A new temporal and social PMF-based method to predict users' interests in micro-blogging

https://doi.org/10.1016/j.dss.2013.02.007Get rights and content

Abstract

Micro-blogging is becoming an increasingly popular social media platform where users can discover interesting information about the real world and especially corporations are able to understand customers' demands. The fast diffusion of information and the convenience of micro-blogging have resulted in large audiences sharing their daily activities, exchanging opinions and establishing friendships with others. By analyzing the user-generated contents, one can explore users' potential interests, which helps micro-blogging provide users with better personalized information services. Users' behaviors are affected by opinions of their friends and changes in their interests over time. Based on these intuitions, in this paper we propose a temporal and social probabilistic matrix factorization model to predict users' potential interests in micro-blogging. By exploiting the matrix factorization technique to learn latent features of users and topics, our model analyzes the impacts of time information and users' activities, including posting of tweets and establishing friendships with others, on the latent feature space of users and topics of their interests. The proposed model provides a unified way to fuse the time information and the social network structure to predict users' future interests accurately. The experimental results on Sina-weibo, one of the most popular micro-blogging sites in China, demonstrate the efficiency and effectiveness of our proposed model.

Highlights

► A new temporal and social PMF-based method is proposed to predict users' interests in micro-blogging. ► The model provides a unified way to fuse the time information and the social network structure. ► The experimental results demonstrate the efficiency and effectiveness of the model.

Introduction

With the rapid development of the Internet, social media has played an increasingly significant role in our everyday lives, providing reports on world events, improving enterprise influence through social media marketing and so on. Micro-blogging is becoming one of the most popular social media platforms where users can share their daily activities, exchange opinions, publish posts on some trending topics and follow others to get relevant information about their interested topics. If A is following B, B is called A's friend, and A is called B's follower. Thus friendships can either be reciprocated or one-way [10]. The convenience and high frequency of updates of information in micro-blogging have attracted a large number of users to actively participate. For example, Sina-weibo, one of the most popular micro-blogging services in China, has had over 300 M unique visitors since December 31, 2011 and around 100 M tweets per day.3 Nowadays, more and more corporations are registering user accounts on Sina-weibo for marketing. For example, Nokia successfully held a product release conference for N8 on August 25, 2010.

Corporations utilize micro-blogging not only to introduce their products but also to formulate customer-driven marketing plans by obtaining rich information such as which features customers consider important in their products, new market dynamics and so on. Micro-blogging has become an important e-commerce marketing channel, and the promotion of merchandise is accessible to micro-blogging users around every corner of this platform. Users' interest plays a vital role in the process of micro-blogging's development [7], which influences the effect of micro-blogging marketing soon afterwards. The research findings in [5] point out that the accurate prediction of users' interest will improve their satisfaction and promote their buying decisions, which will increase the e-commerce business benefits undoubtedly. Decision makers will also benefit from the interest prediction work; Chen and Cheng and Zhao and Lu [4], [28] proposed that decision makers need to grasp users' interest for raising up their satisfaction and providing reasonable results. Asur and Huberman [2] tell us that the box-office revenues of movies can be successfully forecasted in advance of their release by analyzing users' interest in micro-blogging. If the forecasted box-office revenues are below expectations, decision makers can provide ways of film promotion with some incentives in time, or some other methods for coming up to their expectations. All in all, it's valuable and meaningful to predict users' interest in social media, whether for e-commerce business or decision makers. On one hand, it can help micro-blogging systems provide users with better personalized information and advertising services to motivate users to be more active. On the other hand, corporations can easily capture users' future interest and make marketing decisions.

In micro-blogging, trending topics are popular topics, which may be related to emerging events and breaking news or topics under the discussion by a large fraction of micro-blogging users [19]. In Sina-weibo, trending topics are edited and complemented, and users are available to enter into the trending topics and take part in the discussion by publishing posts on them. Trending topics often have a clear meaning [11], mainly relating to entertainment, sports, current events and so on. If the user is interested in a trending topic, he/she may publish posts on it. In other words, if a user has published posts on a trending topic, it shows that the user has interest in this topic. Posts published on some trending topics can well reflect users' interests. Thus, in this paper, we use trending topics to represent users' interests.

Despite the importance of user interest prediction in micro-blogging, existing works on micro-blogging mainly focus on mining users' current interests; little work has been done on prediction of users' potential interests. Nori et al. [20] focused on computing the similarity between a user and a set of resources to predict the user's interests. However, this method ignores the influence of the user's friends on his/her interests. Besides, it doesn't take the evolution of interests into account.

Some researchers have suggested that users are more affected by opinions of their peers than influentials [21], [24], [25]. By comparing quality of recommendations made by recommender systems to recommendations made by users' friends, Sinha et al. [23] showed that users' friends consistently provided better recommendations than recommender systems. In social recommendation, making use of the information in a social graph has recently been receiving increasing research attention. The experimental results [6], [9], [15], [16], [17] show that fusing the social network structure of users with the user–item rating matrix can help make more accurate and personalized recommendations in a social rating network. From this viewpoint, a user's social network affects users' behaviors on the Web.

Additionally, interests of Web users change over time. For time-aware recommendation, it is important to capture users' temporal preferences to make more accurate recommendations. Xiang et al. [12] stated that users' dynamic preferences are affected by both their long-term and short-term preferences. That means their interests may vary over time. In this case, to capture users' temporal preferences, it is necessary to follow the evolution of users' preferences. Generally, recent preferences may play a more important role in predicting current preferences while earlier preferences have relatively smaller contribution to final recommendation. Especially in micro-blogging, the rich information and frequent updates make users' interests more extensive and changeable over time. Therefore, to improve the accuracy of prediction of users' interests in micro-blogging, both the social network structure and time information should be taken into consideration.

