People Recommendation

People-to-people recommendation differ from item recommendations in a number of ways, one of which is that individuals add information to their profile which is often critical in determining a good match. The most critical information can be in the form of free text or personal tags. We explore text-mining techniques to improve classical collaborative filtering methods for a site aimed at matching people who are looking for expert advice on a specific topic. We compare results from a LSA-based text similarity analysis, a simple user-user collaborative filter, and a combination of both methods used to recommend people to meet for a knowledge-sharing website. Evaluations show that LSA similarity has a better precision at low recall rates, whereas collaborative filters have a better precision at higher recall rates. A combination of both can outperform the results of the simpler algorithms.

Traditional recommendation techniques often rely on the user-item rating matrix, which explicitly represents a user's preference among items. Recent studies on recommendations in the scenario of social networks still largely follow this principle. However, the challenge of recommending people to follow in social networks has yet to be studied thoroughly. In this paper, by using the utility instead of ratings and randomly sampling the negative cases in the recommendation log to create a balanced training dataset, we apply the popular matrix factorization techniques to predict whether a user will follow the person recommended or not. The asymmetric factor models are built with an extended item set incorporating the social graph information, which greatly improves the prediction accuracy. Other factors such as sequential patterns, CTR bias, and temporal dynamics are also exploited, which produce promising results on Task 1 of KDD Cup 2012.

On the so-called Web 2.0, numerous Social Networking Sites (SNS) have known a tremendous success as they brought new ways to communicate, interact, learn about people, and share content with them. Besides improving informal social ties, transposing Awareness to the enterprise is very promising for promoting collaboration. However, it can easily lead to information overload, which means productivity loss. With our prototype, we propose a way to stimulate enterprise internal communication while improving the relevance of these interactions with the current working context of their users. This prototype is an efficient tool for cooperation and for discovering relevant people and knowledge.

In this paper we present a methodology for learning user profiles from content shared by people on Social Platforms. Such profiles are specifically tailored to reflect the user's degree of interactivity related to the topics they are writing about. The main novelty in our work is the introduction of Linked Data in the content extraction process and the definition of a specific scores to measure expertise and interactivity.

Web 2.0 technologies provide an opportunity to transform learning into a social experience: social learning can directly benefit from user-generated content analysis. The e-book device is an object central to the learning process; it becomes an ideal medium to promote innovative learning tools. In this paper, we describe how we leverage user generated content in e-books, generally in the form of unstructured annotations, to reduce the latency between professor knowledge, book content and student assimilation.

In this paper we present a methodology for learning user profiles from content shared by people on Social Platforms. Such profiles are specifically tailored to reflect the user's degree of interactivity related to the topics they are writing about. The main novelty in our work is the introduction of Linked Data in the content extraction process and the definition of a specific scores to measure expertise and interactivity.

User profiling techniques have mostly focused on retrieving and representing a user's knowledge, context and interests in order to provide recommendations, personalise search, and build user-adaptive systems. However, building a user profile on a single social network limits the quality and completeness of the profile, especially when interoperability of the profile is key and its reuse on different sites is necessary for providing other types of personalisation. Indeed recent studies have shown that users on the Social Web often use different social networking sites for diverse, and sometimes non-overlapping, purposes and interests. In this paper, we describe our methodology for the automatic creation and aggregation of interoperable and multi-domain user profiles using semantic technologies. Moreover, we propose a user study on different user profiling techniques for social networking websites in general, and for Twitter and Facebook in particular. In this regard, based on the results of our user evaluation, we investigate (i) the accuracy of different methodologies for profiling, (ii) the effect of time decay functions on ranking user interests, and (iii) the benefits of merging different user models using semantic technologies.

In this paper we present a methodology for learning user profiles from content shared by people on Social Platforms. Such profiles are specifically tailored to reflect the user's degree of interactivity related to the topics they are writing about. The main novelty in our work is the introduction of Linked Data in the content extraction process and the definition of a specific scores to measure expertise and interactivity.