Information Retrieval under Constricted Bandwidth

Abstract

Information filtering systems are designed to examine a stream of dynamically generated documents and display only those which are relevant to a user’s interests. Information retrieval systems, by contrast, are designed to respond once to each query based on the contents of a relatively static set of contents.

Most significant distinction between filtering and retrieval is the duration over which the users’ need to settled in for desire of information must be remodeled for rapid access. Users may have many interests, and the lifetime of a particular interest is widely variable in relatively dislocated environments.

In an interactive information retrieval architecture, intelligent agents can be utilized to route queries and update documents to relative locations. Between the “user-agents” (the interactive queries) and the “back-ends” (the information provider), a kernel is needed to be proactive, adaptive, wide spread, heterogeneous, dynamic and able to manage immensely huge interactivities, and to be able to react either gradually or abruptly.

On the other hand, information filtering systems must operate over relatively long time scales, and hence, the ability to observe, model and adapt to their persistence, variation and interaction of interests are important. Some progress has already been made in this direction. Utilizing query recall and modification methods of many information retrieval systems, some explicitly continuing information interests have been identified including the modifications that those interests have undergone.

Although the structure of the resulting search space is becoming more complex, machine learning techniques should obtain benefit from information systems related issues of constraints observed in actual demands/ interests. The richer vocabulary of concepts about those interests can also be used to enhance the quality of user participation in the filter design processes. Several ongoing processes in these aspects have been reviewed with follow-ups and suggestions to improve algorithms by accommodating provision of intelligent dynamic updates.