keyphrases from text streams as features set; (2) cluster keyphrases with synonymy or hypernymy into groups; (3) calculate occurrence frequencies of the groups in sliding windows; (4) evaluate burst groups; (5) burst event as burst groups. We also find the problem about loosing potential burst groups in fixed time window. In order to alleviate this problem, the original time window and the shift time window are good ways to settle the problem. We evaluate the proposed framework on real Google news stream which is suitable for our research. Experimental results show that our framework can detect more descriptive burst events than external events.
|
|---|