Recommender systems have developed in parallel with the web. They were initially based on demographic, content-based and collaborative filtering. Currently, these systems are incorporating social information. In the future, they will use implicit, local and personal information from the Internet of things. This article provides an overview of recommender systems as well as collaborative filtering methods and algorithms; it also explains their evolution, provides an original classification for these systems, identifies areas of future implementation and develops certain areas selected for past, present or future importance.

http://romisatriawahono.net/lecture/rm/survey/information%20retrieval/Bobadilla%20-%20Recommender%20Systems%20-%202013.pdf

Recommender systems survey

Information and Communication Technologies in Tourism

Online vendors employ collaborative filtering algorithms to provide recommendations to their customers so that they can increase their sales and profits. Although recommendation schemes are successful in e-commerce sites, they are vulnerable to shilling or profile injection attacks. On one hand, online shopping sites utilize collaborative filtering schemes to enhance their competitive edge over other companies. On the other hand, malicious users and/or competing vendors might decide to insert fake profiles into the user-item matrices in such a way so that they can affect the predicted ratings on behalf of their advantages. In the past decade, various studies have been conducted to scrutinize different shilling attacks strategies, profile injection attack types, shilling attack detection schemes, robust algorithms proposed to overcome such attacks, and evaluate them with respect to accuracy, cost/benefit, and overall performance. Due to their popularity and importance, we survey about shilling attacks in collaborative filtering algorithms. Giving an overall picture about various shilling attack types by introducing new classification attributes is imperative for further research. Explaining shilling attack detection schemes in detail and robust algorithms proposed so far might open a lead to develop new detection schemes and enhance such robust algorithms further, even propose new ones. Thus, we describe various attack types and introduce new dimensions for attack classification. Detailed description of the proposed detection and robust recommendation algorithms are given. Moreover, we briefly explain evaluation of the proposed schemes. We conclude the paper by discussing various open questions.

Shilling attacks against recommender systems: a comprehensive survey

Providing high quality recommendations is important for e-commerce systems to assist users in making effective selection decisions from a plethora of choices. Collaborative filtering is a widely accepted technique to generate recommendations based on the ratings of like-minded users. However, it suffers from several inherent issues such as data sparsity and cold start. To address these problems, we propose a novel method called ''Merge'' to incorporate social trust information (i.e., trusted neighbors explicitly specified by users) in providing recommendations. Specifically, ratings of a user's trusted neighbors are merged to complement and represent the preferences of the user and to find other users with similar preferences (i.e., similar users). In addition, the quality of merged ratings is measured by the confidence considering the number of ratings and the ratio of conflicts between positive and negative opinions. Further, the rating confidence is incorporated into the computation of user similarity. The prediction for a given item is generated by aggregating the ratings of similar users. Experimental results based on three real-world data sets demonstrate that our method outperforms other counterparts both in terms of accuracy and coverage.

/pdf/merging-trust-in-collaborative-filtering-to-alleviate-data-4gxxqcanll.pdf

Merging trust in collaborative filtering to alleviate data sparsity and cold start

In open and dynamic multiagent systems (MASs), agents often need to rely on resources or services provided by other agents to accomplish their goals. During this process, agents are exposed to the risk of being exploited by others. These risks, if not mitigated, can cause serious breakdowns in the operation of MASs and threaten their long-term wellbeing. To protect agents from the uncertainty in the behavior of their interaction partners, the age-old mechanism of trust between human beings is re-contexted into MASs. The basic idea is to let agents self-police the MAS by rating each other on the basis of their observed behavior and basing future interaction decisions on such information. Over the past decade, a large number of trust management models were proposed. However, there is a lack of research effort in several key areas, which are critical to the success of trust management in MASs where human beings and agents coexist. The purpose of this paper is to give an overview of existing research in trust management in MASs. We analyze existing trust models from a game theoretic perspective to highlight the special implications of including human beings in an MAS, and propose a possible research agenda to advance the state of the art in this field.

http://romisatriawahono.net/lecture/rm/survey/pervasive%20computing/Yu%20-%20Multi-Agent%20Trust%20Management%20-%202013.pdf

