Intelligent Web Agents that Learn to Retrieve and Extract Information

We describe systems that use machine learning methods to retrieve and/or extract textual information from the Web. In particular, we present our Wisconsin Adaptive Web Assistant (WAWA), which constructs a Web agent by accepting user preferences in form of instructions and adapting the agent’s behavior as it encounters new information. Our approach enables WAWA to rapidly build instructable and self-adaptive Web agents for both the information retrieval (IR) and information extraction (IE) tasks. WAWA uses two neural networks, which provide adaptive capabilities for its agents. User-provided instructions are compiled into these neural networks and are modified via training examples. Users can create these training examples by rating pages that WAWA retrieves, but more importantly our system uses techniques from reinforcement learning to internally create its own examples. Users can also provide additional instruction throughout the life of an agent. Empirical results on several domains show the advantages of our approach.