Abstract | ||
---|---|---|
Single pass clustering algorithm is widely used in topic detection and tracking. It is a key part of network topic detection model. In the process of single pass algorithm, clustering results are not satisfactory, and the similarity matching would be reduced. Focusing on these two defects, this paper physically reconstructs web information into a volume, in which every document contains "theme area" and "details area". To improve single pass clustering algorithm, this paper uses "theme area" to detect topics and apply the whole document to distinguish subtopics, while central vector model is used to denote topics. Experimental results indicate that the model based on text reconstruction performs well in detecting network topics and distinguishing subtopics. |
Year | DOI | Venue |
---|---|---|
2013 | null | INFORMATICA-JOURNAL OF COMPUTING AND INFORMATICS |
Keywords | Field | DocType |
topic detection and tracking, single pass algorithm, text reconstruction, network topic detection | Single pass,Pattern recognition,Computer science,Artificial intelligence,Cluster analysis,Similarity matching,Web information | Journal |
Volume | Issue | ISSN |
37 | 4 | 0350-5596 |
Citations | PageRank | References |
0 | 0.34 | 1 |
Authors | ||
6 |
Name | Order | Citations | PageRank |
---|---|---|---|
Zhen-fang Zhu | 1 | 9 | 3.04 |
Peipei Wang | 2 | 6 | 4.81 |
zhiping jia | 3 | 463 | 60.64 |
Hairong Xiao | 4 | 0 | 1.35 |
Guangyuan Zhang | 5 | 11 | 4.34 |
Hao Liang | 6 | 0 | 0.34 |