SocialMF [9] is an effective method for detecting users' interests by exploiting the matrix factorization techniques and analyzing the influence of users' friendships on their interests. Based on this model, we propose a temporal and social probabilistic matrix factorization model (TS-PMF) which fuses on social influence and the time information to predict users' interests in micro-blogging. Following the evolution of users' interests, to import time information in our model, we make the latent features of users and topics associated with their previous latent features by adopting an exponential time decay function. Using this idea, our approach accurately describes the change of the distribution of the latent feature space of users' interests. The proposed model can reflect the impacts of users' interest evolution and users' friendships on their future interests, thus realizing the prediction of users' interests. The experimental results on Sina-weibo demonstrate that our model can improve the quality of prediction.

The remainder of this paper is organized as follows. In Section 2, some related work is discussed. Section 3 introduces the proposed model. Results of the detailed experimental analysis are presented in Section 4. Finally, we conclude the paper and present some directions for future work in Section 5.

Section snippets

Related work

In this work we propose a novel model to predict users' interests in micro-blogging. Our work is related to prediction of user interest in micro-blogging, trust-aware recommendation and time-aware recommendation. In this section we review the related works.

The proposed user interest prediction model

In this section, first we introduce the theoretical background for our proposed model. Second, we illustrate how to fuse social network structure and time information in our TS-PMF model to predict users' interests in micro-blogging.

Experimental analysis

In this section, comprehensive and systematic analyses are conducted to evaluate the proposed user interest prediction model. The process of collection of the dataset used in the empirical work and the evaluation metrics are presented first. Next, we explain the purpose of our experiments in detail. Finally, the performance of our model is compared with results of the other four models; results of the comparison verify the efficacy of the proposed model.

Conclusions and future work

Micro-blogging is one of the most popular social media platforms where the convenience, high update frequency and rich information have attracted millions of active users to join in. Users can publish posts on their daily lives and some trending topics. Trending topics are featured prominently to provide users with an up-to-date glimpse of what is happening in the real world and clearly reflect users' interests. Users enthusiastically follow other users they are interested in to get relevant

Acknowledgments

This research is supported by the NNSFC grants (no. 61172106, no. 71090402, no. 71002064) and the BJNSF grant (no. 4112062).

Hongyun Bao is a Ph.D. candidate in the State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences. She received her B.S. degree in School of Mathematical Sciences from Capital Normal University, China, in 2008. Her research interests include information retrieval and web/text mining.

References (30)

  • H. Ma et al.

    Learning to recommend with trust and distrust relationships

  • J.L. Herlocker et al.

    Evaluating collaborative filtering recommender systems

    ACM Transactions on information systems

    (2004)
  • M. Jamali et al.

    A matrix factorization technique with trust propagation for recommendation in social networks

  • A. Java et al.

    Why we twitter: understanding microblogging usage and communities

  • H. Kwak et al.

    What is Twitter, a social network or a news media?

  • Cited by (64)

    • The evolution of social media influence - A literature review and research agenda

      2022, International Journal of Information Management Data Insights
      Citation Excerpt :

      Van Dam and Van De Velden (2015) had proposed the methodology for identifying the customers through online trajectories. Bao et al., and Gao (2013) had proposed the model for predicting indivdual's interest through their activities on social media platforms. Li and Shiu (2012); Lin et al., and Wu (2015); and Susarla et al., and Tan (2012) had proposed a mechanism for diffusing advertisements and messages over social media platforms.

    • Online hotel rating prediction through a dynamic weighted ordered probit model

      2022, Digital Signal Processing: A Review Journal
      Citation Excerpt :

      A deep MF method is introduced in [31] for nonlinear matrix completion. Other probabilistic MF techniques have been proposed to deal with data sparsity, scalability, and prediction quality [32][33][34]. However, the majority of these methods require training and they cannot support an online recommendation system.

    View all citing articles on Scopus

    Hongyun Bao is a Ph.D. candidate in the State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences. She received her B.S. degree in School of Mathematical Sciences from Capital Normal University, China, in 2008. Her research interests include information retrieval and web/text mining.

    Qiudan Li (corresponding author) is an Associate Professor in the State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences. She received a Ph.D. in Computer Science from Da Lian University of Technology, China, in 2004. Her research interests include web mining and mobile commerce applications. Her articles are published in Communications of the AIS, Decision Support Systems, Journal of the American Society for Information Science and Technology, IEEE Transactions on SMC, and Expert Systems with Applications.

    Stephen Shaoyi Liao is a Professor at the Department of Information Systems and director of Advanced Transportation Information Systems Research Center, City University of Hong Kong. He is also a Visiting Professor and a Ph.D. Supervisor in USTC and Southwest Jiaotong University. He received a bachelor's degree from Beijing University and a Ph.D. from the University of Aix-Marseille III and Institute of France Telecom. His research focuses on use of IT in e-business systems and transportation systems. His articles have been published in MISQ, Decision Support Systems, IEEE Transactions, Communications of the ACM, Information Science, Computer Software and other SCI journals.

    Shuangyong Song is a Ph.D. candidate in the State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences. He received his B.S. degree in Biomedical Engineering from Beijing Jiaotong University, China, in 2007. His research interests include information retrieval and web/text mining.

    Heng Gao is a Master Candidate in the State Key Laboratory of Management and Control for Complex Systems, Institute of Automation, Chinese Academy of Sciences. He received his B.E. degree in Computer Science from China University of Mining and Technology, China, in 2010. His research interests include information retrieval, web/text mining and community question answering.

    1

    Tel.: + 86 10 62636334.

    2

    Tel.: + 852 34427552.

    View full text