Cross-Network Community Sensing for Anchor Link Prediction

Nowadays, people are usually involved in multiple heterogeneous social networks simultaneously. Discovering the anchor links between the accounts owned by the same users across different social networks is crucial for many important inter-network applications, e.g., cross-network link transfer and cross-network recommendation. Many different supervised models have been proposed to predict anchor links so far, but they are effective only when the labeled anchor links are abundant. However, in real scenarios, such a requirement can hardly be met and most anchor links are unlabeled, since manually labeling the inter-network anchor links is quite costly and tedious. To overcome such a problem and utilize the numerous unlabeled anchor links in model building, in this paper, we introduce the active learning based anchor link prediction problem. Different from the traditional active learning problems, due to the one-to-one constraint on anchor links, if an unlabeled anchor link is identified as positive (i.e., existing), all the other unlabeled anchor links incident to account u or account v will be negative (i.e., non-existing) automatically. Viewed in such a perspective, asking for the labels of potential positive anchor links in the unlabeled set will be rewarding in the active anchor link prediction problem. Various novel anchor link information gain measures are defined in this paper, based on which several constraint active anchor link prediction methods are introduced. Extensive experiments have been done on real-world social network datasets to compare the performance of these methods with state-of-art anchor link prediction methods. The experimental results show that the proposed Mean-entropy-based Constrained Active Learning (MC) method can outperform other methods with significant advantages.

Anchor link prediction exacerbates the risk of privacy leakage via the de-anonymization of social network data. Embedding-based methods for anchor link prediction are limited by the excessive similarity of the associated nodes in a latent feature space and the variation between latent feature spaces caused by the semantics of different networks. In this article, we propose a novel method which reduces the impact of semantic discrepancies between different networks in the latent feature space. The proposed method consists of two phases. First, graph embedding focuses on the network structural roles of nodes and increases the distinction between the associated nodes in the embedding space. Second, a federated adversarial learning framework which performs graph embedding on each social network and an adversarial learning model on the server according to the observable anchor links is used to associate independent graph embedding approaches on different social networks. The combination of distinction enhancement and the association of graph embedding approaches alleviates variance between the latent feature spaces caused by the semantics of different social networks. Extensive experiments on real social networks demonstrate that the proposed method significantly outperforms the state-of-the-art methods in terms of both precision and robustness.

Nowadays online users prefer to join multiple social media for the purpose of socialized online service. The problem \textit{anchor link prediction} is formalized to link user data with the common ground on user profile, content and network structure across social networks. Most of the traditional works concentrated on learning matching function with explicit or implicit features on observed user data. However, the low quality of observed user data confuses the judgment on anchor links, resulting in the matching collision problem in practice. In this paper, we explore local structure consistency and then construct a matching graph in order to circumvent matching collisions. Furthermore, we propose graph convolution networks with mini-batch strategy, efficiently solving anchor link prediction on matching graph. The experimental results on three real application scenarios show the great potentials of our proposed method in both prediction accuracy and efficiency. In addition, the visualization of learned embeddings provides us a qualitative way to understand the inference of anchor links on the matching graph.

