Long-term Preferences Research Articles

Structure Aware Incremental Learning with Personalized Imitation Weights for Recommender Systems

Recommender systems now consume large-scale data and play a significant role in improving user experience. Graph Neural Networks (GNNs) have emerged as one of the most effective recommender system models because they model the rich relational information. The ever-growing volume of data can make training GNNs prohibitively expensive. To address this, previous attempts propose to train the GNN models incrementally as new data blocks arrive. Feature and structure knowledge distillation techniques have been explored to allow the GNN model to train in a fast incremental fashion while alleviating the catastrophic forgetting problem. However, preserving the same amount of the historical information for all users is sub-optimal since it fails to take into account the dynamics of each user's change of preferences. For the users whose interests shift substantially, retaining too much of the old knowledge can overly constrain the model, preventing it from quickly adapting to the users’ novel interests. In contrast, for users who have static preferences, model performance can benefit greatly from preserving as much of the user's long-term preferences as possible. In this work, we propose a novel training strategy that adaptively learns personalized imitation weights for each user to balance the contribution from the recent data and the amount of knowledge to be distilled from previous time periods. We demonstrate the effectiveness of learning imitation weights via a comparison on five diverse datasets for three state-of-art structure distillation based recommender systems. The performance shows consistent improvement over competitive incremental learning techniques.

Greenbelt investment with temporal preference in uncertain environments

While greenbelt is recognized as an essential part of sustainable cities, yet its investment with time preference can be affected by environments. Here, we demonstrated that the present bias problem substantially impeded the willingness to pay for greenbelt. We proposed a game with environmental feedbacks where investment in greenbelt was assessed based only on temporal preference. By studying urban environments of Tokyo, we concluded that total investment increased 2.7 times due to long-term preference in real-world scenario. Total investment increased 1.04 times due to urban warming, 1.1 times due to population centrality, and 1.2 times due to urban greening. Also, population aging of Tokyo reduced investment in greenbelt. Our findings highlight the understanding of intergenerational investment in urban policy to support development of adaptation strategy to environments.

A Multi-Task Graph Neural Network with Variational Graph Auto-Encoders for Session-Based Travel Packages Recommendation

Session-based travel packages recommendation aims to predict users’ next click based on their current and historical sessions recorded by Online Travel Agencies (OTAs). Recently, an increasing number of studies attempted to apply Graph Neural Networks (GNNs) to the session-based recommendation and obtained promising results. However, most of them do not take full advantage of the explicit latent structure from attributes of items, making learned representations of items less effective and difficult to interpret. Moreover, they only combine historical sessions (long-term preferences) with a current session (short-term preference) to learn a unified representation of users, ignoring the effects of historical sessions for the current session. To this end, this article proposes a novel session-based model named STR-VGAE, which fills subtasks of the travel packages recommendation and variational graph auto-encoders simultaneously. STR-VGAE mainly consists of three components: travel packages encoder , users behaviors encoder , and interaction modeling . Specifically, the travel packages encoder module is used to learn a unified travel package representation from co-occurrence attribute graphs by using multi-view variational graph auto-encoders and a multi-view attention network. The users behaviors encoder module is used to encode user’ historical and current sessions with a personalized GNN, which considers the effects of historical sessions on the current session, and coalesce these two kinds of session representations to learn the high-quality users’ representations by exploiting a gated fusion approach. The interaction modeling module is used to calculate recommendation scores over all candidate travel packages. Extensive experiments on a real-life tourism e-commerce dataset from China show that STR-VGAE yields significant performance advantages over several competitive methods, meanwhile provides an interpretation for the generated recommendation list.

DCARS: Deep context-aware recommendation system based on session latent context

Recommendation systems (RSs) usually create suggestions based on users’ prior intentions. Users’ interests may evolve due to context change or user-mode change. Discovering such a change is crucial for producing personalized suggestions. Traditional approaches assume that each user has a fixed preference. On the contrary, context-aware recommendation systems (CARSs) use contextual information to detect user intention changes. However, applying contextual information is the main challenge in CARSs, because it is not always feasible to achieve all the users’ contextual information. Furthermore, adding different contexts to RSs grows its dimensionality in multiple applications. Besides, existing CARSs cannot precisely obtain the hierarchical relationships between items and contexts items that influence users’ intentions. They often use short-term interest with either static long-term preference in the recommendation process. To alleviate the mentioned challenges, we propose a novel deep context-aware recommendation system (DCARS) to capture and incorporate user preferences changes in the recommendation process. The proposed method models the latent context among selected items in each session throughout users’ historical interactions and combines users’ short-term and long-term preferences to generate recommendations. Specifically, we suggest a DCARS based on latent representations of sessions derived from users’ activities. The experiment results on benchmark context-aware data sets show that the proposed DCARS model surpasses state-of-the-art approaches.

