This paper describes a multiagent recommender approach based on the collaboration of multiple agents exchanging information stored in their local knowledge bases. A recommendation request is divided into subtasks handled by different agents, each one maintaining incomplete information that may be useful to compose a recommendation. Each agent has a Distributed Truth Maintenance component that helps to keep the integrity of its knowledge base. Further, we show a case study in the tourism domain where agents collaborate to recommend a travel package to the user. In order to help the coordination among the agents during the recommendation, the Distributed Constraint Optimization approach is applied in the search process.
The internet is a rich source of information where the user can find almost everything s/he is looking for. In the last years, the internet has grown exponentially and the information overload problem has appeared. In order to deal with these issues, Recommender Systems have been developed (Resnick et al. 1994). The main characteristic of recommender system is the ability to aggregate information and to match the recommendations with the information people is looking for. Despite recommender systems being efficient, for some recommender applications it is possible that a single information source does not contain the complete information needed for the recommendation.
What recommender systems have in common is an emphasis on leveraging social processes for the purpose of improving information access. Typically, most of the current breed of recommender systems are Internet services with a twofold purpose: providing tailored recommendations and building communities. The issue we focus on here is how to make recommender systems work in organizations and for organizations. Moving from the Internet to Intranets requires shifting the primary focus from sharing recommendations to sharing knowledge and from community-building to community support. Moving recommender systems from the Internet onto Intranets also means turning "leisure-ware" into groupware, creating both new challenges and new opportunities.
Since 2010, we have built and maintained LensKit, an open-source toolkit for building, researching, and learning about recommender systems. We have successfully used the software in a wide range of recommender systems experiments, to support education in traditional classroom and online settings, and as the algorithmic backend for user-facing recommendation services in movies and books. This experience, along with community feedback, has surfaced a number of challenges with LensKit's design and environmental choices. In response to these challenges, we are developing a new set of tools that leverage the PyData stack to enable the kinds of research experiments and educational experiences that we have been able to deliver with LensKit, along with new experimental structures that the existing code makes difficult. The result is a set of research tools that should significantly increase research velocity and provide much smoother integration with other software such as Keras while maintaining the same level of reproducibility as a LensKit experiment. In this paper, we reflect on the LensKit project, particularly on our experience using it for offline evaluation experiments, and describe the next-generation LKPY tools for enabling new offline evaluations and experiments with flexible, open-ended designs and well-tested evaluation primitives.
Students often need guidance in choosing adequate courses to complete their academic degrees. Course recommender systems have been suggested in the literature as a tool to help students make informed course selections. Although a variety of techniques have been proposed in these course recommender systems, combining data mining with user ratings in order to improve the recommendation has never been done before. This paper presents RARE, a course Recommender system based on Association RulEs, which incorporates a data mining process together with user ratings in recommendation. Starting from a history of real data, it discovers significant rules that associate academic courses followed by former students. These rules are later used to infer recommendations. In order to benefit from the current students' opinions, RARE also offers to users the possibility to rate the recommendations, thus leading to an improvement of the rules. Therefore, RARE combines the benefits of both former students' experience and current students' ratings in order to recommend the most relevant courses to its users.