Welcome to the 24th ACM International Conference on User Modeling, Adaptation, and Personalization (UMAP 2016) in Halifax, Canada, July 13-16, 2016. UMAP is the premier international conference for researchers and practitioners working on systems that adapt to individual users or to groups of users. UMAP is the successor of the biennial User Modeling (UM) and Adaptive Hypermedia and Adaptive Web-based Systems (AH) conferences that were merged in 2009. It has traditionally been organized under the auspices of User Modeling Inc. This year (2016) UMAP became an ACM conference, sponsored by ACM SIG CHI and SIG WEB. The conference spans a wide scope of topics related to user modeling, adaptation, and personalization. UMAP 2016 is focused on bringing together cutting-edge research from user interaction and modeling, adaptive technologies, and delivery platforms. It includes high-quality peer-reviewed papers featuring substantive new research in one of five research areas, each chaired by leaders in the field: User Modeling for Recommender Systems (chairs: Alexander Felfernig & Pasquale Lops)Adaptive & Personalized Educational Systems (chairs: Antonija Mitrovic & Kalina Yacef)Personalization in the Social Web & Crowdsourcing Era (chairs: Alessandro Bozzon & Harith Alani)Adaptive, Intelligent, & Multimodal User Interfaces (chairs: Julien Epps & Hatice Gunes)Architectures, Techniques, & Methodologies for UMAP (chairs: Stephan Weibelzahl & Mihaela Cocea) This year we received 123 submissions. In keeping with UMAPs rigorous standards, each paper was carefully reviewed by members of the Program Committee (PC) while the Area Chairs (ACs) coordinated the reviews and provided recommendations to the Program Chairs. The international Program Committee (PC) consisted of 132 members who were assisted by 49 subreviewers. These were leading researchers as well as highly promising young researchers. Papers were assigned to at least 4 members of the PC and to 1 AC member based on their expertise, interests, and other factors. Each paper received at least 3 reviews, and 95% received 4 reviews. After the initial reviews were submitted, the designated AC facilitated discussion amongst reviewers in order to resolve differences and correct misunderstandings. The AC then provided a summative meta-review and a recommendation to the Program Chairs. The final decisions were based on these recommendations, the meta-reviews, and reviewer scores. We accepted 21 long papers (23.9% acceptance rate) and 13 short papers (27.6% acceptance rate) for oral presentation and an additional 17 extended abstracts for poster presentation and inclusion in the proceedings. The program also features posters, demos, and late breaking results, which collectively showcase the wide spectrum of novel ideas and latest results in user modeling, adaptation and personalization. We also invited three distinguished keynote speakers, each illustrating significant issues and prospective directions for the field. Hossein Derakhshan (shared keynote speaker with Hypertext 2016) is an Iranian-Canadian blogger who was imprisoned in Tehran from November 2008 to November 2014. He has been called the "father of Persian blogging" and has helped promote podcasting in Iran. His talk, "Killing the Hyperlink, Killing the Web: the Shift from Library-Internet to Television-Internet," reflects his views on the Internet today.Lada Adamic leads the Product Science group within Facebook's Data Science Team. She is also an adjunct associate professor at the University of Michigan's School of Information and Center for the Study of Complex Systems. Her talk "The Life and Times of Information in Networks" focuses on cascades of information-sharing and resharing within social media.Sandra Carberry was one of the founders of the User Modeling research area at the first workshop in Maria Laach in 1986. As appropriate for the 30th anniversary, her talk "User Modeling: The Past, The Present and The Future" discusses how the field evolved, insights into where the field is headed, and the hottest topics for exploration. The conference includes a doctoral consortium that provides an opportunity for doctoral students to explore and develop their research interests under the guidance of distinguished scholars. This track received 18 submissions of which nine were accepted as full papers and eight as posters. A set of seven workshops and two tutorials round out the program. (Workshop) IFUP: Workshop on Multi-dimension Information Fusion for Modeling and Personalisation (half-day) organized by Robin Burke (DePaul University, USA), Feida Zhu (Singapore Management University, Singapore), Neil Yorke-Smith (American University of Beirut, Lebanon), and Guibing Guo (Northeastern University, China)(Workshop) INRA: News Recommendation and Analytics (half-day) organized by Jon Atle Gulla (Norwegian University of Science and Technology Trondheim, Norway), Luc Martens (Minds-UGent-WiCa Ghent, Belgium), Özlem Özgöbek (Norwegian University of Science and Technology Trondheim, Norway), and Nafiseh Shabib (TNS Gallup, Oslo, Norway)(Workshop) SOAP: Workshop on Surprise, Opposition, and Obstruction in Adaptive and Personalized Systems (half-day) organized by Peter Knees (Johannes Kepler University Linz, Austria), Kristina Andersen (Studio for Electro Instrumental Music, Amsterdam, the Netherlands), Alan Said (Recorded Future, Gothenburg, Sweden), and Marko Tkalcic (Free University of Bozen-Bolzano, Italy)(Workshop) HAAPIE: Human Aspects in Adaptive and Personalised Interactive Environments (half-day) organized by Panagiotis Germanakos (SAP SE, Germany), Marios Belk (Department of Computer Science, University of Cyprus), George Samaras (Department of Computer Science, University of Cyprus), and Vania Dimitrova (University of Leeds, UK)(Workshop) EvalUMAP: Towards comparative evaluation in the user modelling, adaptation and personalization space (full-day) organized by Owen Conlan (Trinity College Dublin, Ireland), Liadh Kelly (Trinity College Dublin, Ireland), Kevin Koidl (Trinity College Dublin, Ireland), Séamus Lawless (Trinity College Dublin, Ireland), Killian Levacher (Trinity College Dublin, Ireland), and Athanasios Staikopoulos (Trinity College Dublin, Ireland)(Workshop) FuturePD: The future of personal data: envisioning new personalized services enabled by Quantified Self technologies (half-day) organized by Amon Rapp (University of Torino, Italy), Federica Cena (University of Torino, Italy), Judy Kay (University of Sidney, Australia), Bob Kummerfeld (University of Sydney, Australia), Frank Hopfgartner (University Gardens Glasgow, UK), Jakob Eg Larsen (Technical University of Denmark, Denmark), and Elise van den Hoven (University of Technology Sydney, Australia). (Workshop) PALE: Personalization Approaches in Learning Environments (full-day) organized by Milos Kravcik (RWTH Aachen University, Germany), Olga C. Santos (UNED, Spain), Jesus G. Boticario (UNED, Spain), and Maria Bielikova (FIIT STUBA, Slovakia)(Tutorial) Semantics-Aware Techniques for Social Media Analysis, User Modeling, and Recommender Systems (half-day) by Pasquale Lops and Cataldo Musto (University of Bari Aldo Moro, Italy)(Tutorial) Games, Gamification and Personalization (half-day) by Amon Rapp (University of Torino, Italy).

