In e-commerce platforms, profit-aware recommender systems aim to improve the platform's profits while maintaining high overall accuracy by recommending items with high profits as top-ranked items. We explore two issues faced by existing model-based profit-aware approaches (i.e., MBAs) when training recommendation models for profit enhancement. First, existing MBAs tend to inaccurately infer the item ranking without considering the user's current preference for each item through their profit-based weighting scheme. Second, through the point-wise learning-to-rank (LTR), the model is optimized solely for the preference score of each item independently rather than being directly optimized for the overall ranking of items. To tackle these issues, we propose a novel MBA that involves three key steps: (S1) defining the Current Preference incorporated with Profit (i.e., CPP) for items; (S2) classifying items through CPP; and (S3) training the model by list-wise LTR based on CPP. Extensive experimental results using real-world platform datasets demonstrate that our approach improves accuracy by approximately 4% and profits by about 24% compared to the best-competing method.

To recommend the points of interest (POIs) that a user would check-in next, most deep-learning (DL)-based existing studies have employed random negative (RN) sampling during model training. In this paper, we claim and validate that, as the training proceeds, such an RN sampling in reality performs as sampling easy negative (EN) POIs (i.e., EN sampling) that a user was highly unlikely to check-in at her check-in time point. Furthermore, we verify that EN sampling is more disadvantageous in improving the accuracy than sampling hard negative (HN) POIs (i.e., HN sampling) that a user was highly likely to check-in. To address this limitation, we present the novel concept of the Degree of Positiveness (DoP), which can be formulated by two factors: (i) the degree to which a POI has the characteristics preferred by a user; (ii) the geographical distance between a user and a POI. Then, we propose a new model-training scheme based on HN sampling by using DoP. Using real-world datasets (i.e., NYC, TKY, and Brightkite), we demonstrate that all the state-of-the-art models trained by our scheme showed dramatic improvements in accuracy by up to about 82.8%.

From the observation that users reading news tend to not click outdated news, we propose the notion of 'lifetime' of news, with two hypotheses: (i) news has a shorter lifetime, compared to other types of items such as movies or e-commerce products; (ii) news only competes with other news whose lifetimes have not ended, and which has an overlapping lifetime (i.e., limited competitions). By further developing the characteristics of the lifetime of news, then we present a novel approach for news recommendation, namely, Lifetime-Aware News reCommEndeR System (LANCER) that carefully exploits the lifetime of news during training and recommendation. Using real-world news datasets (e.g., Adressa and MIND), we successfully demonstrate that state-of-the-art news recommendation models can get significantly benefited by integrating the notion of lifetime and LANCER, by up to about 40% increases in recommendation accuracy.

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).

Online news providers such as Google News, Bing News, and NAVER News collect a large number of news articles from a variety of presses and distribute these articles to users via their portals. Dynamic nature of a news domain causes the problem of information overload that makes it difficult for a user to find her preferable news articles. Motivated by this situation, NAVER Corp., the largest portal company in South Korea, identified four design considerations (DCs) for news recommendation that reflect the unique characteristics of a news domain. In this paper, we introduce a large-scale news recommender system named as AiRS, present how it jointly leverages the four DCs for NAVER News service. Specifically, AiRS first generates candidate articles for recommendation to a target user based on collaborative filtering (CF), quality estimation (QE), and social impact (SI) models; then, it ranks the candidate articles based on the scores computed by considering their multi-type feature scores (e.g., user's section preference and article's recency), finally recommending the top-$k$news articles that a target user is likely to prefer. Also, we present how to build the architecture for online deployment of AiRS at NAVER News. Through extensive offline and online A/B tests using the real-world datasets, we validate that AiRS successfully reflects all of the DCs into the news recommendation process, all design choices employed in AiRS help improve the recommendation accuracy, and AiRS significantly outperforms five state-of-the-art news recommendation approaches in terms of accuracy.

Explainable recommendation has gained great attention in recent years. A lot of work in this research line has chosen to use the knowledge graphs (KG) where relations between entities can serve as explanations. However, existing studies have not considered sentiment on relations in KG, although there can be various types of sentiment on relations worth considering (e.g., a user's satisfaction on an item). In this paper, we propose a novel recommendation framework based on KG integrated with sentiment analysis for more accurate recommendation as well as more convincing explanations. To this end, we first construct a Sentiment-Aware Knowledge Graph (namely, SAKG) by analyzing reviews and ratings on items given by users. Then, we perform item recommendation and reasoning over SAKG through our proposed Sentiment-Aware Policy Learning (namely, SAPL) based on a reinforcement learning strategy. To enhance the explainability for end-users, we further developed an interactive user interface presenting textual explanations as well as a collection of reviews related with the discovered sentiment. Experimental results on three real-world datasets verified clear improvements on both the accuracy of recommendation and the quality of explanations.

In recent years, quantization methods have successfully accelerated the training of large deep neural network (DNN) models by reducing the level of precision in computing operations (e.g., forward/backward passes) without sacrificing its accuracy. In this work, therefore, we attempt to apply such a quantization idea to the popular Matrix factorization (MF) methods to deal with the growing scale of models and datasets in recommender systems. However, to our dismay, we observe that the state-of-the-art quantization methods are not effective in the training of MF models, unlike their successes in the training of DNN models. To this phenomenon, we posit that two distinctive features in training MF models could explain the difference: (i) the training of MF models is much more memory-intensive than that of DNN models, and (ii) the quantization errors across users and items in recommendation are not uniform. From these observations, we develop a quantization framework for MF models, named MASCOT, employing novel strategies (i.e., m-quantization and g-switching) to successfully address the aforementioned limitations of quantization in the training of MF models. The comprehensive evaluation using four real-world datasets demonstrates that MASCOT improves the training performance of MF models by about 45%, compared to the training without quantization, while maintaining low model errors, and the strategies and implementation optimizations of MASCOT are quite effective in the training of MF models. For the detailed information about MASCOT, we release the code of MASCOT and the datasets at: https://github.com/Yujaeseo/lCDM-2021_MASCOT.