The InfoSleuth Agent System

[ Application Features | Underlying Technologies | Index ]

Significant progress has been made recently in the development of technologies for uniform access to heterogeneous information sources and services. Examples of this include HTML/XML for document access, and ODBC for access to relational database systems. However, mere access to information does not always help the user, who needs to be able to locate the right information at the right time, and in an efficient manner. For example, the emerging Internet meta-search engines locate information, but frequently flood the user with an excess of information, most of which is irrelevant to the intended query.

A person typically does not want to be flooded with all of this newly-accessible information, but rather is interested in acquiring the right information, at the right time and at the right level of abstraction, and in his own terms. Thus, there is a "next level" of problems that needs to be addressed, which includes:

Dynamic adaptation to the environment as information sources and services become available or unavailable.
Fusion of information from heterogeneous and multimedia sources, to provide a means of gathering related information that spans multiple information sources.
Abstraction and aggregation of information to levels appropriate to the needs of the user, to save the user from unwanted hand filtering and analysis.
Perpetual queries to allow the user to focus on how information is changing, for instance to be able to look for trends or for unexpected events.

Many of the services required for these features have been developed, but only as independent disciplines without a connected framework. The goal of the InfoSleuth project is to provide an integrated solution to these and other information-related problems, in a manner that can grow and be extended as the individual problems are tackled and solved.

Application Features

Thus, InfoSleuth provides a unifying framework for selectively and dynamically leveraging and combining fuctionality provided by disparate classes of systems, including systems that:

Map or relate data sources to ontological concepts, in the spirit of federated and multi- databases.
Try to locate the right information as do both multidatabase systems and search engines.
Try to classify information according to a hierarchy of concepts, incorporating functionality provided manually by generalized or specialized portal sites.
Provide delivery of personalized information and alerts, such as information channels.
Provide data analysis and decision support as do current data warehousing and data mining tools.
Extract or aggregate information and resolve semantic heterogeneity, as do information retrieval and context mapping tools.

For instance, an InfoSleuth application might

Continually collect information from a specific set of databases, document repositories, and focused areas on the World Wide Web,
Given knoweldge of the structure of the retrieved documents, filter them for specific concepts and data,
Annotate the filtered documents with specific data retrieved from the structured databases and then
Use a data mining tool to look at the correlated data and analyze it for deviation from its normal trend.

Underlying Technologies

InfoSleuth's flexibility comes from its use of several underlying technologies, including agents, ontologies, service matching and interaction templates.

An agent is an autonomous software process that activate external services or encapsulate internal algorithms. Agents also have the facility to communicate with other agents using an agent communication language. They are also capable of delegating tasks that they cannot perform themselves.

An ontology is user-centric descriptions of a specific application domain or service. In an InfoSleuth system, ontologies are used by agents to advertise, or describe, their capabilities, including the services they offer and the information they have access to.

Service matching is the ability to match user requests for services to advertised agent capabilities. InfoSleuth provides the ability to match not only at the level of the interface that an agent presents (which would be similar to current distributed object matching), but also using deeper sets of criteria such as the semantics of how the agent operates.

Interaction templates declare information and control flow related to specific types of services the InfoSleuth agents can provide. These are parametrized with respect to the services needed, which in turn hopefully match to existing agent types.

A good technical overview of the InfoSleuth system can be found in the paper:

Active Information Gathering in InfoSleuth.

Application overviews are provided in the papers:

Use of InfoSleuth to Coordinate Information Acquisition, Tracking and Analysis in Complex Applications and

Agent-based Semantic Interoperability in InfoSleuth

Index to InfoSleuth-Related Material

InfoSleuth Home	This is the link back to the home page for the InfoSleuth project at Telcordia Technologies.
Applications of InfoSleuth	This describes some of the applications and prototypes that have been implemented using the InfoSleuth agent architecture.
Features and Uses of InfoSleuth	This describes some of the ways that an application designer can use InfoSleuth, and the InfoSleuth features that can support your applications.
InfoSleuth-Related Publications	This is a list of selected research publications on InfoSleuth and its related technologies.