| 文献作者 | Behnoush Abdollahi | ||||||||||
| 文献发表年限 | 2017 | 创建时间 | 2018-06-12 | ||||||||
| 文献关键字 | factorization model; explainability; 根据近邻进行解释 | ||||||||||
| 摘要描述 | Accurate model-based Collaborative Filtering (CF) approaches, such as Matrix Factorization (MF), tend to be black-box machine learning models that lack interpretability and do not provide a straightforward explanation for their outputs. Yet explanations have been shown to improve the transparency of a recommender system by justifying recommendations, and this in turn can enhance the user’s trust in the recommendations. Hence, one main challenge in designing a recommender system is mitigating the trade-off between an explainable technique with moderate prediction accuracy and a more accurate technique with no explainable recommendations. In this paper, we focus on factorization models and further assume the absence of any additional data source, such as item content or user attributes. We propose an explainability constrained MF technique that computes the top-n recommendation list from items that are explainable. Experimental results show that our method is effective in generating accurate and explainable recommendations. | ||||||||||
| 文献作者 | Raoua Abdelkhalek | ||||||||||
| 文献发表年限 | 2016 | 创建时间 | 2018-06-09 | ||||||||
| 文献关键字 | Recommender systems · Collaborative filtering · Belief function theory · Uncertainty · Evidential K-Nearest Neighbors；Dempster–Shafer theory; DS证据理论； 证据合成规则 | ||||||||||
| 摘要描述 | Recommender Systems (RSs) in particular the collaborative filtering approaches have reached a high level of popularity. These approaches are designed for predicting the user’s future interests towards unrated items. However, the provided predictions should be taken with restraint because of the uncertainty pervading the real-world problems. Indeed, to not give consideration to such uncertainty may lead to unrep- resentative results which can deeply affect the predictions’ accuracy as well as the user’s confidence towards the RS. In order to tackle this issue, we propose in this paper a new evidential item-based collaborative fil- tering approach. In our approach, we involve the belief function theory tools as well as the Evidential K-Nearest Neighbors (EKNN) classifier to deal with the uncertain aspect of items’ recommendation ignored by the classical methods. The performance of our new recommendation app- roach is proved through a comparative evaluation with several traditional collaborative filtering recommenders. | ||||||||||
| 文献作者 | Raoua Abdelkhalek | ||||||||||
| 文献发表年限 | 2017 | 创建时间 | 2018-06-09 | ||||||||
| 文献关键字 | Dempster–Shafer theory; DS证据理论； 证据合成规则 | ||||||||||
| 摘要描述 | Collaborative Filtering (CF) consists of filtering data, predicting users’ preferences and providing recommendations accordingly. Commonly, neighborhood-based CF methods predict the future ratings based on similar users (user-based) or similar items (item- based) to perform recommendations. However, the reliability of the information provided by these pieces of evidence as well as the final predictions cannot be fully trusted. Incorporating trust in the recommendation process can be argued to be an important challenge in Recommender Systems (RSs). To tackle these issues, we propose new CF approaches under the belief function framework. The final prediction is obtained by fusing evidences from similar items or similar users using Dempster’s rule of combination. The prediction process of our evidential approaches is able to provide the users with a global overview of their possible preferences. This would lead to increase their confidence towards the system as well as their satisfaction. In this paper, we mainly highlight the benefits of incorporating uncertainty in CF approaches using the belief function theory. We present the preliminary results and also discuss our ongoing works, as well as the challenges in the future. | ||||||||||
| 文献作者 | Anupam Datta, Sophia Kovaleva, Piotr Mardziel, Shayak Sen | ||||||||||
| 文献发表年限 | 2018 | 创建时间 | 2018-06-04 | ||||||||
| 文献关键字 | 隐因子解释; interpretation; shadow model; Quantitative input influence; QII; Interpreting； 仿真实验; simulation; mapping | ||||||||||
| 摘要描述 | Many machine learning systems utilize latent factors as internal representations for making predictions. Since these latent factors are largely uninterpreted, however, predictions made using them are opaque. Collaborative filtering via matrix factorization is a prime example of such an algorithm that uses uninterpreted latent features, and yet has seen widespread adoption for many recommendation tasks. We present Latent Factor Interpretation (LFI), a method for interpreting models by leveraging interpretations of latent factors in terms of human- understandable features. The interpretation of latent factors can then replace the uninterpreted latent factors, resulting in a new model that expresses predictions in terms of interpretable features. This new model can then be interpreted using recently developed model explanation techniques. In this paper, we develop LFI for collaborative filtering based recommender systems. We illustrate the use of LFI interpretations on the MovieLens dataset, integrating auxiliary features from IMDB and DB tropes, and show that latent factors can be predicted with sufficient accuracy for replicating the predictions of the true model. | ||||||||||
| 文献作者 | Armelle Brun; Marharyta Aleksandrova | ||||||||||
| 文献发表年限 | 2014 | 创建时间 | 2018-06-02 | ||||||||
| 文献关键字 | representative users; interpretable; cold-start problem; | ||||||||||
