2011 7 th International Conference on IT in Asia (CITA) A Recommender System Infrastructure to Allow Educational Metadata Reasoning Subtitle as needed (paper subtitle) Tiago Thompsen Primo, Rosa Vicari Instituto de Inforática UFRGS Porto Alegre, Rio Grande do Sul {ttprimo, rosa}@inf.ufrgs.br Abstract—This work presents a recommender infrastructure for educational material that is described with metadata. For its operation we propose the use of the OBAA standard, an extension to IEEE LOM that provides interoperability among hardware platforms. We also suggest the need in reusing the user profiles information that are available through FOAF metadata and extend them with personalized educational information. In order to test this infrastructure, we proposed Lassique, an application that makes reasoning over a metadata ontology to filter educational material suggested by a Collaborative Filtering Algorithm. Keywords- Recommender Systems; Semantic Web; Educational Standards I. INTRODUCTION Recommender Systems (RS) has been a research topic by some time [1]. Those methods suggest items based on user profile observations. They can be classified as Collaborative Filtering (CF) methods, where the recommendations are made by the similarity between the users evaluations on items; content Based Filtering where the recommendations are made considering the likeness with the users history with items description, or by mixing them, which is known as Hybrid Methods. Such methods have been successfully used in entertainment domains, but they need improvement in educational domain. Different from songs or movies, educational material is designed based on pedagogical strategies and learning styles. Due to its complex nature, it is common that this materials require description that associated with metadata information. This work focuses on the process of reasoning over metadata information in order to support educational RS applications. To accomplish this goal, the work focuses mainly on five definitions: 1) Metadata schema definition; 2) Educational content repository definition; 3) User profiles store/retrieve definition; 4) Recommendation algorithm definition; 5) Metadata reasoning type definition. The OBAA standard is chosen for the metadata schema [2]. This standard is derived from the IEEE LOM standard, which extends metadata information in order to adequate it to Brazilian educational context. These extensions encompass pedagogical aspects and multi-platform interoperability. The educational materials are available in Brazilian repositories. The materials are geographically distributed and described with the OBAA standard. The educational materials can be videos, presentations, books and so on. An educational material can be regarded by any content that helps in the learning processes. To describe the user profiles, we propose the use of the Semantic Web Standard FOAF with some extensions that is inherited from the OBAA in collaboration with the educational domains. The use of this Semantic Web technology can provide compatibility with applications of the WEB 2.0 and obtain access to specific user information that are available on other web sites. This alternative can also be used as an option for a common RS problem defined as cold- start 1 . The proposed reasoning model can be used with any RS algorithm. It is part of this infrastructure where the educational material is processed according to a specific user profile. For example, a user that access educational materials from his cell phone must receive only the compatible materials with such device. In order to use the proposed infrastructure we are developing an prototype prototype called Lassique. This prototype makes recommendations according to the user hardware platform and learning styles. The current prototype is developed as a set of rules. It is able to distinguish educational recommendations that are compatible with the user platform (e.g. mobile, digital TV and Internet). We use CF algorithm for the recommendation process. The rest of the paper is distributed mainly on: theoretical core as composed in section II. We present our major motivational references and related work. The RS infrastructure Core is composed in section III. We present our RS infrastructure based on bibliographical study, The Lassique in section IV is the case study of the proposed infrastructure. 1 The cold-start problem is related to the fact that an RS does not know the ser interests on the moment that he/she joins an environment 978-1-61284-130-4/11/$26.00 ©2011 IEEE