MITLIBRARIES DUPL 111 3 9080 01918....3Hil5l9III6 Basement D28 M414 D£WEY ..MIS*- To appear in the Journal at Applied Intelligence, September 2000 Context Knowledge Representation and Reasoning in the Context Interchange System Stephane Bressan', Cheng Goh2 Natalia Levina, Stuart Madnick, Ahmed , Shah, Michael Siegel Sloan WP# 4133 CISLWP #00-04 August, 2000 MIT Sloan School ofManagement 50 Memorial Drive Cambridge, Massachusetts 02142-1347 MASSACHUSETTS INSTITUTE OFTECHNOLOGY E OCT 2 5 2000 LIBRARIES Context Knowledge Representation and Reasoning in the Context Interchange System Stephane Bressan1, Cheng Goh2, Natalia Levina, Stuart Madnick, Ahmed Shah, Michael Siegel Massachusetts Institute ofTechnology, Cambridge, MA 02139 Abstract The Context Interchange Project presents a unique approach to the problem ofsemantic conflict resolution among multiple heterogeneous data sources. The system presents a semantically meaningful view of the data to the receivers (e.g. user applications) for all the available data sources. The semantic conflicts are automatically detected and reconciled by a Context Mediator using the context knowledge associated with both the data sources and the data receivers. The results are collated and presented in the receiver context. The cunent implementation of the system provides access to flat files, classical relational databases, on-line databases, and web services. An example application, using actual financial information sources, is described along with a detailed description ofthe operation ofthe system foran example query. Keywords: context, integration, mediation, semantic heterogeneity, web wrapping 1. Introduction In recent years the amount ofinformation available has grown exponentially. While the availability ofso much information has helped people become self-sufficient and get access to all the information handily, this has created anotherdilemma. All these data sources and the technologies that are employed by the data source providers do not provide sufficient logical connectivity (the ability to exchange data meaningfully). Logical connectivity is crucial because users ofthese sources expect each system to understand requests stated in their own terms, using theirown concepts ofhow the world is defined and structured. As a result, any data integration effort must be capableof reconciling semantic conflicts among sources and receivers. This problem is generally referred to asthe need for semantic interoperabilityamongdistributed data sources. The Context Interchange Project at MIT [1,2] is studying the semantic integration ofdisparate information sources. Like other information integration projects (the SIMS project at ISI [3], the TSIMMIS project at Stanford [4], the DISCO project at Bull-INRIA [5], the Information Manifold projectat AT&T [6], the Garlic project at IBM [7], the Infomasterproject at Stanford [8]), we have adopted a Mediation architecture as outlined in Wiederhold's seminal paper [9]. In section 2, we present a motivational scenario ofa usertryingto access information from various actual data sources and the problems faced. Section 3 describes the current implementation ofthe Context mediation system. Section 4 presents a detailed discussion ofthe various subsystems, highlightingthe context knowledge representation and reasoning, usingthe scenario outlined in section 2. Section 5 concludes ourdiscussion. 2. Why Context Mediation ? - An Example Scenario Consider an example ofa financial analyst doing research on Daimler Benz. She needs to find out the net income, net sales, and total assets ofDaimler BenzCorporation forthe yearending 1993. In addition to that, she needs to know the closing stock price ofDaimler Benz. She normally uses the financial data stored in the Worldscope3 database. She recalls Jill, herco-workertelling her about two otherdatabases, Datastrearn4 and Disclosure5 and how they contained much ofthe infonnation that Jill needed. She starts offwith Worldscope database. She knows that Worldscope has total assets for all the companies. She brings up a query tool and issues aquery: Now at the National University ofSingapore. 2 We dedicate this work to the memory ofCheng Hian Goh (1965-1999). 3 The Worldscopedatabase is an extract from the Worldscope financial data source 4 The Datastrearn database is an extract from the Datastrearn financial data source. select company_name total_assets from worldscope , where company_name = " DAIMLER-BENZ AG" ; She immediately gets back the result: DAIMLER-BENZAG 5659478 Satisfied, she moves on and figures out after looking at the data information forthe new databases that she can get the data on net income from Disclosureand net sales from Datastream. Fornet income, she issues the query: select company_name net_income from disclosure , where company_name = " DAIMLER-BENZ AG" ; The query does not return any records. Puzzled, she checks fortypos andtries again. She knows thatthe information exists. Shetries one more time, this time entering apartial name for DAIMLER BENZ. select company_name net_income from disclosure , where company_name like " DAIMLER%" ; She gets the record back: DAIMLER BENZCORP 615000000 She now realizes that the data sources do not conform to the same standards, as it becomes obvious from the names. Cautious, she presses on and issues the third query: select name, total_sales from datastream where name like " DAIMLER%" ; She gets the result: DAIMLER-BENZ 9773092 Company-Name Woikkcupc