Formal Concept Analysis for Domain-Specific Document Retrieval Systems

摘要

Domain-specific information retrieval normally depends on general search engines, or systems which support browsing using handcrafted organisation of documents, but such systems are costly to build and maintain. An alternative approach for specialised domains is to build a retrieval system incrementally and dynamically by allowing users to evolve their own organisation of documents and to assist them in ensuring improvement of the system's performance as it evolves. This paper describes a browsing mechanism for such a system based on the concept lattice of Formal Concept Analysis (FCA) in cooperation with incremental knowledge acquisition mechanisms. Our experience with a prototype suggests that a browsing scheme for a specific domain can be able to be collaboratively created and maintained by multiple users over time. It also appears that the concept lattice of FCA is a useful way of supporting the flexible open management of documents required by individuals, small communities or in specialised domains.