SIMS Group Home Page

The SIMS Group of Projects at ISI

The SIMS Group consists of several related research projects in ISI's Intelligent Systems Division that are investigating different aspects of the problem of retrieving and integrating data distributed over multiple heterogeneous information sources. The group began by working with well-structured databases, and has expanded over the years to deal also with more loosly structured text sources and Web pages.

The work on accessing information over the Web has developed into the Ariadne project.

SIMS Overview

SIMS is an intermediate layer - a mediator - between information sources and humans users or applications programs. Queries to SIMS are in a uniform language, independent of the distribution of information over sources, of the various query languages, the location of sources, etc. SIMS determines which data sources to use, how to obtain the desired information, how and where to temporarily store and manipulate data, and how to maintain an acceptable level of efficiency in performing its task.

In contrast to the standard approach to this problem - building custom systems - SIMS applies and extends a variety of AI techniques and systems to build an intelligent and dynamic interface to diverse information sources.

A model is constructed to describe the domain about which information is stored in the information sources, as well the structure and contents of the sources themselves. The domain model is a declarative description of the objects and activities possible in the application domain as seen by a typical user. For each information source the model indicates the data-model used, query language, network location, size estimates, etc., and describes the contents of its fields in relation to the domain model. The user formulates queries using terms from the domain, without needing to know about specific information sources. SIMS' models of different information sources are independent, greatly easing the process of extending the system. An example of modelling is given here.

SIMS uses a planner to generate a sequence of queries to individual information sources that satisfy a user's query. The planner first selects information sources to be used in answering a query. It then orders sub-queries to the appropriate information sources, selects the location for processing intermediate data, and determines which sub-queries can be executed in parallel.

Change to information sources is handled by changing models only. The changes will be considered by the planner in producing future plans that utilize information from the modified sources. This greatly facilitates extensibility.

Diagram of the SIMS Architecture.

SIMS Information

Go to the ISI homepage.


visitors since July 8, 1996

José Luis Ambite (ambite@isi.edu).