Poster/Demo Session

sponsored by LinkedIn

Date: Wednesday, Oct 8, 19:00-22:00
Location: Alexandria/Balboa

Short Papers, Posters and Demos will be presented during the banquet. Short Papers will be presented as Posters.

  • An Analysis of Users’ Propensity Toward Diversity in Recommendations

    by Tommaso Di Noia, Vito Ostuni, Jessica Rosati, Paolo Tomeo and Eugenio Di Sciascio
    Providing very accurate recommendations to end users has been nowadays recognized to be just one of the main tasks a recommender systems must be able to perform. While predicting relevant suggestions, attention needs to be paid in their diversification in order to avoid monotony in recommendation. In this paper we focus on modeling users’ inclination toward selecting diverse items, where diversity is computed by means of content-based item attributes. We then exploit such modeling to present a novel approach to re-rank the list of Top-N items predicted by a recommendation algorithm, in order to foster diversity in the final ranking. Experimental evaluation proves the effectiveness of the proposed approach.

  • Clinical Online Recommendation with Subgroup Rank Feedback

    by Yanan Sui and Joel Burdick
    Many real applications in experimental design need to make decisions online. Each decision leads to a stochastic reward with initially unknown distribution. New decisions are made based on the observations of previous rewards. To maximize the total reward, one needs to solve the tradeoff between exploring different strategies and exploiting currently optimal strategies. This kind of tradeoff problems can be formalized as Multi-armed bandit problem. We recommend strategies in series and generate new recommendations based on noisy rewards of previous strategies. When the reward for a strategy is difficult to quantify, classical bandit algorithms are no longer optimal. This paper, studies the Multi-armed bandit problem with feedback given as a stochastic rank list instead of quantified reward value. We propose an algorithm for this new problem and show its optimality. A real application of this algorithm on clinical treatment is helping paralyzed patient to regain the ability to stand on their own feet.

  • Convex AUC Optimization for Top-N Recommendation with Implicit Feedback

    by Fabio Aiolli
    In this paper, an effective collaborative filtering algorithm for top-N item recommendation with implicit feedback is proposed. The task of top-N item recommendation is to predict a ranking of items (movies, books, songs, or products in general) that can be interesting for a given user based on earlier preferences of the user. In this paper we focus on implicit feedback where these preferences are given in the form of binary events/ratings. Differently from state-of-the-art methods our proposed method is designed to optimize the obtained AUC directly within a margin maximization paradigm. In particular, the solution consists of a simple constrained quadratic optimization problem, one for each user. The experiments performed on several benchmarks show that our method significantly outperforms state-of-the-art matrix factorization methods in terms of AUC of the obtained predictions.

  • Cross-Domain Recommendations Without Overlapping Data: Myth or Reality?

    by Paolo Cremonesi and Massimo Quadrana
    Cross-domain recommender systems adopt different tech- niques to transfer learning from source domain to target domain in order to alleviate the sparsity problem and improve accuracy of recommendations. Traditional techniques require the two domains to be linked by shared characteristics associated to either users or items. In collaborative-filtering (CF) this happens when the two domains have overlapping users or item (at least partially). Recently, Li et al. [7] introduced codebook transfer (CBT), a cross-domain CF technique based on co-clustering, and presented experimental results showing that CBT is able to transfer knowledge between non-overlapping domains. In this paper, we disprove these results and show that CBT does not transfer useful knowledge when source and target domains do not overlap.

  • CSLIM: Contextual SLIM Recommendation Algorithms

    by Yong Zheng, Bamshad Mobasher and Robin Burke
    In addition to user preference profiles, context-aware recommender systems (CARS) take contextual conditions into account when providing item recommendations. In recent years, context-aware matrix factorization (CAMF) has emerged as an extension of the matrix factorization technique that also incorporates contextual conditions. In this paper, we introduce another matrix factorization approach for contextual recommendations, the contextual SLIM (CSLIM) recommendation approach. It is derived from the sparse linear method (SLIM) which was designed for Top-N recommendations in traditional recommender systems. Based on the experimental evaluations over several context-aware data sets, we demonstrate that CLSIM can be an effective approach for context-aware recommendations, in many cases outperforming state-of-the-art CARS algorithms in the Top-N recommendation task.

