2004-11-15

Finding scientific topics

Other keywords: webmining, knowledge extraction

From the PNAS Mapping Knowledge Domains, we find the link under the title of this post. The topic has to do with various means, including probabilistic, by which scientific topics can be mined from a body of literature. I think this idea applies to those notions whereby subject identity is based on various properties, some of which are detected by datamining techniques. Requisite quote:
A first step in identifying the content of a document is determining which topics that document addresses. We describe a generative model for documents, introduced by Blei, Ng, and Jordan [Blei, D. M., Ng, A. Y. & Jordan, M. I. (2003) J. Machine Learn. Res. 3, 993-1022], in which each document is generated by choosing a distribution over topics and then choosing each word in the document from a topic selected according to this distribution. We then present a Markov chain Monte Carlo algorithm for inference in this model. We use this algorithm to analyze abstracts from PNAS by using Bayesian model selection to establish the number of topics. We show that the extracted topics capture meaningful structure in the data, consistent with the class designations provided by the authors of the articles, and outline further applications of this analysis, including identifying "hot topics" by examining temporal dynamics and tagging abstracts to illustrate semantic content.