SoPRa: A New Social Personalized Ranking Function for Improving Web Search Mohamed Reda Bouadjenek * PRiSM Laboratory Versailles University mrb@prism.uvsq.fr Hakim Hacid * Sidetrade 114 Rue Gallieni, 92100 Boulogne-Billancourt, France hhacid@sidetrade.com Mokrane Bouzeghoub PRiSM Laboratory Versailles University mokrane.bouzeghoub@ prism.uvsq.fr ABSTRACT We present in this paper a contribution to IR modeling by proposing a new ranking function called SoPRa that consid- ers the social dimension of the Web. This social dimension is any social information that surrounds documents along with the social context of users. Currently, our approach re- lies on folksonomies for extracting these social contexts, but it can be extended to use any social meta-data, e.g. com- ments, ratings, tweets, etc. The evaluation performed on our approach shows its benefits for personalized search. Categories and Subject Descriptors: H.3.3 [Informa- tion Systems]: Information Search and Retrieval General Terms: Algorithms, Experimentation. Keywords: Information Retrieval, Social networks. 1. INTRODUCTION Nowadays, the Web is becoming more and more complex with the socialization and interaction between individuals and objects. This evolution is known as social Web, which includes linking people through the World Wide Web. This is mainly done through platforms such as Facebook, Twitter, or YouTube, where users can comment, spread, share and tag information and resources. The social Web leaded to facilitate the implication of users in the enrichment of the social context of web pages 1 . Especially, it allows users to freely tag web pages with annotations. These annotations can be easily used to get an intuition about the content of web pages to which they are related. Hence, several research works ([4, 7, 9, 19]) reported that adding tags to the content of a document enhances the search quality, as they are good summaries for documents. In particular, tags are useful for documents that contain few terms. * This work has been mainly done when the authors was at Bell Labs France, Centre de Villarceaux. 1 In this paper, we also refer to web pages as documents or resources. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full cita- tion on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or re- publish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from permissions@acm.org. SIGIR’13, July 28–August 1, 2013, Dublin, Ireland. Copyright 2013 ACM 978-1-4503-2034-4/13/07 ...$15.00. In such a context, classic models of Information Retrieval (IR) should be adapted by considering (i) the social context that surrounds web pages and resources, e.g. their anno- tations, their associated comments, their ratings, etc. and (ii) the social context of users, e.g. their used tags, their comments, their trustworthiness, etc. Exploiting social in- formation has a number of advantages (for IR in particular). First, feedback information in social networks is provided directly by the user, so user interests accurate information can be harvested as people actively express their opinions on social platforms. Second, a huge amount of social infor- mation is published and available with the agreement of the publishers. Exploiting this information should not violate user privacy, in particular social tagging information, which doesn’t contain sensitive information about users. Finally, social resources are often publicly accessible, as most of so- cial networks provide APIs to access their data. In this paper, we are interested in improving the IR model by proposing a new ranking function for documents, while considering the social context of the Web. The approach we are proposing relies on social annotations, which are associ- ated to documents in bookmarking systems but can consider other social metadata, e.g. comments, tweets, etc. In this context, we propose the following contributions: (1) A Social Personalized Ranking function called SoPRa. (2) A method for weighing user profiles and social docu- ments. (3) An extension of SoPRa by considering tagging users individually. (4) An intensive evaluation of SoPRa. The rest of this paper is organized as follows: in Sec- tion 2.1, we present the fundamental concepts, and we for- mally define the problem we tackle. Section 3 presents the related work. Section 4 introduces our approach for rank- ing documents. Experiments are discussed in Section 5. We conclude and provide some future directions in Section 6. 2. BACKGROUND 2.1 Background and notation Social bookmarking systems are based on the techniques of social tagging. The principle is to provide the user with a mean to freely annotate resources on the Web with tags, e.g. URIs in delicious, or images in Flickr. These annotations can be shared with others. This unstructured approach to classification is often referred to as a folksonomy. A folkson- omy is based on the notion of bookmark, which is formally defined as follows: 861