The days of the critic may soon be numbered if a new system that will allow Internet users to have an input into the reviewing process comes to fruition.

Ulysses, a project at UCD, aims to develop a system, using artificial intelligence, that would collate the views of any number of people into one single or best review.

The system will give its subscribers, who would submit reviews, a sense of ownership over the review process and perhaps make such reviews less subjective. It will also create a practically maintenance free system for the companies running the sites that carry the reviews.

Ulysses is a database-driven entertainment website that employs user profiling and case-based reasoning to provide users with personalised guides for Dublin.

Ms Marie Angela Ferrario, who is undertaking the project as her PhD project, said the core concept was that users were allowed to update the database.

Although similar technology already exists, Ms Ferrario is going one step further by developing a system that will monitor the behaviour of normal users and check their interaction with the system, to ensure the accuracy of the information a user is inputting.

For example, she said, a review of a nightclub by someone of 83 years of age would have to be considered unsuitable, because it would offer a very different perspective from those of people who normally frequented the place.

Everybody from authors and musicians to film-makers and restaurateurs could soon be living in fear of review on the Internet.

Ulysses is part of the CBRNET projects, which investigate the development of the next generation of Web-based Case-based reasoning (CBR) software. It is a collaborative project with Trinity College Dublin.

The case-based reasoning systems, on which Ulysses is based, act very much like humans in order to solve real-life problems.

A human being is much more efficient at taking decisions in a situation if they have memories of similar circumstances.

By readapting this experience and knowledge to a new situation they can solve a problem quicker than if they were approaching it from scratch.

Likewise CBR systems solve new problems by utilising specific knowledge of past experience and using a basic competence that is encoded within a corpus of previous problem-solving episodes called casebase.

CBR is an incremental system that learns and builds up experience as it goes along, as each experience is retained and is available for future use.

The system will probably be most suited to online communities where you have the same users visiting the same site regularly, with similar interests and views to other people in the community. One of the main problems that such a system would face would be the prospect of individual users submitting deliberately malicious or untruthful reviews just to either mess up the system or because of a personal dislike they may have for the subject of the review, be it a restaurant, bar or writer.

However, the system will build a profile of the person submitting reviews and decide whether the views they express are consistent, and are therefore a reliable source of opinion. If one user suddenly wrote a highly critical review of something that the system had identified as a personal preference, the alarm bells would go off and the review would be ignored.

The system would then identify the most consistent and therefore trustworthy reviewers and use their views on certain topics more heavily than those of somebody who was erratic.

The system also has a personalisation layer and collaborative filtering, which gives the system its ability to bring together groups of users with similar preferences and tastes. This would mean that people could receive a personalised stream of reviews of restaurants that they would be more likely to be interested in, and the system may suggest new films, books or bars that they would be more likely to enjoy.

The challenge of actually getting Internet users to submit reviews on a regular basis on a range of different topics to a site may be the greatest obstacle to Ulysses's success.

However, the system's developers believe that incentive and discount schemes for areas within the user's interests, which would reward people for submitting reviews, may get around this problem.

Ms Ferrario said that many databases such as music databases on the Internet are not able to have constantly up-to-date reviews of the latest releases because of the sheer volume of new material.

However, a collaborative review system could overcome this, as such a system would potentially have access to millions of reviewers.

The project has been running for two years and Ms Ferrario hopes to have the system fully developed by April of next year.

The peer review concept on which Ulysses is based could also be applicable to corporate environments for problem-solving issues by gathering a wide range of views and finding the most acceptable way, for all contributors, of approaching a problem.

Knowledge management and maintenance is a central issue for today's large-scale information systems, especially for the current generation of database-driven information applications on the Internet.

The sheer volume of information that these systems have to deal with means that significant resources have to be devoted to maintaining such systems.

The conventional approach has been to employ a human maintainer whose job it is to monitor and update the information database.

Although distributed maintenance has been used recently on the Internet, keeping users maintaining the system and the quality of maintenance presents problems.

Ulysses is seeking to find an intelligent way, using artificial intelligence, of overcoming these problems.

Ulysses' website can be accessed at http://boffin.ucd.ie/Ulysses2000/.