TOAS intelligence mining; Analysis of natural language processing and computational linguistics

Abstract

The Technology Opportunities Analysis System (TOAS), being developed under a Defense Advanced Research Projects Agency (DARPA) project, enables mining of text files using bibliometrics. TOAS, a software system, extracts useful information from literature abstract files, which have identified fields that repeat in each abstract record of specific databases, such as Engineering Index (ENGI), INSPEC, Business Index, U.S. Patents, and the National Technical Information Service (NTIS) Research Reports. The TOAS applies various technologies, which include natural language processing (NLP), computational linguistics (CL), fuzzy analysis, latent semantic indexing, and principle components analysis (PCA). This software system combines simple operations (i.e., listing, counting, list comparisons and sorting of search term retrieved consolidated records' field results) with complex matrix manipulations, statistical inference and artificial intelligence approaches to reveal patterns and provide insights from large amounts of information, primarily related to technology-oriented management issues. The authors apply the TOAS tool on its own root technologies, NLP and computational linguistics—two apparently synonymous terms. These terms, however, when used in a literature search of the same abstract databases, ENGI and INSPEC, provide distinctly different search results with only 10% to 25% search result abstract records overlap. This paper introduces TOAS, summarizes analyses comparing NLP and CL, and then discusses the underlying development implications.