The online social network alignment problem aims at inferring the anchor links connecting the shared users across social networks, which are usually subject to the one-to-one cardinality constraint. Several existing social network alignment models have been proposed, many of which are based on the supervised learning setting. Given a set of labeled anchor links, a group of features can be extracted manually for the anchor links to build these models. Meanwhile, such methods may encounter great challenges in the application on real-world social network datasets, since manual feature extraction can be extremely expensive and tedious for the social networks involving heterogeneous information. In this paper, we propose to address the heterogeneous social network alignment problem with a deep learning model, namely DETA (Deep nETwork Alignment). Besides a small number of explicit features, DETA can automatically learn a set of latent features from the heterogeneous information. DETA models the anchor link one-to-one cardinality constraint as a mathematical constraint on the node degrees. Extensive experiments have been done on real-world aligned heterogeneous social network datasets, and the experimental results have demonstrated the effectiveness of the proposed model compared against the existing state-of-the-art baseline methods.

Dynamic link prediction is essential for a wide range of domains, including social networks, bioinformatics, knowledge bases, and recommender systems. Existing works have demonstrated that structural information and temporal information are two of the most important information for this problem. However, existing works either focus on modeling them independently or modeling the temporal dynamics of a single structural scale, neglecting the complex correlations among them. This paper proposes to model the inherent correlations among the evolving dynamics of different structural scales for dynamic link prediction. Following this idea, we propose an Attentional Multi-scale Co-evolving Network (AMCNet). Specifically, We model multi-scale structural information by a motif-based graph neural network with multi-scale pooling. Then, we design a hierarchical attention-based sequence-to-sequence model for learning the complex correlations among the evolution dynamics of different structural scales. Extensive experiments on four real-world datasets with different characteristics demonstrate that AMCNet significantly outperforms the state-of-the-art in both single-step and multi-step dynamic link prediction tasks.

Mind maps have not received much attention in the user modeling and recommender system community, although mind maps contain rich information that could be valuable for user-modeling and recommender systems. In this paper, we explored the effectiveness of standard user-modeling approaches applied to mind maps. Additionally, we develop novel user modeling approaches that consider the unique characteristics of mind maps. The approaches are applied and evaluated using our mind mapping and reference-management software Docear. Docear displayed 430,893 research paper recommendations, based on 4,700 user mind maps, from March 2013 to August 2014. The evaluation shows that standard user modeling approaches are reasonably effective when applied to mind maps, with click-through rates (CTR) between 1.16% and 3.92%. However, when adjusting user modeling to the unique characteristics of mind maps, a higher CTR of 7.20% could be achieved. A user study confirmed the high effectiveness of the mind map specific approach with an average rating of 3.23 (out of 5), compared to a rating of 2.53 for the best baseline. Our research shows that mind map-specific user modeling has a high potential, and we hope that our results initiate a discussion that encourages researchers to pursue research in this field and developers to integrate recommender systems into their mind mapping tools.KeywordsMind mapUser modelingRecommender systems

