Shuaiqiang Wang

A survey of serendipity in recommender systems

We summarize most efforts on serendipity in recommender systems.We compare definitions of serendipity in recommender systems.We classify the state-of-the-art serendipity-oriented recommendation algorithms.We review methods to assess serendipity in recommender systems.We provide the future directions of serendipity in recommender systems. Recommender systems use past behaviors of users to suggest items. Most tend to offer items similar to the items that a target user has indicated as interesting. As a result, users become bored with obvious suggestions that they might have already discovered. To improve user satisfaction, recommender systems should offer serendipitous suggestions: items not only relevant and novel to the target user, but also significantly different from the items that the user has rated. However, the concept of serendipity is very subjective and serendipitous encounters are very rare in real-world scenarios, which makes serendipitous recommendations extremely difficult to study. To date, various definitions and evaluation metrics to measure serendipity have been proposed, and there is no wide consensus on which definition and evaluation metric to use. In this paper, we summarize most important approaches to serendipity in recommender systems, compare different definitions and formalizations of the concept, discuss serendipity-oriented recommendation algorithms and evaluation strategies to assess the algorithms, and provide future research directions based on the reviewed literature.

Singular value decomposition based minutiae matching method for finger vein recognition

Recently, finger vein recognition has received considerable attention in the biometric recognition field. Originating from fingerprint recognition, minutiae-based methods are recognized as an important branch, which attempts to discover minutia patterns from finger vein images for matching and recognition. However, the accuracy of these methods is generally unsatisfactory. One of the most challenging problems is that, the correspondences of two minutia sets are difficult to obtain resulting from the rotation, translation and deformation of the finger vein images. Another critical problem is that, the current available feature descriptors for minutia representation are weak and insufficient. In this paper, we propose SVDMM, a singular value decomposition (SVD)-based minutiae matching method for finger vein recognition, which involves three stages: (I) minutia pairing, (II) false removing and (III) score calculating. In particular, stage I discovers minutia pairs via SVD-based decomposition of the correlation-weighted proximity matrix. Stage II removes false pairs based on the local extensive binary pattern (LEBP) for increasing the reliability of the correspondences. Stage III determines the matching score of the input and template images by the ‘average’ matching degree of all their precise minutia pairs. Extensive experiments demonstrate that our work not only performs better than the similar works in the literature, but also has great potential to achieve comparable performance to other categories of state-of-the-art methods.

VSRank: A Novel Framework for Ranking-Based Collaborative Filtering

Collaborative filtering (CF) is an effective technique addressing the information overload problem. CF approaches generally fall into two categories: rating based and ranking based. The former makes recommendations based on historical rating scores of items and the latter based on their rankings. Ranking-based CF has demonstrated advantages in recommendation accuracy, being able to capture the preference similarity between users even if their rating scores differ significantly. In this study, we propose VSRank, a novel framework that seeks accuracy improvement of ranking-based CF through adaptation of the vector space model. In VSRank, we consider each user as a document and his or her pairwise relative preferences as terms. We then use a novel degree-specialty weighting scheme resembling TF-IDF to weight the terms. Extensive experiments on benchmarks in comparison with the state-of-the-art approaches demonstrate the promise of our approach.

Summarizing web forum threads based on a latent topic propagation process

With an increasingly amount of information in web forums, quick comprehension of threads in web forums has become a challenging research problem. To handle this issue, this paper investigates the task of Web Forum Thread Summarization (WFTS), aiming to give a brief statement of each thread that involving multiple dynamic topics. When applied to the task of WFTS, traditional summarization methods are cramped by topic dependencies, topic drifting and text sparseness. Consequently, we explore an unsupervised topic propagation model in this paper, the Post Propagation Model (PPM), to burst through these problems by simultaneously modeling the semantics and the reply relationship existing in each thread. Each post in PPM is considered as a mixture of topics, and a product of Dirichlet distributions in previous posts is employed to model each topic dependencies during the asynchronous discussion. Based on this model, the task of WFTS is accomplished by extracting most significant sentences in a thread. The experimental results on two different forum data sets show that WFTS based on the PPM outperforms several state-of-the-art summarization methods in terms of ROUGE metrics.

Adapting vector space model to ranking-based collaborative filtering

Collaborative filtering (CF) is an effective technique addressing the information overload problem. Recently ranking-based CF methods have shown advantages in recommendation accuracy, being able to capture the preference similarity between users even if their rating scores differ significantly. In this study, we seek accuracy improvement of ranking-based CF through adaptation of the vector space model, where we consider each user as a document and her pairwise relative preferences as terms. We then use a novel degree-specialty weighting scheme resembling TF-IDF to weight the terms. Then we use cosine similarity to select a neighborhood of users for the target user to make recommendations. Experiments on benchmarks in comparison with the state-of-the-art methods demonstrate the promise of our approach.

