As the number of TV shows increases, designing recommendation systems to provide users with their favorable TV shows becomes more important. In a TV show domain, watching a TV show (i.e., giving implicit feedback to the show) among the TV shows broadcast at the same time frame implies that the currently watching show is the winner in the competition with others (i.e., losers). However, in previous studies, such a notion of limited competitions has not been considered in estimating the user’s preferences for TV shows. In this paper, we propose a new recommendation framework to take this new notion into account based on pair-wise models. Our framework is composed of the following ideas: (i) identify winners and losers by determining pairs of competing TV shows; (ii) learn the pairs of competing TV shows based on the confidence for the pair-wise preference between the winner and the loser; (iii) recommend the most favorable TV shows by considering the time factors with respect to users and TV shows. Using a real-world TV show dataset, our experimental results show that our proposed framework consistently improves the accuracy of recommendation by up to 38%, compared with the best state-of-the-art method. The code and datasets of our framework are available in an external link (https://github.com/hongkyun-bae/tvshow_rs).

The political stance prediction for news articles has been widely studied to mitigate the echo chamber effect – people fall into their thoughts and reinforce their pre-existing beliefs. The previous works for the political stance problem focus on (1) identifying political factors that could reflect the political stance of a news article and (2) capturing those factors effectively. Despite their empirical successes, they are not sufficiently justified in terms of how effective their identified factors are in the political stance prediction. Motivated by this, in this work, we conduct a user study to investigate important factors in political stance prediction, and observe that the context and tone of a news article (implicit) and external knowledge for real-world entities appearing in the article (explicit) are important in determining its political stance. Based on this observation, we propose a novel knowledge-aware approach to political stance prediction (KHAN), employing (1) hierarchical attention networks (HAN) to learn the relationships among words and sentences in three different levels and (2) knowledge encoding (KE) to incorporate external knowledge for real-world entities into the process of political stance prediction. Also, to take into account the subtle and important difference between opposite political stances, we build two independent political knowledge graphs (KG) (i.e., KG-lib and KG-con) by ourselves and learn to fuse the different political knowledge. Through extensive evaluations on three real-world datasets, we demonstrate the superiority of KHAN in terms of (1) accuracy, (2) efficiency, and (3) effectiveness.

In this article, we propose a framework for embedding-based community detection on signed networks, namelyAdversarial learning ofBalanced triangle forCommunity detection, in short${{\sf ABC}}$. It first represents all the nodes of a signed network as vectors in low-dimensional embedding space and conducts a clustering algorithm (e.g.,k-means) on vectors, thereby detecting a community structure in the network. When performing the embedding process,${{\sf ABC}}$learns only the edges belonging to balanced triangles whose edge signs follow the balance theory, significantly excluding noise edges in learning. To address the sparsity of balanced triangles in a signed network,${{\sf ABC}}$learns not only the edges in balancedreal-triangles but those in balancedvirtual-triangles that do not actually exist but are produced by our generator. Finally,${{\sf ABC}}$employs adversarial learning to generate more-realistic balanced virtual-triangles with less noise edges. Through extensive experiments using seven real-world networks, we validate the effectiveness of (1) learning edges belonging to balanced real/virtual-triangles and (2) employing adversarial learning for signed network embedding. We show that${{\sf ABC}}$consistently and significantly outperforms the state-of-the-art community detection methods in all datasets.

One of the significant social problems emerging in modern society is mental illness, and a growing number of people are seeking psychological help. Art therapy is a technique that can alleviate psychological and emotional conflicts through creation. However, the expression of a drawing varies by individuals, and the subjective judgments made by art therapists raise the need to secure an objective assessment. In this paper, we present M2C (Multimodal classification with 2-stage Co-attention), a deep learning model that predicts stress from art therapy psychological test data. M2C employs a co-attention mechanism that combines two modalities-drawings and post-questionnaire answers-to complement the weaknesses of each, which corresponds to therapists' psychometric diagnostic processes. The results of the experiment show that M2C yielded higher performance than other state-of-the-art single- or multi-modal models, demonstrating the effectiveness of the co-attention approach that reflects the diagnosis process.

