Semantic representation analysis: A general framework for individualized, domain-specific and context-sensitive semantic processing

Abstract

Language agnostic methods for semantic extraction, encoding, and applications are an increasingly active research area in computational linguistics. This paper introduces an analytic framework for vector-based semantic representation called semantic representation analysis (SRA). The rationale for this framework is considered, as well as some successes and future challenges that must be addressed. A cloud-based implementation of SRA as a domain-specific semantic processing portal has been developed. Applications of SRA in three different areas are discussed: analysis of online text streams, analysis of the impression formation over time, and a virtual learning environment called V-CAEST that is enhanced by a conversation-based intelligent tutoring system. These use-cases show the flexibility of this approach across domains, applications, and languages. © 2014 Springer International Publishing Switzerland.