A Rule-Based Semiautomated Approach to Building Natural Language Question Answering (NLQA) Systems

This paper presents a rule-based approach to natural language question answering that can be easily implemented for any domain. We discuss the framework in the context of a National Science Foundation funded project Universal Virtual Laboratory (UVL). UVL is a virtual electrical engineering (EE) laboratory for able and disabled individuals to construct, simulate and understand the characteristics of basic electrical circuits. Like any real-life laboratory, the UVL has a teaching assistant (TA), to whom students can ask questions related to basic EE and on using the laboratory in general. The virtual TA is built using the framework that is discussed in the paper. The framework relies on careful design rather than complex grammar or statistics. It can be used for building rapid language comprehension applications specific to a particular domain by using the appropriate heuristics/keywords of that domain. Also introduced is the automation tool that can be used to implement the framework for any domain.