Domain-dependent information gathering agent

摘要

A universal agent should be capable of gathering information from arbitrary heterogeneous sites and offer intelligent information services on its own based on information so gathered. We present a domain-dependent agent for information gathering. It can visit an arbitrary domain-related site by observing a user perform the first query. By understanding key concepts of the first query, the agent performs subsequent queries autonomously. When a user asks the agent about a particular item, the agent gathers relevant information from various sites. The major advantage of the agent is a semi-automatic creation of a wrapper around a particular site with few human interventions. We have implemented two versions of such information gathering agents: ShinA (SHoppINg Assistant) for e-trading tasks, and EMA (EMployment Agent), which performs employment and job related functions over the Internet.