Recommender systems and 2.0 web

Manuela I. Martín Vicente | Researcher

Recommender systems arose in view of the information overload present in numerous domains, which makes it difficult for the users to identify those items (commercial products, TV contents, educational courses, etc.) that are relevant to them.

These tools provide the users with personalized suggestions, selecting, from among the large amount of available options, those items that best match the preferences of each individual. For this purpose, the personalization strategies of these systems rely on the information gathered in personal profiles.

Generating and updating such profiles are key tasks in any recommender system, because they determine the quality of the resulting suggestions. However, information about users’ interests is limited to that acquired from their interaction with the system, either explicitly (through quantitative or qualitative ratings) or implicitly (e.g. purchase histories in an e-commerce site).

In the Web 2.0, users are no longer mere consumers of information, but also producers and distributors. Millions of people spend long hours in diverse social media sites (blogs, social networks, specialized forums, etc.), communicating and sharing information in the same way they do in the real world. Hence their great potential as a source of knowledge for recommender systems.

Nowadays, one of the biggest challenges is the analysis of the plethora of data available in social media. On the one hand, since users’ posts often reflect their interests, it will be possible to add new preferences to their profiles without requiring their interaction with the system. On the other hand, a great amount of useful information can be inferred from the different connections established among users in social media, which can be incorporated to the system in order to enhance the recommendation process.

Gradiant has opened several lines of work in the personalization field and it is currently developing a project in business intelligence that includes social media profiling (CELTIC).