WebKB-2 is a knowledge server
that permits Web users to browse and update private knowledge bases on
their machines, or alternatively, a large shared knowledge base on the server machine.
In this last case, each object - a category (type or individual), a link between
categories (e.g. subtypeOf, exclusion, locationOf, substanceOf),
or a more complex statement (graph) - has an associated
source (user and/or document), and a Web user can
1) query, re-use and add new categories, links or graphs,
2) remove those he/she has created,
3) "correct other graphs" (represent alternative beliefs) by using
relations such as pm#corrective_specialization.
The ontology of the shared knowledge base is currently an integration of various top-level ontologies and a lexical ontology derived from an extension and correction of the noun-related part of WordNet 1.7 (for a published article, click here). The semantics of some categories from WordNet had to be modified in order to fix inconsistencies (all the modifications have been recorded and justified). The semantics of categories from other sources (e.g. Sowa, Dolce, the Lifecycle Integration Schema, the Natural Semantic Metalanguage, OWL, DAML+OIL, KIF and the Dublin Core) have been kept (no direct/indirect link or definition has been modified, and hopefully the links between categories of different sources are mostly correct). Since the categories from the various ontologies are well connected and since the creators of the categories, links and statements are recorded, 1) the categories and statements from the various ontologies complement, illustrate or precise each other, 2) lexical conflicts are avoided, 3) semantics conflicts or redundancies can be detected when new categories or statements are added, thus leading the knowledge providers to be more precise and keep the knowledge base well inter-connected (as opposed to the traditional approach where ontologies are more-or-less independently developed and must then be selected, interpreted and merged by each knowledge engineer wanting to re-use ontologies for an application), 4) the categories and statements may be filtered on their creators (according to their identifiers, types or more complex information on them), 5) the contribution of each knowledge provider is acknowledged.
The Standard Upper Ontology Working Group
has voted this
Multi-Source Ontology (MSO) as one of the materials to work on
(click here for the main messages on the SUO mailing list
about the MSO of WebKB-2 and its underlying approach, the most important
messages being
the one of December 2nd 2003
and the one of May 12th 2004).
WebKB-2 and its MSO may indeed be an instrument for ontology creators
to interconnect their ontologies with other ones, and let people query these
ontologies and re-use or complement them. Importantly, no central authority is
relied on, and conflicts can be solved without discussion nor compromise (but
also, the protocols are not sufficient to "force" the knowledge providers to
keep the shared ontology "well connected", some minimal cooperative spirit and
fairness is necessary).