  • Dynamics of Human Trust in Recommender Systems

    by Jason Harman, John O’Donovan, Tarek Abdelzahe and Cleotilde Gonzalez
    The trust that humans place on recommendations is key to the success of recommender systems. The formation and decay of trust in recommendations is a dynamic process influenced by context, human preferences, accuracy of recommendations, and the interactions of these factors. This paper describes two psychological experiments (N=400) that evaluate the evolution of trust in recommendations over time, under personalized and non-personalized recommendations by matching or not matching a participant’s profile. Main findings include: Humans trust inaccurate recommendations more than they should; when recommendations are personalized, they lose trust in inaccurate recommendations faster than when recommendations are not personalized; and participants learn to select the options that provide best outcomes increasingly over time when they use personalized recommendations, while they are unable to learn if the recommendations are not personalized. We make connections to the possible implications that these psychological findings to the design of recommender systems.

  • Eliciting the Users’ Unknown Preferences

    by Julia Neidhardt, Rainer Schuster, Leonhard Seyfang and Hannes Werthner
    Personalized recommendation strongly relies on an accurate model to capture user preferences; eliciting this information is, in general, a hard problem. In the field of tourism this initial profiling becomes even more challenging. It has been shown that particularly in the beginning of the travel decision making process, users themselves are often not conscious of their needs and are not able to express them. In this paper, the basics of a picture-based approach are introduced that aims at revealing implicitly given user preferences. Based on a set of travel related pictures selected by a user, an individual travel profile is deduced. This is accomplished by mapping those pictures onto seven basic factors that reflect different travel behavioral aspects. Also tourism products can be represented by this simple seven factor model. Thus, this model constitutes the basis of our recommendation algorithm. In addition, users experience this non-verbal way of interaction as exiting and inspiring.

  • Emphasize, Don’t Filter! Displaying Recommendations in Twitter Timelines

    by Wesley Waldner and Julita Vassileva
    This paper describes and evaluates a method for presenting recommendations that will increase the efficiency of the social activity stream while preserving the users’ accurate awareness of the activity within their own social networks. With the help of a content-based recommender system, the application displays the user’s home timeline in Twitter as three visually distinct tiers by emphasizing more strongly those Tweets predicted to be more interesting. Pilot study participants reported that they were able to read the interesting Tweets while ignoring the others with relative ease and that the recommender accurately categorized their Tweets into three tiers.

  • `Free Lunch’ Enhancement for Collaborative Filtering with Factorization Machines

    by Babak Loni, Alan Said, Martha Larson and Alan Hanjalic
    The advantage of Factorization Machines over other factorization models is their ability to easily integrate and efficiently exploit auxiliary information to improve Collaborative Filtering. Until now, this auxiliary information has been drawn from external knowledge sources beyond the user-item matrix. In this paper, we demonstrate that Factorization Machines can exploit additional representations of information inherent in the user-item matrix to improve recommendation performance. We refer to our approach as “Free Lunch” enhancement since it leverages clusters that are based on information that is present in the user-item matrix, but not otherwise directly exploited during matrix factorization. Borrowing clustering concepts from codebook sharing, our approach can also make use of “Free Lunch” information inherent in a user-item matrix from a auxiliary domain that is different from the target domain of the recommender. Our approach improves performance both in the joint case, in which the auxiliary and target domains share users, and in the disjoint case, in which they do not. Although “Free Lunch” enhancement does not apply equally well to any given domain or domain combination, our overall conclusion is that Factorization Machines present an opportunity to exploit information that is ubiquitously present, but commonly under-appreciated by Collaborative Filtering algorithms.

  • Implicit vs. Explicit Trust in Social Matrix Factorization

    by Soude Fazeli, Babak Loni, Alejandro Bellogin, Hendrik Drachsler and Peter Sloep
    Incorporating social trust in Matrix Factorization (MF) methods demonstrably improves accuracy of rating prediction. Such approaches mainly use the trust scores explicitly expressed by users. However, it is often challenging to have users provide explicit trust scores of each other. There exist quite a few works, which propose Trust Metrics (TM) to compute and predict trust scores between users based on their interactions. In this paper, we first evaluate several TMs to find out which one can best predict trust scores compared to the actual trust scores explicitly expressed by users. And, second, we propose to incorporate these trust scores inferred from the candidate implicit TMs into social matrix factorization (MF). We investigate if incorporating the implicit trust scores in MF can make rating prediction as accurate as the MF on explicit trust scores. The reported results support the idea of employing implicit trust into MF whenever explicit trust is not available, since the performance of both models is similar.

