Presentation on theme: "DAML Tools for Intelligent Information Annotation, Sharing and Retrieval UMBC Johns Hopkins University Applied Physics Lab MIT Sloan School July 19, 2001."— Presentation transcript:
DAML Tools for Intelligent Information Annotation, Sharing and Retrieval UMBC Johns Hopkins University Applied Physics Lab MIT Sloan School July 19, 2001
UMBC/JHU/MIT Team UMBC, JHU/APL, and MIT/Sloan are working together on a set of issues UMBC Integrating communicating agents, DAML and web applications Tim Finin, Charles Nicholas, Yun Peng, Anupam Joshi, Scott Cost JHU APL DAML and information retrieval Jim Mayfield, Paul McNamee, Wayne Bethea MIT Sloan School DAML, rules based technology and distributed belief Benjamin Grosof To be integrated in agent-based applications involving search and using rule-based reasoning. DEMO DEMO
UMBC WHAT'S HOT? ITTALKS as a useful DAML-based application –Automatic classification of talks and users wrt DAML topic ontologies; multiple topic ontologies, with manual and mapping between them. –Agents get DAML talk notifications and make entries on a users calendar if it matches interests, location and schedule. –Agents use DAML as ACL and incorporate a DAML reasoning engine (XSB, YAJXB, RDF API) –A DAML distributed trust model including permissions, obligations, authorization and delegation. DAMLized Jini registration service DAMLized Bluetooth service discovery protocol CHALLENGES! Need for a range of DAML reasoning engines –Offering varying degrees of services and completeness –Implementations in Java? Full DAML may be too large for some interesting applications –e.g., using in Jini for light-weight devices –e.g., use in Bluetooth service discovery protocol Lack of DAML rules –Required for distributed trust policies.
JHU WHAT'S HOT? Search on DAML, text, or both Normalized DAML tags are used as indexing terms. Queries may include DAML, text, or both. Automated relevance feedback adds related words and DAML tags to the user's query. System finds DAML tags or words most closely related to a given word, phrase, or DAML tag. Ontology mapping Given a node in an ontology, system finds other DAML tags and words that characterize it, then finds other ontology nodes that have similar characterizations. Provides a simple ontology mapping capability. CHALLENGES! DAML quantity There is still not very much DAML on the Web DAML quality Monolithic files: many DAML statements are concentrated in a few large files Structured database nature: many DAML tags always co-occur with exactly the same set of tags, making it difficult to distinguish the tags statistically Is DAML markup?: much DAML has very few accompanying words
MIT Sloan WHAT'S HOT? Webizing and extending logic programs Web-powerful: URI's for: relations, functions, rulesets, reasoning Web-friendly syntax: exploit/support both XML and RDF, use namespaces Ontologies: basic connection: link ontologies for rules, and vice versa Object-oriented: basic features: subclass... with exceptions; member roles Infrastructure for inter-site communication of premises and conclusions (+justifications) –inferencing as a web service, with... –translation among rule languages –merging web-named/accessible rulesets with prioritized conflict handling –hybrid reasoning algorithms & semantics Applications: describe contracts and web services as rulesets and decide to buy/choose or sell via inferencing about the deal CHALLENGES! Heterogeneity of commercial rule languages, yet need deep common semantics –potential de facto standards jockeying by vendors Multiplicity of protocols for procedural invocation: –CGI, SOAP, RMI, CORBA,... Unordered (RDF) vs. ordered (XML) graph semantics for Webized syntax Little theory available for tight composition of Rules plus Description Logic -- yet want deep combined knowledge representation semantics Current gulf to databases/XMLQuery; sociologically + technically unconnected
Papers 1.X. Luan, Y. Peng, and T. Finin, Agent Consumer Reports: of the Agents, by the Agents, and for the Agents, 2nd Asia-Pacific Conference on Intelligent Agent Technology, Maebashi City, Japan, Oct, 2001. 2.T. Finin, A. Joshi, L. Kagal, O. Ratsimore, V. Korolev, and H. Chen, Information Agents for Mobile and Embedded Devices, 5th Int. Workshop on Cooperative Information Agents, Sept. 6-8, 2001 Modena, Italy. 3.L. Kagal, T. Finin, and Y. Peng, A Framework for Distributed Trust, Workshop on Autonomy, Delegation, and Control: Interacting with Autonomous Agents, IJCAI-2001, Seattle, Aug. 2001. 4.R. S. Cost, T. Finin, A. Joshi, Y. Peng, F. Perich, C. Nicholas, H. Chen, L. Kagal, Y. Zou, and S. Tolia, ITTALKS: A Case Student in how the Semantic Web Helps, Semantic Web Workshop, July 2001, Stanford. 5.F. Perich, R.S. Cost, T. Finin, A. Joshi, Y. Peng, C. Nicholas, H. Chen, L. Kagal, Y. Zou, and S. Tolia, ITTALKS: An Application of Agents in the Semantic Web, Workshop on Engineering Societies in the Agents' World, 7 July 2001, Prague. 6.S. Avancha, A. Joshi and T. Finin, Enhancing Bluetooth Service Discovery Protocol, Report TR- CS-01-08, CSEE, UMBC, June 2001. 7.L. Kagal, Scott Cost, H. Chen, T. Finin, Yun Peng, An Infrastructure for Distributed Trust Management, Workshop on Norms and Institutions in Multiagent Systems, Autonomous Agents 2001, Montreal, May 2001.