Disconnected Web Browsing

December 1, 1999 - May 1, 2001

This project addresses issues related to personalization and asynchronous operation to support wireless access to the web. Currently, the protocols associated with information browsing have been designed to work in wired environments that are capable of transmitting and receiving large amounts of data without significant delays. However, due to the various resource constraints of the mobile hosts, these protocols do not operate well in a wireless environment. Also, the current model of web browsing leaves the burden of finding relevant information on the user. The user has to search through the documents following links, and find the information he/she needs thereby wasting bandwidth and other resources on tranmitting data that is useless to the user over the wireless links. Moreover, the synchronous nature of the HTTP protocol requires a continuous connection for the duration of the information retrieval process.

We are addressing these issues by developing a proxy based system that replaces the exisiting http client server system with a client-intelligent proxy- intelligent server system, with the clients, proxies and servers constituting a cooperative multiagent system that provide active transcoding, information retrieval, personalization and disconnected management. W3IQ proxy splits information access into a set of asynchronous transaction like operations and also maintains state information while disconnections occur. W3IQ proxy allows the user to specify requests as queries, rather than explicit URLs and takes up the burden of locating the right information on behalf of the user. Thus the W3IQ agent acts as an information broker and a recommender.

The system will also support other interaction formats. For example, the users could ask for monitoring changes in certain web pages, ask to be informed whenever new movies are played at his theatre of choice and also be sent a review of the movie and its rating.

We are also working on Automatic Profile Development by monitoring the ratings of users and dynamically forming groups of users with similar interests within a proxy and also across all the peer proxies. These groups get updated (new groups are created, existing groups are merged, users moved between the groups) as and when more content is rated.

OWL Tweet