Keyword-Based, Context-Aware Selection of Natural Language Query Patterns

Pervasive access to distributed data sources by means of mobile devices is becoming a frequent realistic operational context in many application domains. In these scenarios data access may be thwarted by the scarce knowledge that users have of the application and of the underlying data schemas and complicated by limited query interfaces, due to the small size of the devices.A viable solution to this problem could be expressing the queries in natural language; however, in applications like medical emergencies, data management systems must obey requirements such as very fast and precise data access which make this solution infeasible.To reduce the time needed to get answers to user queries, the paper proposes a lightweight, context-aware approach based on the combination of keywords with natural language queries. The method employs ontologies and query patterns to support the users in formulating the most appropriate query for retrieving the desired data. Precision and query efficiency are further improved by focusing searches only to the data which are meaningful w.r.t. the current context, thus supporting the users' situation awareness.The approach has been integrated in the SAFE system, developed for mobile and Web, and has been applied in cardiology to support medical personnel in emergency interventions on patients affected by chronic cerebro-vascular diseases. Experimental results have shown that the proposed solution significantly reduces the time to get useful data w.r.t. traditional form-based approaches.