  • Inferring User Interests in the Twitter Social Network

    by Parantapa Bhattacharya, Muhammad Zafar , Niloy Ganguly, Saptarshi Ghosh, Krishna Gummadi
    We propose a novel mechanism to infer topics of interest of individual users in the Twitter microblogging site. We observe that in Twitter, a user generally follows experts on various topics of her interest to acquire authentic information on those topics. We use a methodology based on social annotations (proposed earlier by us) to deduce the topical expertise of popular Twitter users, and then transitively infer the interests of the users who are following them. This methodology of inferring user-interests is a sharp departure from the traditional techniques of inferring interests of a user from the tweets that she posts or receives. The topics of interest inferred by the proposed methodology are found to be much superior than the topics extracted by state-of-the-art techniques such as using topic models (Labeled LDA) on tweets. Based upon the proposed methodology, we build a system Who Likes What, which can infer the interests of millions of Twitter users. To our knowledge, this is the first methodology / system that can infer interests for such large number of Twitter users, which would be particularly beneficial in developing personalized recommendation services over the Twitter platform.

  • Modeling the Dynamics of User Preferences in Coupled Tensor Factorization

    by Dimitrios Rafailidis and Alexandros Nanopoulos
    In several applications, user preferences can be fairly dynamic, since users tend to exploit a wide range of items and modify their tastes accordingly over time. In this paper, we model continuous user-item interactions over time using a tensor that has time as a dimension (mode). To account for the fact that user preferences are dynamic and change individually, we propose a new measure of user-preference dynamics (UPD) that captures the rate with which the current preferences of each user have been shifted. We generate recommendations based on factorizing the tensor, by weighting the importance of past user preferences according to their UPD values. We additionally exploit users’ side data, such as demographics, which can help improving the accuracy of recommendations based on a coupled, tensor-matrix factorization scheme. Our empirical evaluation uses a real data set from last.fm, which allows us to demonstrate that user preferences can become very dynamic. Our experimental results show that the proposed method, by taking into account these dynamics, outperforms several baselines.

  • Multi-Criteria Journey Aware Housing Recommender System

    by Elizabeth Daly, Adi Botea, Akihiro Kishimoto and Radu Marinescu
    Recommender systems can be employed to assist users in complex decision making processes. This paper presents a multi-criteria housing recommender system which takes into account not just features of a home, such as rent, but also the transportation links to user specified locations. First, we describe an efficient multi-hop journey time calculator. Second, we introduce a mechanism to find the optimal solutions for multi-criteria evaluation, where a balanced trade-off between the target goals is found. Finally, we present a user study to demonstrate the potential of such a system.

  • PERSPeCT: Collaborative Filtering for Tailored Health Communications

    by Roy Adams, Rajani Sadasivam, Kavitha Balakrishnan, Rebecca Kinney, Thomas Houston and Benjamin Marlin
    The goal of computer tailored health communications (CTHC) is to elicit healthy behavior changes by sending motivational messages personalized to individual patients. One prominent weakness of many existing CTHC systems is that they are based on expert-written rules and thus have no ability to learn from their users over time. One solution to this problem is to develop CTHC systems based on the principles of collaborative filtering, but this approach has not been widely studied. In this paper, we present a case study evaluating nine rating prediction methods for use in the Patient Experience Recommender System for Persuasive Communication Tailoring, a system developed for use in a clinical trial of CTHC-based smoking cessation support interventions.

  • Preference Elicitation for Narrowing the Recommended List for Groups

    by Lihi Naamani-Dery, Meir Kalech, Lior Rokach and Bracha Shapira
    A group may appreciate recommendations on items that fit their joint preferences. When the members’ actual preferences are unknown, a recommendation can be made with the aid of collaborative filtering methods. We offer to narrow down the recommended list of items by eliciting the users’ actual preferences. Our final goal is to output top-k preferred items to the group out of the top-N recommendations provided by the recommender system (k < N), where one of the items is a necessary winner. We propose an iterative preference elicitation method, where users are required to provide item ratings per request. We suggest a heuristic that attempts to minimize the preference elicitation effort under two aggregation strategies. We evaluate our methods on real-world Netflix data as well as on simulated data which allows us to study different cases. We show that preference elicitation effort can be cut in up to 90% while preserving the most preferred items in the narrowed list.

