A news aggregator website that ropes in articles from 10 Maltese websites and automatically categorises them as positive, negative or neutral has been developed by a team of students.

The News Marketplace website – www.thenewsmarketplace.com – that was launched on December 31, is now open for feedback, while the team of 10 enthusiastic second-year BSc ICT students continues to brainstorm ways to improve the innovative tool.

The main idea behind this web science assignment was to archive all local articles in one user-friendly place and categorise them semantically.

The website relies on foreign third-party engines, that analyse the articles’ sentiment. It has a searchable archive database that runs on filters chosen by web users.

The services employed for the website were all free of charge. However, the students had to pay for the domain name and webhosting.

Facing time and financial constrains, the students’ resources were limited. However, the tools they employed were quite flexible, which allowed them to refine their website.

The team is still battling with various hitches. Short stories do not encompass enough data for reasonable study, which hinders fair analysis. Background information might also change the nature of a story, where an article about someone’s death might automatically be labelled as positive because comments about the person’s life change its feel.

Such textual analysis is still at the experimental stages in Malta, and discrepancies between the language used in Maltese news articles and that employed by the classifier make it more difficult to decipher the nature of the articles.

Thus, the team would like to manually train the classifier itself or employ an alternative engine within its budget.