Spatially explicit, quantitative reconstruction of past vegetation based on pollen or charcoal data as a tool for autecology of trees

ContextThe determination of autecological preferences based on long-term vegetation dynamics is hampered by the lack of realistic estimates for past occurrence and abundance patterns. Palaeoecological record has still rather character of points than spatially continuous maps.ObjectivesTo infer long-term autecological preferences of trees from reconstructed vegetation. Compare reconstructions based on pollen and charcoals.MethodsWe employed to the regional training set of 58 sites the Extended Downscaling Approach (EDA) using nine topographic factors clustered in 8 habitat classes, data on pollen productivity estimates, fossil pollen, charcoal sequences from soil and archaeological contexts. Based on abundances and habitat preferences from the last 9 millennia, we calculated the autecological preferences of tree taxa, using multivariate statistics.ResultsThe significant spatiotemporal patterns between soil-charcoal and pollen-based EDA validated the reconstruction, the use of both records in the EDA, and the EDA model. One of the topographic indices—vertical distance to channel network—evidenced the following: the closest taxon to the groundwater is Picea; Abies, Betula, Pinus and Quercus have intermediate distances; Fagus grows far from the channel network and Corylus even further.ConclusionsThe EDA model linked past forest composition to realistic topography. Such a spatially explicit reconstruction produced by our new algorithm allows inferring the relationship between past plant communities and environmental variables. The long-term preferences of tree species to habitat characteristics match their current autecological demands. This might be a breakthrough in quantitative plant paleoecology.

Spatio-temporal intention learning for recommendation of next point-of-interest

ABSTRACT Next point-of-interest (POI) recommendation has been applied by many internet companies to enhance the user travel experience. Recent research advocates deep-learning methods to model long-term check-in sequences and mine mobility patterns of people to improve recommendation performance. Existing approaches model general user preferences based on historical check-ins and can be termed as preference pattern models. The preference pattern is different from the intention pattern, in that it does not emphasize the user mobility pattern of revisiting POIs, which is a common behavior and kind of intention for users. An effective module is needed to predict when and where users will repeat visits. In this paper, we propose a Spatio-Temporal Intention Learning Self-Attention Network (STILSAN) for next POI recommendation. STILSAN employs a preference-intention module to capture the user’s long-term preference and recognizes the user’s intention to revisit some specific POIs at a specific time. Meanwhile, we design a spatial encoder module as a pretrained model for learning POI spatial feature by simulating the spatial clustering phenomenon and the spatial proximity of the POIs. Experiments are conducted on two real-world check-in datasets. The experimental results demonstrate that all the proposed modules can effectively improve recommendation accuracy and STILSAN yields outstanding improvements over the state-of-the-art models.

UPTDNet: A User Preference Transfer and Drift Network for Cross-City Next POI Recommendation

Cross-city point of interest (POI) recommendation for tourists in an unfamiliar city has high application value but is challenging due to the data sparsity. Most existing models attempt to alleviate the sparsity problem by learning the user preference transfer and drift. However, they either fail to simultaneously model the preference transfer and drift in both long- and short-term user preferences or cannot accomplish the task of the next POI recommendation, which is crucial for a wide spectrum of applications ranging from transportation and urban planning to advertising. To address the limitation, we proposed a user preference transfer and drift network (UPTDNet) for cross-city next POI recommendation. UPTDNet excels at cross-city recommendations by learning the transfer and drift of both long- and short-term preferences. For short-term preference, dual recurrent neural network-based (RNN-based) branches are designed to model preference transfer from tourist’s current city and drift among different user roles. For long-term preference, a mapping function and user similarity calculation are employed for preference transfer from the tourist’s home city and drift among individual users. Experiments are conducted on the Gowalla and Foursquare datasets, and the results show that UPTDNet consistently and significantly outperforms state-of-the-art models by an average of 10.22% to 22.63% in the next POI recommendation task. Ablation study and further analysis validate the effectiveness and plausibility of considering both user preference transfer and drift in the cross-city recommendation.