Spatio-temporal models, which combine GNNs (Graph Neural Networks) and RNNs (Recurrent Neural Networks), have shown state-of-the-art accuracy in traffic speed prediction. However, we find that they consider the spatial and temporal dependencies between speeds separately in the two (i.e., space and time) dimensions, thereby unable to exploit the joint-dependencies of speeds in space and time. In this paper, with the evidence via preliminary analysis, we point out the importance of considering individual dependencies between two speeds from all possible points in space and time for accurate traffic speed prediction. Then, we propose an Individual Spatio-Temporal graph (IST-graph) that represents the Individual Spatio-Temporal dependencies (IST-dependencies) very effectively and a Spatio-Temporal Graph ATtention network (ST-GAT), a novel model to predict the future traffic speeds based on the IST-graph and the attention mechanism. The results from our extensive evaluation with five real-world datasets demonstrate (1) the effectiveness of the IST-graph in modeling traffic speed data, (2) the superiority of ST-GAT over 5 state-of-the-art models (i.e., 2-33% gains) in prediction accuracy, and (3) the robustness of our ST-GAT even in abnormal traffic situations.

Many global automakers strive to develop technologies towards the next-generation of intelligent transportation systems (ITS). One of the primary goals of ITS is predicting future traffic speeds to optimize a driver's route, which can lead to not only alleviating traffic flow but also increasing user satisfaction with an ITS service. While prior studies have applied deep learning models to traffic speed prediction and improved model performance, existing models did not well capture abrupt speed changes. In this paper, we propose a novel model, named as adversarial prediction of traffic speed (APOTS), based on adversarial training, data augmentation, and hybrid deep learning modeling. Through the experiments with real traffic data provided by Hyundai Motor Company, we demonstrate that APOTS effectively learns dynamics of traffic speed changes and predicts traffic speed up to 40% higher in accuracy than existing prediction models.

As an emotion plays an important role in people’s everyday lives and is often mirrored in their social media use, extensive research has been conducted to characterize and model emotions from social media data. However, prior research has not sufficiently considered trends of social media use—the increasing use of images and the decreasing use of text—nor identified the features of images in social media that are likely to be different from those in non-social media. Our study aims to fill this gap by (1) considering the notion of visual hints that depict contextual information of images, (2) presenting their characteristics in positive or negative emotions, and (3) demonstrating their effectiveness in emotion prediction modeling through an in-depth analysis of their relationship with the text in the same posts. The results of our experiments showed that our visual hint-based model achieved 20% improvement in emotion prediction, compared with the baseline. In particular, the performance of our model was comparable with that of the text-based model, highlighting not only a strong relationship between visual hints of the image and emotion, but also the potential of using only images for emotion prediction which well reflects current and future trends of social media use.

In this paper, we propose a framework for embedding-based community detection on signed networks. It first represents all the nodes of a signed network as vectors in low-dimensional embedding space and conducts a clustering algorithm (e.g., k-means) on vectors, thereby detecting a community structure in the network. When performing the embedding process, our framework learns only the edges belonging to balanced triangles whose edge signs follow the balance theory, significantly excluding noise edges in learning. To address the sparsity of balanced triangles in a signed network, our framework learns not only the edges in balanced real-triangles but those in balanced virtual-triangles that are produced by our generator. Finally, our framework employs adversarial learning to generate more-realistic balanced virtual-triangles with less noise edges. Through extensive experiments using seven real-world networks, we validate the effectiveness of (1) learning edges belonging to balanced real/virtual-triangles and (2) employing adversarial learning for signed network embedding. We show that our framework consistently and significantly outperforms the state-of-the-art community detection methods in all datasets.

Fashion style analysis is of the utmost importance for fashion professionals. However, it has an issue of having different style classification criteria that rely heavily on professionals’ subjective experiences with no quantitative criteria. We present FANCY (Fashion Attributes detectioN for Clustering stYle), a human-centered, deep learning-based framework to support fashion professionals’ analytic tasks using a computational method integrated with their insights. We work closely with fashion professionals in the whole study process to reflect their domain knowledge and experience as much as possible. We redefine fashion attributes, demonstrate a strong association with fashion attributes and styles, and develop a deep learning model that detects attributes in a given fashion image and reflects fashion professionals’ insight. Based on attribute-annotated 302,772 runway fashion images, we developed 25 new fashion styles (FANCY dataset 1). We summarize quantitative standards of the fashion style groups and present fashion trends based on time, location, and brand.

