Integrating Knowledge Base Retrieval with Web Search using Semantic Roles

An integrated search of the Web together with a local knowledge base is essential for specialized search applications in various domains for which rich local knowledge bases are available. Yet, there is no satisfactory solution to the problem either in research prototypes or commercial products that meets the requirements of high precision and effectiveness achievable only through the application of the semantic richness of the domain (as opposed to a general-purpose keyword-based search). We find three major problems in current solutions for such a domain-specific integrated search: (i) absence of links between conceptual elements of the knowledge base (i.e., the ontology) and textual elements in documents, (ii) drawbacks of keyword-based Web search engines such as lack of semantics and poor precision, (iii) the inadequacy of hyperlinks in terms of capturing the context and semantics of the links in addition to the lack of guidance for the user while navigating the Web together with a local knowledge repository. We present a solution to the above problems which caters to the needs of a knowledge worker in retrieving relevant information from both a local knowledge base and the Web using a richly interconnected link structure called Superlinks with well-defined semantic roles . We also illustrate a prototype implementation in the example domain of Indian medicinal herbs in Ayurveda.