Computer Science
8 years ago

New method extracts topic transitions from news articles with high accuracy.

Extraction of topic transition from document stream based on hierarchical clustering
Masaaki Kikuchi, Masayuki Okamoto, Tomohiro Yamasaki

Paper Summary

The article presents a method to extract keywords showing topic changes in a stream of documents like news articles. The method uses hierarchical clustering and C-value method to create compound words. In a user evaluation of 640 topics over 32 days, it was found that 94.3% of topics were understandable as news, and 68.6% of topics showed topic transitions.

Related papers

Theme background image
Computer Science
8 years ago

Revolutionize Information Organization with Hierarchical Document Clustering!

Data Mining Techniques
Multi-label Classification
Text Compression
1 view
Theme background image
Physics and Astronomy
4 years ago

New document clustering method improves efficiency and quality of information retrieval.

Statistical Mechanics
Data Clustering
Trajectory Analysis
1 view
Theme background image
Computer Science
5 years ago

Unveiling the Future: Text Clustering Revolutionizes Topic Identification in Data Retrieval

Keyword Extraction
Multi-label Classification
1 view