A Survey of Multi-Agent Trust Management Systems

Disasters and Recommender System: Setting the Research Agenda for Developing Nations

Recommender systems have become a crucial tool for electronic commerce. These systems help in improving customer satisfaction by providing personalized services. Much of the research in recommender systems is done in solving the top--N recommendation problem where the objective is to identify a set of N items the user will most likely prefer. In this paper, we propose a weighted class based hybrid algorithm to solve this problem. Our algorithm combines the strength of both content based and collaborative filtering approaches. In this method, firstly, classes are defined by applying content based and collaborative filtering techniques on ratings data. Subsequently, user's existing profile of items is examined to determine the weightage of each class. Items are then recommended to the user on the basis of weightage attached to each class. We show through experimental results that our algorithm consistently outperforms existing algorithms when applied on MovieLens data set.

Weighted class based hybrid algorithm for top-N recommender systems

INTRODUCTION The approach used for solving the cold start problem in a trust-aware recommender system is to ask a cold start user to select few existing users in the system that they trust. Using the ratings of the items rated by the selected trusted users, recommendations are generated for the cold start user. An implicit assumption made in this solution is that cold start users or users new to the system have knowledge of the existing users in the system. Moreover, even if the cold start user has knowledge of the existing users, it is assumed that he is aware of the ratings given by the existing users to different items. To overcome the shortcomings of the proposed approach, recent research have started focusing on the problem: to whom should the new users connect to? (Victor, Cornelis, De Cock, & Teredesai, 2008). The suggested approach is to identify key figures in the trust network based on their impact on coverage and accuracy of recommendations made and then ask the cold start user to choose few of these users presented before them as trusted users. In this paper, we analyze the trust connections of the cold start users in the dataset from Massa and Avesani (2006b), a real life recommender system dataset, to examine "What are the characteristics of those users that cold start users actually trust?" The objective of our work is to provide insights on the type of users that should be preferred while creating a list of prospective trusted users to be presented before a cold start user. Our paper first discusses the existing parameters that form the basis of selecting prospective trusted users and the reasoning behind it. We then analyze the set of users trusted by cold start users using four parameters: number of outgoing links, number of incoming links, number of items rated, and a hybrid parameter defined as preference score. Using rank correlation coefficient as the metric we show that in the Epinions dataset (Massa & Avesani, 2006a), users with higher number of incoming links are preferred as trusted users as compared to other three parameters. We also examine which parameter gives the best predictive accuracy when selected as trusted user. Our results show that those cold start users that trust users with high preference scores have the benefit of getting more accurate recommendations. TRUST IN WEB BASED SOCIAL NETWORKS Web based social networks (Boyd & Ellison, 2007; Golbeck, 2005) can be defined as follows: Services that are accessible over the web and allow individuals to (1) construct a public or semipublic profile within a system designed specifically to support social network connections, (2) articulate a list of other users with whom they share a connection, and (3) browse their list of connections and those made by others within the system. The study of web based social networks (WBSN) primarily focuses on connections between people. Scholars have examined the behavior of people in social network, how connections are formed and their evolution. WBSNs are a source of rich behavioral data. Data in WBSNs is extracted from user profiles and explicitly made connections between users. In WBSNs, much of the products, services, and features are based around social connections. Social connections are based on trust. Trust-aware recommender systems exploit these social connections or trust data as well profile data of users to create intelligent applications that provide personalized recommendations to users in WBSNs that are relevant and trustworthy. Trust is a social phenomenon. As a social concept, the many facets and influences of social trust has been extensively addressed in sociology and social psychology literature. It is a complex notion as a result it has many subtly different definitions. In Luhmann (1979), trust has been described as a tool for complexity reduction. Mayer, Davis and Schoorman (1995), proposed a trust model that is designed to focus on trust in an organizational setting. …

/pdf/trusting-the-trusted-an-empirical-analysis-2qcgcszhfo.pdf

Sanjog Ray

Papers

Disasters and Recommender System: Setting the Research Agenda for Developing Nations

Weighted class based hybrid algorithm for top-N recommender systems

Trusting the Trusted: An Empirical Analysis