SanMove: next location recommendation via self-attention network

PurposeThis paper aims to address the following issues: (1) most existing methods are based on recurrent network, which is time-consuming to train long sequences due to not allowing for full parallelism; (2) personalized preference generally are not considered reasonably; (3) existing methods rarely systematically studied how to efficiently utilize various auxiliary information (e.g. user ID and time stamp) in trajectory data and the spatiotemporal relations among nonconsecutive locations.Design/methodology/approachThe authors propose a novel self-attention network–based model named SanMove to predict the next location via capturing the long- and short-term mobility patterns of users. Specifically, SanMove uses a self-attention module to capture each user's long-term preference, which can represent her personalized location preference. Meanwhile, the authors use a spatial-temporal guided noninvasive self-attention (STNOVA) module to exploit auxiliary information in the trajectory data to learn the user's short-term preference.FindingsThe authors evaluate SanMove on two real-world datasets. The experimental results demonstrate that SanMove is not only faster than the state-of-the-art recurrent neural network (RNN) based predict model but also outperforms the baselines for next location prediction.Originality/valueThe authors propose a self-attention-based sequential model named SanMove to predict the user's trajectory, which comprised long-term and short-term preference learning modules. SanMove allows full parallel processing of trajectories to improve processing efficiency. They propose an STNOVA module to capture the sequential transitions of current trajectories. Moreover, the self-attention module is used to process historical trajectory sequences in order to capture the personalized location preference of each user. The authors conduct extensive experiments on two check-in datasets. The experimental results demonstrate that the model has a fast training speed and excellent performance compared with the existing RNN-based methods for next location prediction.

Dual-path recommendation algorithm based on CNN and attention-enhanced LSTM

ABSTRACT To recommend useful information to users more efficiently, this paper proposes a dual-path recommendation algorithm which combines multilayer Convolutional Neural Network (CNN) and attention-enhanced long short-term memory network (Attention-LSTM). Firstly, the matrix factorisation technique is used for learning the long-term preferences of users. Secondly, a dual-path network based on CNN and LSTM is constructed to perform feature extraction on the rating matrix. The dual-path network can learn the long-term preferences of users while capturing their dynamic preferences in changing preferences. The algorithm is tested on the public dataset MovieLens-1M, and the MAE value reflects the accuracy of the algorithm.

GNN-based long and short term preference modeling for next-location prediction

Next-location prediction is a special task of the next POIs recommendation. Different from general recommendation tasks, next-location prediction is highly context-dependent: (1) sequential dependency, i.e., the sequential locations checked in by a user have high correlation; (2) temporal dependency, i.e. check-in preferences are identified as different days or nights; (3) spatial dependency, i.e. users prefer to visit closer locations. Recent studies have been very successful in predicting users' next location by comprehensively considering user preferences. Nonetheless, these methods not only fail to capture temporal dependencies but also fail to capture location topology information. To fill this gap, we propose a GNN-based model which converts POIs into a low-dimensional metric and integrates users' long-term and short-term preferences to comprehensively represent dynamic preferences. The model consists of graph neural networks for long-term preference modeling and LSTM for short-term preference modeling. Comprehensive experiments are conducted on two real-world datasets, and results demonstrate the effectiveness of our approach over state-of-the-art methods for next-location prediction.

A Deep Reinforcement Learning Real-Time Recommendation Model Based on Long and Short-Term Preference

With the development of Internet technology, the problem of information overload has increasingly attracted attention. Nowadays, the recommendation system with excellent performance in information retrieval and filtering would be widely used in the business field. However, most existing recommendation systems are considered a static process, during which recommendations for internet users are often based on pre-trained models. A major disadvantage of these static models is that they are incapable of simulating the interaction process between users and their systems. Moreover, most of these models only consider users’ real-time interests while ignoring their long-term preferences. This paper addresses the abovementioned issues and proposes a new recommendation model, DRR-Max, based on deep reinforcement learning (DRL). In the proposed framework, this paper adopted a state generation module specially designed to obtain users’ long-term and short-term preferences from user profiles and user history score item information. Next, Actor-Critical algorithm is used to simulate the real-time recommendation process.Finally, this paper uses offline and online methods to train the model. In the online mode, the network parameters were dynamically updated to simulate the interaction between the system and users in a real recommendation environment. Experimental results on the two publicly available data sets were used to demonstrate the effectiveness of our proposed model.

Can Direct Subsidies or Tax Incentives Improve the R&D Efficiency of the Manufacturing Industry in China?

