SCALABLE FEATURE-SELECTION, CLASSIFICATION AND SIGNATURE GENERATION FOR ORGANIZING LARGE TEXT DATABASES INTO HIERARCHICAL TOPIC TAXONOMIES

We explore how to organize large text databases hierarchically by topi c to aid better searching, browsing and filtering. Many corpora, such as internet directories, digital libraries, and patent databases are m anually organized into topic hierarchies, also called taxonomies. Simi lar to indices for relational data, taxonomies make search and access more efficient. However, the exponential growth in the volume of on-li ne textual information makes it nearly impossible to maintain such tax onomic organization for large, fast-changing corpora by hand. We descr ibe an automatic system that starts with a small sample of the corpus in which topics have been assigned by hand, and then updates the datab ase with new documents as the corpus grows, assigning topics to these new documents with high speed and accuracy. To do this, we use techniq ues from statistical pattern recognition to efficiently separate the f eature words, or discriminants, from the noise words at each node of t he taxonomy. Using these, we build a multilevel classifier. At each no de, this classifier can ignore the large number of ''noise'' words in a document. Thus. the classifier has a small model size and is very fa st. 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 classe s to which it belongs. We describe the design and implementation of ou r system, stressing how to exploit standard, efficient relational oper ations 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 sear ching and filtering capabilities.