Unsupervised Non-overlapping Cross-domain Recommendation (UNCR) is the task that recommends source domain items to the target domain users, which is more challenging as the users are non-overlapped, ...and its learning process is unsupervised. Unsupervised Non-overlapping Cross-domain Recommendation UNCR is still unsolved due to the following: (1) Previous studies need extra auxiliary information to learn transferable features when aligning two domains, which is unrealistic and hard to obtain due to privacy concerns. (2) Since the adoption of the shared network, existing works cannot well eliminate the domain-specific features in the common feature space, which may incorporate domain noise and harm the cross-domain recommendation. In this work, we propose a domain adaption-based method, namely DA-DAN, to address the above challenges. Specifically, to let DA-DAN be free of auxiliary information, we learn users’ preferences by only exploring their sequential patterns, and propose an improved self-attention layer to model them. To well eliminate the domain-specific features from the common feature space, we resort to a dual generative adversarial network with a multi-target adversarial loss, where two generators and discriminators are leveraged to model each domain separately. Experimental results on three real-world datasets demonstrate the advantage of DA-DAN compared with the state-of-the-art recommendation baselines. Moreover, our source codes have been publicly released.1
The objective of cross-domain sequential recommendation is to forecast upcoming interactions by leveraging past interactions across diverse domains. Most methods aim to utilize single-domain and ...cross-domain information as much as possible for personalized preference extraction and effective integration. However, on one hand, most models ignore that cross-domain information is composed of multiple single-domains when generating representations. They still treat cross-domain information the same way as single-domain information, resulting in noisy representation generation. Only by imposing certain constraints on cross-domain information during representation generation can subsequent models minimize interference when considering user preferences. On the other hand, some methods neglect the joint consideration of users’ long-term and short-term preferences and reduce the weight of cross-domain user preferences to minimize noise interference. To better consider the mutual promotion of cross-domain and single-domains factors, we propose a novel model (C2DREIF) that utilizes Gaussian graph encoders to handle information, effectively constraining the correlation of information and capturing useful contextual information more accurately. It also employs a Top-down transformer to accurately extract user intents within each domain, taking into account the user’s long-term and short-term preferences. Additionally, entropy regularized is applied to enhance contrastive learning and mitigate the impact of randomness caused by negative sample composition.
•A Gaussian distribution-based graph encoder strengthens cross-domain constraints.•A Top-down transformer is used to capture users’ authentic intentions.•An entropy loss is introduced to enhance contrastive learning.
Two characteristics of location-based services are mobile trajectories and the ability to facilitate social networking. The recording of trajectory data contributes valuable resources towards ...understanding users’ geographical movement behaviors. Social networking is possible when users are able to quickly connect to anyone nearby. A social network with location based services is known as location-based social network (LBSN). As shown in Cho et al. 2013, locations that are frequently visited by socially related persons tend to be correlated, which indicates the close association between social connections and trajectory behaviors of users in LBSNs. To better analyze and mine LBSN data, we need to have a comprehensive view of each of these two aspects, i.e., the mobile trajectory data and the social network.
Specifically, we present a novel neural network model that can jointly model both social networks and mobile trajectories. Our model consists of two components: the construction of social networks and the generation of mobile trajectories. First we adopt a network embedding method for the construction of social networks: a networking representation can be derived for a user. The key to our model lies in generating mobile trajectories. Second, we consider four factors that influence the generation process of mobile trajectories: user visit preference, influence of friends, short-term sequential contexts, and long-term sequential contexts. To characterize the last two contexts, we employ the RNN and GRU models to capture the sequential relatedness in mobile trajectories at the short or long term levels. Finally, the two components are tied by sharing the user network representations. Experimental results on two important applications demonstrate the effectiveness of our model. In particular, the improvement over baselines is more significant when either network structure or trajectory data is sparse.
Compared with only pursuing recommendation accuracy, the explainability of a recommendation model has drawn more attention in recent years. Many graph-based recommendations resort to informative ...paths with the attention mechanism for the explanation. Unfortunately, these attention weights are intentionally designed for model accuracy but not explainability. Recently, some researchers have started to question attention-based explainability because the attention weights are unstable for different reproductions, and they may not always align with human intuition. Inspired by the counterfactual reasoning from causality learning theory, we propose a novel explainable framework targeting path-based recommendations, wherein the explainable weights of paths are learned to replace attention weights. Specifically, we design two counterfactual reasoning algorithms from both path representation and path topological structure perspectives. Moreover, unlike traditional case studies, we also propose a package of explainability evaluation solutions with both qualitative and quantitative methods. We conduct extensive experiments on four real-world datasets, the results of which further demonstrate the effectiveness and reliability of our method.
The graph-based recommendation systems achieve significant success, yet they are accompanied by malicious attacks. In most scenes, attackers will inject crafted fake profiles into the recommendation ...system to boost the ranking of their target items in the recommendation lists. For an exploration of potential attacks hidden in real life, researchers have proposed various attacks with severe threats. However, current efforts in exploring attack strategies neglect the stealthiness aspect of the attack, i.e., the perturbation of recommendation performance after an attack can be quite noticeable, potentially alerting the defenders. To fill this research gap, this paper introduces a novel attack framework named InfoAtk, designed to conduct attacks while ensuring stealthiness. Specifically, given the dependency of precise recommendation predominantly on representations, the framework employs contrastive learning techniques to align representations before and after the attack, thereby augmenting stealthiness. Additionally, we optimize the representation of target items to outrank the last items in users’ recommendation lists, thereby promoting the visibility of the target item to increase the attack’s effectiveness. Extensive experiments on four public datasets validate the stealthiness and effectiveness of our proposed attack framework.
