Scalable feature selection, classification and signature generation for organizing large text databases into hierarchical topic taxonomies

Chakrabarti, Soumen ; Dom, Byron ; Agrawal, Rakesh ; Raghavan, Prabhakar (1998) Scalable feature selection, classification and signature generation for organizing large text databases into hierarchical topic taxonomies The VLDB Journal The International Journal on Very Large Data Bases, 7 (3). pp. 163-178. ISSN 1066-8888

Full text not available from this repository.

Official URL: http://doi.org/10.1007/s007780050061

Related URL: http://dx.doi.org/10.1007/s007780050061

Abstract

We explore how to organize large text databases hierarchically by topic to aid better searching, browsing and filtering. Many corpora, such as internet directories, digital libraries, and patent databases are manually organized into topic hierarchies, also called taxonomies. Similar to indices for relational data, taxonomies make search and access more efficient. However, the exponential growth in the volume of on-line textual information makes it nearly impossible to maintain such taxonomic organization for large, fast-changing corpora by hand. We describe an automatic system that starts with a small sample of the corpus in which topics have been assigned by hand, and then updates the database with new documents as the corpus grows, assigning topics to these new documents with high speed and accuracy. To do this, we use techniques from statistical pattern recognition to efficiently separate the feature words, or discriminants, from thenoise words at each node of the taxonomy. Using these, we build a multilevel classifier. At each node, this classifier can ignore the large number of “noise” words in a document. Thus, the classifier has a small model size and is very fast. Owing to the use of context-sensitive features, the classifier is very accurate. As a by-product, we can compute for each document a set of terms that occur significantly more often in it than in the classes to which it belongs. We describe the design and implementation of our system, stressing how to exploit standard, efficient relational operations like sorts and joins. We report on experiences with the Reuters newswire benchmark, the US patent database, and web document samples from Yahoo!. We discuss applications where our system can improve searching and filtering capabilities.

Item Type:Article
Source:Copyright of this article belongs to Springer Nature Switzerland AG
Keywords:Feature Selection;Digital Library;Statistical Pattern;Document Sample;Feature Word
ID Code:130996
Deposited On:02 Dec 2022 05:40
Last Modified:02 Dec 2022 05:40

Repository Staff Only: item control page