An approach to KNOW-WHO using RDF

Kunio Matsui

Know-Who (looking for experts with the specific skill, for instance) is one of the functions to realize Knowledge Management(KM). Among various KM functions, Know-Who is difficult to implement, because it requires managing tacit knowledge in people as well as explicit knowledge in DBs, documents, and so on. The central problem is how to collect and represent tacit knowledge. There are already some products with Know-Who function. In most products, however, personal skill information is manually declared as a set of attribute-value pair, and stored in RDBs. This method has the following two problems. 1. It requires high maintenance costs. As personal knowledge is changing day by day, the database must be updated continually. However, skilled people are always busy, and have no extra time for maintenance. 2. A more complex data structure is required to represent personal knowledge. Simple attribute-value pair structure is not enough to do that. My talk is about our approach, called ¡ÈWorkWare++¡É to solve above problems in Know-Who. Using NLP techniques, WorkWare++ attaches semantic links between objects in the groupware such as documents, people, meetings, and so on, semi-automatically. Those links are stored as RDF (Resource Description Framework) that is the standard metadata format in the Semantic Web. Our Know-Who function is implemented as a combination of full-text searching and information visualization of the RDF data.