Mingyang proposed his idea for the Search Agent - specifically, a semantic search which is returns more specific results as well as relatedness information between returned "JONS" (Jans + Ontology Description).
Dr. Reddy proposed the establishment of "rules" concerning various aspects of particular JAN types. As an example, we discussed a "meeting minutes" JAN which always has the attribute "number of participants." This would allow a "fill-in-the-blank" sort of interface. In particular, we might want to provide the user with a non-programming method of establishing these "types."
Xiaoyan went more in depth on her Ontology ideas, specifically the reasoning methods involved.
Mingchu discussed his ideas on how a JAN should be constructed, and how one might use "key phrase extraction" to obtain semantic information. In particular he spoke on the linking agent (e.g. "what should be linked"?)
Dr. Reddy went on to talk about two potential types of linking - "general" and "domain-centric". "General" linking is ad-hoc and might be described as "is-a," "part-of," "member-of," etc. "Domain-centric" is more concerned with what kind of JAN we are dealing with. For example, we might have a JAN_Meeting, which has a Meeting_minutes which is related to it. This could be described as "Meeting_minutes_of_JAN_Meeting," like an attribute. So, we are attempting to define what is the role that is being played. He also mentioned that an important difference between our research and others is that our ontology is organic - i.e. it grows and changes over time, rather than having been assumed to have simply always been there.
Daniel discussed the Work-Pattern-Centric Knowledge Advantage Machine.