Information Extraction and Database Techniques: A User-Oriented Approach to Querying the Web

摘要: We propose a novel approach to querying the Web with system named AKIRA (Agentive Knowledge-based Information Retrieval Architecture) which combines advanced technologies from and Extraction together Database techniques. The former enable access explicit as well implicit structure of documents organize them into hierarchy concepts metaconcepts; latter provide tools for data-manipulation. useroriented approach: given user's query, extracts target (structure expressed in query) uses standard retrieval techniques potentially relevant documents. content these is processed using extraction (along flexible agentive structure) filter relevance extract or matching structure. information garnered used populate smart-cache (an object-oriented database) whose schema inferred This smart-cache, thus defined posteriori, populated queried an expression PIQL, our query language. integrates complementary maximum flexibility user offer transparent