User modeling is a key topic in many applications, mainly social networks and information retrieval systems. To assess the effectiveness of a user modeling approach, its capability to classify personal characteristics (e.g., the gender, age, or consumption grade of the users) is evaluated. Due to the fact that some of the attributes to predict are multiclass (e.g., age usually encompasses multiple ranges), assessing fairness in user modeling becomes a challenge since most of the related metrics work with binary attributes. As a workaround, the original multiclass attributes are usually binarized to meet standard fairness metrics definitions where both the target class and sensitive attribute (such as gender or age) are binary. However, this alters the original conditions, and fairness is evaluated on classes that differ from those used in the classification. In this article, we extend the definitions of four existing fairness metrics (related to disparate impact and disparate mistreatment) from binary to multiclass scenarios, considering different settings where either the target class or the sensitive attribute includes more than two groups. Our work endeavors to bridge the gap between formal definitions and real use cases in bias detection. The results of the experiments, conducted on four real-world datasets by leveraging two state-of-the-art graph neural network-based models for user modeling, show that the proposed generalization of fairness metrics can lead to a more effective and fine-grained comprehension of disadvantaged sensitive groups and, in some cases, to a better analysis of machine learning models originally deemed to be fair. The source code and the preprocessed datasets are available at the following link: https://github.com/erasmopurif/toward-responsible-fairness-analysis.

Trust prediction in online social networks is crucial for information dissemination, product promotion, and decision making. Existing work on trust prediction mainly utilizes the network structure or the low-rank approximation of a trust network. These approaches can suffer from the problem of data sparsity and prediction accuracy. Inspired by the homophily theory, which shows a pervasive feature of social and economic networks that trust relations tend to be developed among similar people, we propose a novel deep user model for trust prediction based on user similarity measurement. It is a comprehensive data sparsity insensitive model that combines a user review behavior and the item characteristics that this user is interested in. With this user model, we firstly generate a user's latent features mined from user review behavior and the item properties that the user cares. Then we develop a pair-wise deep neural network to further learn and represent these user features. Finally, we measure the trust relations between a pair of people by calculating the user feature vector cosine similarity. Extensive experiments are conducted on two real-world datasets, which demonstrate the superior performance of the proposed approach over the representative baseline works.

With the popularity of social media, opinion mining has gradually become a popular research field. Among these fields, sentiment analysis mining is an important research direction in the field of opinion mining. It aims to reveal the public's sentiment tendency, and attitude towards specific topics or events by analyzing text data generated by users on online platforms and digital media. However, the large amount of opinion data usually lacks effective annotation, which limits the learning and construction of opinion models. Therefore, focusing on the problem of the scarcity of labeled data in opinion analysis, this paper proposes a mining method for public opinion sentiment analysis based on multi-model fusion transfer learning, that can make full use of the limited labeled data to improve the learning efficiency of sentiment features by integrating the advantages of different models. Additionally, it introduces a transfer learning strategy to enable the models of the target domains to perform better in the absence of labeled data. Furthermore, the attention mechanism is combined to strengthen the acquisition of key features of the emotional colors and improve the accuracy of sentiment analysis. Specifically, the paper uses the ERNIE model to generate dynamic representations of the text word vectors in the dataset. It also uses TextCNN and BiGRU to construct a joint model for extracting local and overall features of the text word vectors. The parameters of the feature layer of the trained model are migrated to the target domain through transfer learning. The attention mechanism is combined with the model to identify the extreme elements of the sentiment. Finally, the local and overall features are fused to achieve comprehensive mining of public opinion and emotional information. This method can effectively improve the accuracy and generalization of public opinion analysis in cases of data scarcity. In the experimental part, the paper conducts comparisons and analyses in eight aspects: word embedding model, model combination, attention mechanism, transfer learning, source domain dataset, target domain dataset, model training, and baseline model. The four indicators, namely accuracy, precision, recall, and F1-measure are used to evaluate the performance of the method. The experiments are thorough and detailed, demonstrating the effective improvement of opinion mining performance.