  • Recommendation-based Modeling Support for Data Mining Processes

    by Dietmar Jannach and Simon Fischer
    RapidMiner is a software tool that allows users to define data mining processes based on a visual model and implements a variety of so-called operators for data extraction, manipulation, model learning and analysis. The large number of available operators can however make it challenging for the process designer to identify the appropriate operator for the problem at hand. At the same time, some operators are only meaningful when combined with certain others. In this work, we evaluate different strategies of recommending additional operators to the user during the design of the process. The recommendation models are learned using a pool of several thousand existing data mining processes and evaluated in an offline experimental design. The results indicate that good predictive accuracy can already be achieved with comparably simple co-occurrence based algorithms.

  • Scalable Audience Targeted Models for Brand Advertising on Social Networks

    by Kunpeng Zhang
    People are using social media to generate, share, and communicate information with each other. Finding actionable insights from such big data has attracted a lot of research attentions on, for example, finding targeted user groups based on their historical on-line activities. However, existing machine learning algorithms fail to keep up with the increasing large data volume. In this paper, we develop a scalable regression-based algorithm called distributed iterative shrinkage-thresholding algorithm (DISTA) that can identify potential users. Our experiments conducted on Facebook data containing billions of users and associated activities show that DISTA with feature selection not only enables on-line audience-targeted approach for precise marketing but also performs efficiently on parallel computers.

  • Social Collaborative Filtering for Cold-start Recommendations

    by Suvash Sedhain, Scott Sanner, Darius Braziunas, Lexing Xie and Jordan Christensen
    We examine the cold-start recommendation task in an online retail setting for users who have not yet purchased (or interacted in a meaningful way with) any available items but who have granted access to limited side information, such as basic demographic data (gender, age, location) or social network information (Facebook friends or page likes). We formalize neighborhood-based methods for cold-start collaborative filtering in a generalized matrix algebra framework that does not require purchase data for target users when their side information is available. In real-data experiments with 30,000 users who purchased 80,000+ books and had 9,000,000+ Facebook friends and 6,000,000+ page likes, we show that using Facebook page likes for cold-start recommendation yields up to a 3-fold improvement in mean average precision (mAP) and up to 6-fold improvements in Precision@k and Recall@k compared to most-popular-item, demographic, and Facebook friend cold-start recommenders. These results demonstrate the substantial predictive power of social network content, and its significant utility in a challenging problem – recommendation for cold-start users.

  • Switching Hybrid for Cold-Starting Context-Aware Recommender Systems

    by Matthias Braunhofer, Victor Codina and Francesco Ricci
    Finding effective solutions for cold-starting context-aware recommender systems (CARSs) is important because usually low quality recommendations are produced for users, items or contextual situations that are new to the system. In this paper, we tackle this problem with a switching hybrid solution that exploits a custom selection of two CARS algorithms, each one suited for a particular cold start situation, and switches between these algorithms depending on the detected recommendation situation (new user, new item or new context). We evaluate the proposed algorithms in an offline experiment by using various contextually-tagged rating datasets. We illustrate some significant performance differences among the considered algorithms and show that they can be effectively combined into the proposed switching hybrid to cope with different types of cold-start problems.

  • Using Graded Implicit Feedback for Bayesian Personalized Ranking

    by Lukas Lerche and Dietmar Jannach
    In many application domains of recommender systems, explicit rating information is sparse or non-existent. The preferences of the current user have therefore to be approximated by interpreting his or her behavior, i.e., the implicit user feedback. In the literature, a number of algorithm proposals have been made that rely solely on such implicit feedback, among them Bayesian Personalized Ranking (BPR). In the BPR approach, pairwise comparisons between the items are made in the training phase and an item i is considered to be preferred over item j if the user interacted in some form with i but not with j. In real-world applications, however, implicit feedback is not necessarily limited to such binary decisions as there are, e.g., different types of user actions like item views, cart or purchase actions and there can exist several actions for an item over time. In this paper we show how BPR can be extended to deal with such more fine-granular, graded preference relations. An empirical analysis shows that this extension can help to measurably increase the predictive accuracy of BPR on realistic e-commerce datasets.

Back to Program

Diamond Supporters
 
Platinum Supporters
 
 
 
 
Gold Supporters
 
 
Silver Supporter