Research

Using AI to Guide Users through the Data Maze

Professor Leake

One of the great challenges when dealing with large-scale data is providing accurate ways to collect the data and associated metadata when the source is human response or gathered from on-line instrument streams. Intelligent interfaces that employ artificial intelligence methods such as Conversational Case-Based Reasoning (CCBR) can ease the burden on the ingest clients (including humans) by technology that engages the client in a guided conversation, with questions selected strategically, based on existing data and the developing description of the encounter. CCBR and data mining methods could exploit the extensive data to provide suggestions and comparison points for diagnoses and treatments, but to apply such techniques requires research to address the unprecedented scale of this data.

Case-based reasoning principles can also facilitate search in other contexts. For example, the combination of CBR with concept-map-based knowledge modeling tools can provide novel interfaces to retrieve on-point resources, helping users to navigate large resource libraries. However, how to provide the needed knowledge models is a challenging research problem.

Professor Leake is a leading expert on case-based reasoning and has extensive experience with applications involving remote user-technician dialogue to diagnose problems by using CBR to guide human reporting of problem symptoms and provide relevant prior examples, as well as for systems to support knowledge-model-based retrieval.