This study examines the impact of the 'diversity' of product recommendations on the 'preference' of a customer, using online/offline data from a leading fashion company. First, through interviews with fashion professionals, we categorized the characteristics of customers into four types - gift, coordinator, carry-over, and trendsetter. Then, using a hybrid filtering method, we increased the accuracy and diversity of recommended products. We derived 13 salient features that reflect customer behavior based on the Purchase Funnel model and built a classification model that predicts a customer's preference rates. Second, we conducted two large-scale user tests with 20,000 real customers to verify the effectiveness of our recommendation system. Study results empirically demonstrated the importance of diversity of recommended products. The more diverse the product recommendations were, the higher the purchase rate, the average purchase amount, and the cross purchase rate were observed. In addition, we tracked the customers? purchase for two months after the user tests and found that diverse product exposure positively influenced customer retention (e.g., repurchase rate, amount).

As social network services (SNS) are expanding from friend-based to interest-based, users form a new type of relationships, namely interest-based relationships, with friends and others through social activities (e.g., likes, comments). Although such relationships are highlighted in the common-identity theory and have important values in theoretical and practical aspects, little evidence exists in the literature pertaining to the explanation of social activities as a central component for social network analysis and an association with friendship. In this paper, we build like networks in Instagram and analyze them through the lens of two salient aspects - friendship and interest - that constitute social networks. Our study results (1) show ambiguous interpretations of the like activities between users who are friends, based on the comparative analysis between friend- and non-friend-based like networks, and (2) demonstrate strong signals of the hashtag characterizing the interest-based relationships in users and content. Our research substantiates and gives insights on the common-identity theory applied in online social networks through data-driven, empirical analysis.

Motivated by a success of generative adversarial networks (GAN) in various domains including information retrieval, we propose a novel signed network embedding framework, ASiNE, which represents each node of a given signed network as a low-dimensional vector based on the adversarial learning. To do this, we first design a generator G+ and a discriminator D+ that consider positive edges, as well as a generator G - and a discriminator D- that consider negative edges: (1) G+/G- aim to generate the most indistinguishable fake positive/negative edges, respectsupively; (2) D+/D aim to discriminate between real positive/negative edges and fake positive/negative edges, respectively. Furthermore, under ASiNE, we propose two new strategies for effective signed network embedding: (1) an embedding space sharing strategy for learning both positive and negative edges; (2) a fake edge generation strategy based on the balance theory. Through extensive experiments using five real-life signed networks, we verify the effectiveness of each of the strategies employed in ASiNE. We also show that ASiNE consistently and significantly outperforms all the state-of-the-art signed network embedding methods in all datasets and with all metrics in terms of accuracy of sign prediction.

As the number of TV channels increases, it is becoming important to recommend TV shows that users prefer to watch. To this end, we investigate the inherent characteristics of implicit feedback given in the TV show domain, and identify the challenges for building an effective TV show recommendation. Based on the unique characteristics, we define a user's watchable interval, the most important and novel concept in understanding users' true preferences. In order to reflect this new concept into the TV show recommendation, we propose a novel framework based on collaborative filtering. Our framework is composed of (1) preference estimation based on a user's watchable interval, (2) preference prediction based on confidence exploiting watchable episodes, and (3) top-N recommendation considering TV show's staying and remaining times. Using a real-world TV show dataset, we demonstrate that our framework effectively solves the challenges and significantly outperforms other existing state-of-the-art methods.

Social media platforms such as Twitter and Facebook are commonly used to disseminate up-to-date news information, but they also contain a lot of noise and irrelevant content. The contents of social media platforms are typically filtered by followship or friendship oriented relationships, and is almost always driven by trending news topics at the national scale, making it difficult for users to gather useful information that is most pertinent to a local community context. Research has utilized content analysis techniques to gain insights on the sentiment expressed about political topics on social media sites. However, there has been little attempt to understand how users would perceive this information if opinions and sentiments about news topics were externalized and made aware to them. We designed Community Poll, a smartphone application that aggregates local news feeds with relevant tweets about the local news topics. A Public Attitude Meter is calculated based on the sentiment score of the tweets for each of the local news topic presented in the system. We conducted a 2-week deployment with 16 users about their perception of the system. The users reported that Community Poll helps them digest locally relevant news topics, and quickly gather public opinions associated with the topics. They reported that being aware of public sentiment encouraged them to more actively participate in discussions on social media. Curiosity about a score-based representation is an important element that drove them to consume local news topics that they wouldn’t otherwise be exposed to. Interestingly, although being aware of public sentiment served to reaffirm people’s positions on the local topics, users expressed concerns about how sentiment awareness might bias other people’s judgments regarding news topics.