Using Allegro CL, Associative Cognition, Inc., (ACI), developed two cutting edge applications as part of its family of information retrieval and presentation products. TopicHound, a document classifier, and HighlightHound, a document summarizer, are, according to Dr. Tim Musgrove, Vice President of Artificial Intelligence Development, tools for data filtering and presentation.
TopicHound is an automated, customizable document classifier that runs on the Windows 95/NT platform. It rapidly and intelligently analyzes documents, determines their topic, and sorts them into default or user-defined categories. Companies that own a server with high throughput will find TopicHound to be a useful alternative to manual classification.
TopicHound is easy to install, train, and retrain. Furthermore, it offers useful features such as multiple classification and adjustable prediction parameters. For example, a user can create the topics "Digital Cameras" and "Deep-Sea Exploration" and then classify a web site dealing with deep-sea digital photography under both topics. Users can also adjust TopicHound so that it makes a qualified prediction when it is unsure of a document's classification.
HighlightHound, which can be used effectively in conjunction with TopicHound, is an intelligent text highlighter and summarizer. Using complex algorithms, it selects only the most critical sentences in a document and can combine them into a handy summary.
TopicHound and HighlightHound were both developed with Franz Inc.'s Allegro CL, which supports ANSI Standard CLOS (Common Lisp Object System). Says Dr. Musgrove, "We used Lisp because it streamlines the development cycle. Lisp is clearly the most powerful, flexible, and portable language for artificial intelligence applications. It would have been incredibly slow to do all the algorithmic research required for these kinds of complex applications in any other language." ACI will release Beta versions of TopicHound and HighlightHound in January 1998.
For more information about Mindmaker, Inc., you may visit their web site at http://www.mindmaker.com.
|Copyright © 2020 Franz Inc., All Rights Reserved | Privacy Statement|