| 摘要描述 | Matrix factorization has proven to be one of the most accurate recommendation approach. However, it faces one main shortcoming: the latent features that result from factoriza- tion, and that represent the underlying relation between users and items, are not directly interpretable. Some works focused on their interpretation, particularly with non-negative matrix factorization. In these works, features are viewed as groups of users, groups of items or as attributes of items, but such interpretations require human expertise. In this paper, we propose to interpret features as users, referred to as representative users. This interpretation relies on the study of the matrices that result from the factorization and on their link with the original rating matrix. Such an interpretation is not only performed automatically, as it does not require any human expertise, but it helps also to explain the recommendations made to users. In addition, we see it as a way to alleviate the new item cold-start problem, without requiring any information about the content of the items. The experiments conducted on several benchmark datasets confirm that the features discovered by a non-negative matrix factorization can be actually interpreted as users and that the representative users (the interpretations of the features), are a reliable source of information that allows to accurately estimate ratings on new items. They are thus a promising way to solve the new item cold-start problem. | ||||||||||
| 文献作者 | Marharyta Aleksandrova | ||||||||||
| 文献发表年限 | 2014 | 创建时间 | 2018-06-02 | ||||||||
| 文献关键字 | Recommender systems, matrix factorization, features interpretation. | ||||||||||
| 摘要描述 | Matrix factorization (MF) is a powerful approach used in recommender systems. One main drawback of MF is the difficulty to interpret the automatically formed features. Following the intuition that the relation between users and items can be expressed through a reduced set of users, referred to as representative users, we propose a simple modification of a traditional MF algorithm, that forms a set of features corresponding to these representative users. On one state of the art dataset, we show that the proposed representative users-based non-negative matrix factorization (RU-NMF) discovers interpretable features, while slightly (in some cases insignificantly) decreasing the accuracy. | ||||||||||
| 文献作者 | Dimitrios Rafailidis | ||||||||||
| 文献发表年限 | 2017 | 创建时间 | 2018-05-26 | ||||||||
| 文献关键字 | Learning to rank; social relationships; collaborative filtering; trust-based | ||||||||||
| 摘要描述 | The sparsity of users’ preferences can significantly degrade the quality of recommendations in the collaborative filtering strategy. To account for the fact that the selections of social friends and foes may improve the recommendation accuracy, we propose a learning to rank model that exploits users’ trust and distrust relationships. Our learning to rank model focusses on the performance at the top of the list, with the recommended items that end- users will actually see. In our model, we try to push the relevant items of users and their friends at the top of the list, while ranking low those of their foes. Furthermore, we propose a weighting strategy to capture the correlations of users’ preferences with friends’ trust and foes’ distrust degrees in two intermediate trust- and distrust-preference user latent spaces, respectively. Our experiments on the Epinions dataset show that the proposed learning to rank model significantly outperforms other state-of-the-art meth- ods in the presence of sparsity in users’ preferences and when a part of trust and distrust relationships is not available. Furthermore, we demonstrate the crucial role of our weighting strategy in our model, to balance well the influences of friends and foes on users’ preferences. | ||||||||||
| 文献作者 | Anaı̈s Ollagnier, Sébastien Fournier, and Patrice Bellot | ||||||||||
| 文献发表年限 | 2018 | 创建时间 | 2018-05-25 | ||||||||
| 文献关键字 | 【随意看】Recommender systems, Text mining, Digital libraries, Bib- liographic information, Bibliometrics. | ||||||||||
| 摘要描述 | The iterative continuum of scientific production generates a need for filtering and specific crossing of ideas and papers. In this paper, we present BIBLME RecSys software which is dedicated to the analysis of bibliographical references extracted from scientific collections of papers. Our goal is to provide users with paper suggestions guided by the papers they are reading and by the references they contain. To do so, we propose a new approach based on a new bibliometric measure. We propose to determine the impact, the inner representativeness, of each bibliographical reference according to their occurrences in the paper the user is reading. By means of this approach, we suggest central references of the author’s paper. As a result, we obtain papers that are related to the paper selected by the user according to the influence of references on it. We evaluate the recommendation in the context of a digital library dedicated to humanities and social sciences. | ||||||||||
| 文献作者 | Toine Bogers | ||||||||||
| 文献发表年限 | 2014 | 创建时间 | 2018-05-25 | ||||||||
| 文献关键字 | 【随意看】content-based | ||||||||||
| 摘要描述 | While content-based recommendation has been applied successfully in many different domains, it has not seen the same level of attention as collaborative filtering techniques have. However, there are many recommendation domains and applications where content and metadata play a key role, either in addition to or instead of ratings and implicit usage data. For some domains, such as movies, the relationship between content and usage data has seen thorough investigation already, but for many other domains, such as books, news, scientific articles, and Web pages we still do not know if and how these data sources should be combined to provide the best recommendation performance. The CBRecSys 2014 workshop aimed to address this by providing a dedicated venue for papers dedicated to all aspects of content-based recommender systems. | ||||||||||
| 文献作者 | |||||||||||
| 文献发表年限 | 2010 | 创建时间 | 2018-04-28 | ||||||||
| 文献关键字 | 笛卡尔积特征组合；特征哈希；Feature Hashing；Hash trick | ||||||||||
| 摘要描述 | 1）笛卡尔积特征组合 2）特征哈希 | ||||||||||
