In this paper, we propose a new neural network architecture for item recommendation with structural information. Our model, structural neural recommender (SNR) is based on neural networks and operates on a hierarchy paradigm, aiming to explore the effectiveness of incorporating different structural information for recommendation. Many recent state-of-the-art neural network based recommendation models exploit the nonlinear transformations for modeling the complex user-item interaction patterns and user historical behaviors, ignoring the item-item structural relationship. This structural information, however, is uncomplicated to derive and useful for inferring item characteristics. To utilize this information, SNR simultaneously learns representation from user-item interactions and item-item relationships. Empirical studies on eight real-world datasets demonstrate the effectiveness of incorporating such structural information, by outperforming classic and recent baselines. We also conduct detail ablation studies and hyper-parameter analysis to provide further understanding towards the behaviors of our model. Following the model development, we conduct a field experiment to demonstrate that the effectiveness of algorithmic recommender systems can further increase by using different types of message framing when communicating recommendations to consumers. Our results suggest that recommendations framed with a relevance appeal (e.g. “Top 5 brands for you”) are more effective in general, yet recommendations that are framed with a popularity appeal (“Top 5 most popular brands”) are more effective for customers who were acquired via social media (versus non-social media) advertising or for those who have stronger (versus weaker) social orientation.

Recommender systems based on Collaborative Filtering (CF) algorithms have established an extensive means for retailers to suggest personalized item lists that will maximize each consumer’s utility. Nevertheless, in the mobile game industry, which characterized by the intense competition from the avalanche of other game options and fast-changing demands of game users, there has been no marked success with recommender systems. Instead, app stores merely show summaries of general market trends without any individual-level information, fail to suggest personalized lists based on preferences of the future. For modeling dynamics of game usage, we assume that an individual’s preferences on games can be represented as the proportion of each game’s running time, which can be calculated in daily basis by the individual’s usage time for each game apps divided by the individual’s total capacity. Then, we construct a tensor filled with the induced preferences. For the next step, we apply Bayesian probabilistic tensor factorization (BPTF), an extension of Singular Value Decomposition (SVD) to consider dynamic pattern, to restore all the empty entries of the tensor. Each restored component becomes an estimate of each user’s preference on each game at certain period. For empirical analysis, we use mobile log data in app-level for total 1,000 panels over 2 years. Top 100 mobile games in cumulative usage time are treated as focal apps in this study, making the dimension of the tensor by 1000 (users) * 100 (focal games) * 730 (2 years). We compare the model performance by the root-mean-squared error (RMSE) with that of baseline model, the static counterpart in collaborative filtering algorithm (Salakhutdinov, and Mnih, 2008). The results showed that our model (BPTF) defeats the baseline throughout overall user-game pairs, especially outperforming under the conditions that there are severe fluctuations in daily usage pattern and when the life span of newly adopted apps are relatively short. Furthermore, we compose personalized suggestions, which consists of the top-10 highly likable lists according to the predicted usage patterns for each individual, and compare the performance with that of the established general recommender system in app stores. For that matter, our suggestion also outweighed the existing recommender system by the typical performance metrics that commonly used in the mobile game industry.

In this study, a novel and flexible recommender system was developed, based on product taxonomy and usage patterns of users. The proposed system consists of the following four steps : (i) estimation of the product-preference matrix, (ii) construction of the product-preference matrix, (iii) estimation of the popularity and similarity levels for sought-after products, and (iv) recom- mendation of a products for the user. The product-preference matrix for each user is estimated through a linear combination of clicks, basket placements, and purchase statuses. Then the preference matrix of a particular genre is constructed by computing the ratios of the number of clicks, basket placements, and purchases of a product with respect to the total. The popularity and similarity levels of a user’s clicked product are estimated with an entropy index. Based on this information, collaborative and content-based filtering is used to recommend a product to the user. To assess the effectiveness of the proposed approach, an empirical study was conducted by constructing an experimental e-commerce site. Our results clearly showed that the proposed hybrid method is superior to conventional methods.

In this paper, the author devised new decision recommendation ordering method of items attributed by age to improve accuracy of recommender system. In conventional recommendation system, recommendation order is decided by high order of preference prediction. However, in this paper, recommendation accuracy is improved by decision recommendation order method that reflect age attribute of target customer and neighborhood in preference prediction. By applying decision recommendation order method to recommender system, recommendation accuracy is improved more than conventional ordering method of recommendation.

In this paper, the author proposed following two methods to improve the accuracy of the recommender system. First, in order to classify the users more accurately, the author used a EMC(Expanded Moving Center) heuristic algorithm which improved clustering accuracy. Second, the author proposed the Neighborhood-oriented preference prediction method that improved the conventional preference prediction methods, so the accuracy of the recommender system is improved. The test result of the recommender system which adapted the above two methods suggested in this paper was improved the accuracy than the conventional recommendation methods.

노인의 일상생활을 편안하고 즐겁게 지낼 수 있도록 도와주는 동반자 로봇의 기능 중 부정적인 감정/정서상태 개선을 위한 Infotainment Service를 소개한다. 노인의 일상생활 중에서 부정적인 감정/정서상태를 정의하고, 이들의 정서상태를 개선할 수 있는 방법들을 모색한다. 인지행동치료에서의 배경지식을 기반으로 노인의 부정적인 정서상태를 개선할 수 있는 애니메이션 클립들을 제작, 편집하여 검증해 보기로 한다. 또한, 검증된 애니메이션 클립들을 이용하여 바람직한 감정상태로의 전이를 위한 감성 컨텐츠를 제공하는 기능을 도출한다. 구체적으로 일련의 실험적 접근방법을 토대로 제작, 편집한 애니메이션 클립을 이용하여 영화의 감정요소를 분석할 수 있는 도구를 설계하고, 기존의 선호도를 고려한 영화추천 시스템을 확장한 감정요소를 고려한 영화 추천시스템을 제안한다.