Over the past eight years we have developed statistical methods for characterizing, classifying, and retrieving brief natural-language messages. Our goal was to provide a tool for people who had to deal with enormous numbers of heterogeneous documents, using ill-defined criteria of relevance and interest. Initially, we worked with a large, general-purpose system, the On-Line Pattern Analysis and Recogition System (OLPARS). More recently, we have developed a system called Message Extraction Through Estimation of Relevance (METER). Our present aim is the construction of a Testbed system for further studies and the development of more complex systems.
/lp/association-for-computing-machinery/statistical-techniques-for-free-text-processing-gNFonvYA5F