A Hybrid Multigroup Coclustering Recommendation Framework Based on Information Fusion

Collaborative Filtering (CF) is one of the most successful algorithms in recommender systems. However, it suffers from data sparsity and scalability problems. Although many clustering techniques have been incorporated to alleviate these two problems, most of them fail to achieve further significant improvement in recommendation accuracy. First of all, most of them assume each user or item belongs to a single cluster. Since usually users can hold multiple interests and items may belong to multiple categories, it is more reasonable to assume that users and items can join multiple clusters (groups), where each cluster is a subset of like-minded users and items they prefer. Furthermore, most of the clustering-based CF models only utilize historical rating information in the clustering procedure but ignore other data resources in recommender systems such as the social connections of users and the correlations between items. In this article, we propose HMCoC, a Hybrid Multigroup CoClustering recommendation framework, which can cluster users and items into multiple groups simultaneously with different information resources. In our framework, we first integrate information of user--item rating records, user social networks, and item features extracted from the DBpedia knowledge base. We then use an optimization method to mine meaningful user--item groups with all the information. Finally, we apply the conventional CF method in each cluster to make predictions. By merging the predictions from each cluster, we generate the top-n recommendations to the target users for return. Extensive experimental results demonstrate the superior performance of our approach in top-n recommendation in terms of MAP, NDCG, and F1 compared with other clustering-based CF models.

Learning to rank for hybrid recommendation

Most existing recommender systems can be classified into two categories: collaborative filtering and content-based filtering. Hybrid recommender systems combine the advantages of the two for improved recommendation performance. Traditional recommender systems are rating-based. However, predicting ratings is an intermediate step towards their ultimate goal of generating rankings or recommendation lists. Learning to rank is an established means of predicting rankings and has recently demonstrated high promise in improving quality of recommendations. In this paper, we propose LRHR, the first attempt that adapts learning to rank to hybrid recommender systems. LRHR first defines novel representations for both users and items so that they can be content-comparable. Then, LRHR identifies a set of novel meta-level features for learning purposes. Finally, LRHR adopts RankSVM, a pairwise learning to rank algorithm, to generate recommendation lists of items for users. Extensive experiments on benchmarks in comparison with the state-of-the-art algorithms demonstrate the performance gain of our approach.

Challenges of Serendipity in Recommender Systems

Most recommender systems suggest items similar to a user profile, which results in boring recommendations limited by user preferences indicated in the system. To overcome this problem, recommender systems should suggest serendipitous items, which is a challenging task, as it is unclear what makes items serendipitous to a user and how to measure serendipity. The concept is difficult to investigate, as serendipity includes an emotional dimension and serendipitous encounters are very rare. In this paper, we discuss mentioned challenges, review definitions of serendipity and serendipity-oriented evaluation metrics. The goal of the paper is to guide and inspire future efforts on serendipity in recommender systems.

Distributed collaborative filtering with singular ratings for large scale recommendation

Collaborative filtering (CF) is an effective technique addressing the information overloading problem, where each user is associated with a set of rating scores on a set of items. For a chosen target user, conventional CF algorithms measure similarity between this user and other users by utilizing pairs of rating scores on common rated items, but discarding scores rated by one of them only. We call these comparative scores as dual ratings, while the non-comparative scores as singular ratings. Our experiments show that only about 10% ratings are dual ones that can be used for similarity evaluation, while the other 90% are singular ones. In this paper, we propose SingCF approach, which attempts to incorporate multiple singular ratings, in addition to dual ratings, to implement collaborative filtering, aiming at improving the recommendation accuracy. We first estimate the unrated scores for singular ratings and transform them into dual ones. Then we perform a CF process to discover neighborhood users and make predictions for each target user. Furthermore, we provide a MapReduce-based distributed framework on Hadoop for significant improvement in efficiency. Experiments in comparison with the state-of-the-art methods demonstrate the performance gains of our approaches.

Learning to rank using evolutionary computation: immune programming or genetic programming?

Nowadays ranking function discovery approaches using Evolutionary Computation (EC), especially Genetic Programming (GP), have become an important branch in the Learning to Rank for Information Retrieval (LR4IR) field. Inspired by the GP based learning to rank approaches, we provide a series of generalized definitions and a common framework for the application of EC in learning to rank research. Besides, according to the introduced framework, we propose RankIP, a ranking function discovery approach using Immune Programming (IP). Experimental results demonstrate that RankIP evidently outperforms the baselines. In addition, we study the differences between IP and GP in theory and experiments. Results show that IP is more suitable for LR4IR due to its high diversity.