LoTrust: A Social Trust Level Model based on Time-Aware Social Interactions and Interests Similarity Ahlem Kala¨ ı MIR@CL Laboratory SFAX University, Tunisia ahlem.kalai@gmail.com Abdelghani Wafa MIR@CL Laboratory SFAX University, Tunisia abdelghani wafa@hotmail.fr Corinne Amel Zayani MIR@CL Laboratory SFAX University, Tunisia corinne.zayani@isecs.rnu.tn Ikram Amous MIR@CL Laboratory SFAX University, Tunisia ikram.amous@isecs.rnu.tn Abstract—With the immense growth of online social applica- tions, trust plays a more and more important role in connecting users to each other, sharing their personal information and attracting him to receive recommendations. Therefore, how to obtain trust relationships through mining online social networks became a critical issue. To calculate the level of trust between two users, many computational trust models are proposed which mainly rely on the social network structure, the explicit trust from user to another, the users’ behaviors, or the users’ sim- ilarity, etc. However, the majority of these models ignored the temporal factor. In this paper, we propose a trust relationship detection mechanism from an egocentric social network in order to compute the trust level between an active user and his directed friends. We propose a Level of social Trust model, that we called LoTrust, which is suitable for personalized recommendation purpose. This computational model founded on novel trust metric which is based not only on the users’ interests similarity according to their semantic social profiles (RDF/FOAF), but also takes into account the time factor of the users’ active interactions (e.g comments, share photo, wall posts, messages). We perform experiments on real life dataset extracted from Facebook. The experimental results demonstrated how our LoTrust model produces satisfactory results than other computational models. I. I NTRODUCTION The Web has dramatically evolved to an interactive and social environment called Web -based social networks (e.g Facebook 1 , Twitter 2 ). These Online Social Networks (OSNs) provide a space in which people can share information and can connect with one another. In this open environment, the user-generated content (e.g discussion, social profiles, video and photo feeds, reviews and ratings of anything) is very tremendous [11] and created by the different users’ activities or behaviors. This content can be reliable or untrustworthy to the users. Despite all measures taken for privacy and security in OSNs, there is no certainty of trust [34]. For this reason, trust [10] [9] plays an important role in addressing both information overload and credibility problems [52] [37]. Trust is a filed research which has recently been attracting scientists from many domains including sociology, psychology, economics, 1 http://www.facebook.com 2 http://www.twitter.com and computer science [33]. In our research context, we are in- terested how detect and measure trust in OSNs. With so much user-interaction and user-generated content, the establishing of online trust mining mechanisms [52] is emerged in recent years. If trust can be detected accurately, the user can then use this trust to make decisions. In social Web, trust is a complex concept influenced by many factors (personality and social) which online systems cannot yet model it completely [52]. In this context, some research work proposed a computational trust models [33] which can improve the social recommender systems [49]. Other existing work proposed a mining trust mechanisms from OSNs [52] [48]. The majority of these researches only rely on the network structure in order to generate trusted graph like TidalTrust [10], SWTrust [17], or based on the explicit trust like TrustWalker [16], Epinions 3 . For this reason, the accuracy trust inferring cannot be guaranteed due to lack of some relevant and proper information. Reputation of one another [19] [46], profile similarity [50] [5] [12], explicit rating (e.g. TidalTrust [30]) are frequently used information as influence factors which affect the trust between users. Since various OSNs support different types of social activities or interactions (e.g. tag photo, post comment, write on friend’s wall, send message) which are performed by the users, some research studies [24] [22] [42] [29] [31] have then used these interactions as another factor to compute the social trust. One shortcoming of all of the above studies that they have neglected the time factor. In reality, every interaction between two friends occurs at a given time, in a given situation (or context) and in a particular location. Therefore, the interactions change over time. Hence, the time is an important factor to capture the change in the behavior of an individual. Josang et al. [18] proposed to rank the friends by age of their social friendships and they considered the newer friends as the most trustworthy. On the contrary, Moghaddam et al. [27] are considered the older friends more trusted than newer ones. We think that these assumptions are not always valid depending on the 3 http://epinions.com