At present, Graph Neural Network (GNN) methods usually follow the node centered message passing process and rely heavily on smooth node characteristics rather than graph structure. In view of this limitation, based on the heuristic method and graph attention mechanism, a feature fusion link prediction model (SAFL) combined with graph neural network is proposed. This model extracts the enclosing subgraphs around the target, combines the attention mechanism to assign neighbor weights to learn useful structural features, considers the impact of different nodes on the link, and fuses the graph neural network with the characteristics of input nodes to predict the link. The experiment on OGB dataset shows that the proposed link prediction model based on heuristic method enhances the graph structure characteristics, effectively represents the connectivity of enclosing subgraphs, and achieves better performance in link prediction.

In this paper we will describe how we model personal data in smart environments, which are user adaptive systems. Personal services in such smart environments can access user data by a consistent interface of our user modelling system. Considering inter-home mobility (i.e. users moving across multip

Nextpoint-of-interest (POI) recommendation has received widespread attention in recent years due to its superiority of recommending where users will go to next. However, there exist two limitations in many recommendation methods: (1) the hardness of modeling users' short-term preferences adaptively based on input sequence; (2) the efficient learning of the joint information between users' long- and short-term preferences. To this end, we propose an attention mechanism and adaptive convolution actuated fusion network (AMACF) innovatively, which optimizes forecast effectiveness of user preference. To better model some contextual information such as category, temporal, we utilize long- and short-term memory network to learn contextual features of POIs in historical check-ins and embed self-attention mechanism to capture users' preference in the long-term module. In the short-term module, for capturing the complicated interest, the adaptive convolution network (Ada-CN) is novelly proposed, which applies the attention mechanism to aggregate multiple parallel convolution kernels selectively and adjust to short-term preference according to users' continuously updated check-ins. Furthermore, an attention-based fusion mechanism is designed to combine long- and short-term preferences, where contributions to the next POI of different users are evaluated sufficiently. Experimental results on two real-world data sets indicates that AMACF outperforms baseline methods for next POI recommendation.

In recent years, the recommender system has been widely used in online platforms, which can extract useful information from giant volumes of data and recommend suitable items to the user according to user preferences. However, the recommender system usually suffers from sparsity and cold-start problems. Cross-domain recommendation, as a particular example of transfer learning, has been used to solve the aforementioned problems. However, many existing cross-domain recommendation approaches are based on matrix factorization, which can only learn the shallow and linear characteristics of users and items. Therefore, in this article, we propose a novel autoencoder framework with an attention mechanism (AAM) for cross-domain recommendation, which can transfer and fuse information between different domains and make a more accurate rating prediction. The main idea of the proposed framework lies in utilizing autoencoder, multilayer perceptron, and self-attention to extract user and item features, learn the user and item-latent factors, and fuse the user-latent factors from different domains, respectively. In addition, to learn the affinity of the user-latent factors between different domains in a multiaspect level, we also strengthen the self-attention mechanism by using multihead self-attention and propose AAM++. Experiments conducted on two real-world datasets empirically demonstrate that our proposed methods outperform the state-of-the-art methods in cross-domain recommendation and AAM++ performs better than AAM on sparse and large-scale datasets.

User alignment (UA), a central issue for social network analysis, aims to recognize the same natural persons across different social networks. Existing studies mainly focus on the positive effects of incorporating user attributes and network structure on UA. However, there have been few in-depth studies into the existing challenges for the joint integration of different types of text attributes, the imbalance between user attributes and network structure, and the utilization of massive unidentified users. To this end, this paper presents a high-accuracy embedding model named Joint embedding of Attributes and Relations for User Alignment (JARUA), to tackle the UA problem. First, a mechanism that can automatically identify the granularity of user attributes is introduced for handling multi-type user attributes. Second, a graph attention network is employed to extract the structural features and is integrated with user attributes features. Finally, an iterative training algorithm with quality filters is introduced to bootstrap the model performances. We evaluate JARUA on two real-world data sets. Experimental results demonstrate the superiority of the proposed method over several state-of-the-art approaches.