•We explore the stealthiness of recommender attack to fill this research gap.•We introduce a poisoning attack method based on a bi-level optimization framework.•We leverage contrastive learning to align item embeddings for stealthiness.•We boost attack effectiveness with minimum cost by optimizing target item embedding.•Our experiments show the stealthiness and effectiveness of our attack.
With the widespread popularization of social network platforms, user-generated content and other social network data are growing rapidly. It is difficult for social users to select interested ...contents from the numerous social data. To alleviate information overload problem and enhance overall user experience of social networks, recommendation systems relying on historical behavioural data and social friendship relations of users, are widely used in social networks. Although researches on social recommendations have been conducted in recent years, recommendation systems of social networks still suffer from several challenges, such as data sparsity and lower performance. Since graph neural network has huge advantages in graph data learning by aggregating neighbors representations of the central node, it has been gathering pace in recent years. In this survey, we review graph neural network based literature for solving recommendation problems in social networks. We first introduce backgrounds of graph neural network and recommendation systems in social networks. Then, for different types of recommendation problems in social networks, we review different graph neural network based recommendation methods briefly. In particular, we first review GNN-based methods for general social recommendation and then review GNN-based methods for different social recommendation scenarios (such as friend recommendation and point-of-interest recommendation). Finally, we briefly discuss promising future directions of the graph neural network based recommendation in social networks.
Recommender systems exploit interaction history to estimate user preference, having been heavily used in a wide range of industry applications. However, static recommendation models are difficult to ...answer two important questions well due to inherent shortcomings: (a) What exactly does a user like? (b) Why does a user like an item? The shortcomings are due to the way that static models learn user preference, i.e., without explicit instructions and active feedback from users. The recent rise of conversational recommender systems (CRSs) changes this situation fundamentally. In a CRS, users and the system can dynamically communicate through natural language interactions, which provide unprecedented opportunities to explicitly obtain the exact preference of users. Considerable efforts, spread across disparate settings and applications, have been put into developing CRSs. Existing models, technologies, and evaluation methods for CRSs are far from mature. In this paper, we provide a systematic review of the techniques used in current CRSs. We summarize the key challenges of developing CRSs in five directions: (1) Question-based user preference elicitation. (2) Multi-turn conversational recommendation strategies. (3) Dialogue understanding and generation. (4) Exploitation-exploration trade-offs. (5) Evaluation and user simulation. These research directions involve multiple research fields like information retrieval (IR), natural language processing (NLP), and human-computer interaction (HCI). Based on these research directions, we discuss some future challenges and opportunities. We provide a road map for researchers from multiple communities to get started in this area. We hope this survey can help to identify and address challenges in CRSs and inspire future research.
With the expansion of the online environment, recently, recommendation systems have become established as an essential element of any online service. Following this trend, the issue of how to present ...recommended information effectively to users is attracting attention not only in companies but also in the research field. This paper presents a between-subjects study that aimed to elicit whether there is a difference in consumer’s attitudes towards the user-centric and content-centric recommendation approaches based on their level of psychological ownership towards the online service. Our findings indicated that users with high psychological ownership toward the online service prefer the user-centric recommendation approach, while users with low psychological ownership prefer the content-centric recommendation approach.
In this study, we show that individual users’ preferences for the level of diversity, popularity, and serendipity in recommendation lists cannot be inferred from their ratings alone. We demonstrate ...that we can extract strong signals about individual preferences for recommendation diversity, popularity and serendipity by measuring their personality traits. We conducted an online experiment with over 1,800 users for six months on a live recommendation system. In this experiment, we asked users to evaluate a list of movie recommendations with different levels of diversity, popularity, and serendipity. Then, we assessed users’ personality traits using the Ten-item Personality Inventory (TIPI). We found that ratings-based recommender systems may often fail to deliver preferred levels of diversity, popularity, and serendipity for their users (e.g. users with high-serendipity preferences). We also found that users with different personalities have different preferences for these three recommendation properties. Our work suggests that we can improve user satisfaction when we integrate users’ personality traits into the process of generating recommendations.
We frequently rely on suggestions in an online context, including those from search engine results, e-commerce product recommendations, movie recommendations, and so forth. These suggestions are made ...in response to our actions, such as when we search for products, songs, or movies, or at the very least, when we visit the website in order to receive a recommendation. These recommendations do not take into account time-sensitive factors, such as when we often utilise these services during the day or the week. We prefer to repeat some activities, like listening to music at a specific time of day. We presented a methodology in this paper that addresses two critical issues in personalised recommendation. 1) recommending the right items to the right people at the right time, and 2) estimating when a user will return to a service or product after performing repeated actions. The scholarly community has not yet examined this work in personalised recommendations. To do this, we offered the Hawkes process, in which we employed a customised initial intensity based on Hierarchical Poisson Factorization, and for a dynamic activity, we considered a sinusoidal function coupled with exponential effect decay. This is consistent with user activity cycles, such as music listening during a specific time of day. This is the first framework we've used, which is based on a probabilistic matrix factorization