The understanding of the impact of different government support methods on R&D efficiency is of great significance for evaluating the performance of innovation policies in various countries. We selected 31 manufacturing industries in China from 2009 to 2015, used the stochastic frontier analysis (SFA) method to measure R&D efficiency, and used tobit regression method to examine the relationship between direct government subsidies and preferential tax policies and manufacturing R&D efficiency. The results reveal that the overall R&D efficiency of China’s manufacturing industry was low, but it has been steadily increasing, and the R&D efficiency of emerging industries was significantly higher than that of traditional industries. Tax incentives played a stable and significant role in promoting R&D efficiency in manufacturing. Affected by factors such as the government’s long-term preference and information asymmetry, direct subsidies had no significant impact on the current R&D efficiency of the manufacturing industry, and began to play a positive role after a two-year lag. Based on the above research findings, this paper suggests that progressive preferential tax rates can be designed according to the “base + increment” approach for tax preferential policies. At the same time, different proportions of tax cuts should be set for enterprises of different sizes and levels of innovation, and the focus should be on small and medium-sized enterprises and emerging industries. In terms of direct funding subsidies, the government should not only increase the support for basic research, but also give more preference to enterprises that receive tax incentives for research and development, so as to enhance the complementary effect of the two types of subsidy policies. The marginal contribution of this paper mainly includes three aspects: First, based on the Chinese situation, the impact of direct government subsidies and tax incentives on the R&D efficiency of the manufacturing industry is tested. Second, we present the evidence that direct government funding subsidies “crowd out” enterprise R&D funds. Thirdly, we describe the influence of enterprise scale, innovation level, ownership, and management ability on R&D efficiency of the manufacturing industry, and put forward the possible influence mechanism.

A Graph Positional Attention Network for Session-Based Recommendation

The main idea of a session-based recommendation system is to model the user’s historical click sequence and then summarize user preferences and predict the items the user will interact with. The session recommendation model based on graph neural networks has attracted much attention in recent years because it can accurately obtain the local relationship between items. However, the traditional session recommendation model based on graph neural Networks lack the use of user’s higher-order features or fail to address the impact of item position information on the current session, which are both critical to the recommendation system. In addition, some models proposed the position information while neglects the click frequency information. We propose a graph network recommendation model called GPAN based on position attention in response to the abovementioned problems. Specifically, we propose a novel high-low order session perceptron that uses the perceptron to model undirected and directed graphs separately to obtain high and low order item representations in a session. For position information, we designed a position layer to calculate independently. Finally, the user’s short-term preference and long-term preference are aggregated to obtain the recommendation sequence. The results through a large number of experiments on three real datasets show that the performance of the proposed GPAN model is the best.

The Next Step: Current Sexual Lifestyles and Future Intentions

Using data from a nationally representative survey of sexual behavior in the United Kingdom, this visualization shows participants’ current sexual lifestyles and their ideal lifestyles five years from now. It demonstrates strong preferences for relationship progression toward cohabitation, marriage, and monogamy. Only a small minority of participants express a long-term preference for casual relationships, no sexual relationship, or any nonmonogamous lifestyle. Of currently nonmonogamous participants, a majority indicate a preference for a monogamous future, although a significant minority of nonmonogamous participants who are cohabiting or married wish to keep their lifestyles in five years’ time.

Motivating Personal Climate Action through a Safety and Health Risk Management Framework.

There is overwhelming evidence the impacts of climate change present a probable threat to personal health and safety. However, traditional risk management approaches have not been applied to ameliorate the crises. The purpose of this study was to assess the impact on personal motivation for action of a communication intervention that framed climate change as a safety issue that can be mitigated through a safety and health risk management framework. Participants' perception of climate change in terms of its anthropogenicity, context and importance, perception as a personal threat, belief in the efficacy of human action, motivating drivers for action, knowledge of climate change impacts, perceived personal barriers to climate action, and short- and long-term preferences for mitigating actions were evaluated. In addition, this study assessed the role of personal worldview on motivation for climate action. Through an online survey instrument embedded with a communication/education intervention, data were collected from N = 273 participants. Pre and post-intervention responses were assessed using Wilcoxon signed-rank tests and descriptive statistics. A path analysis assessed the influence of anthropogenicity, personal impact, and human efficacy beliefs on participant motivation for action. Multi-regression analyses and descriptive statics were used to evaluate the role of worldview on participant motivation for climate action. Personal motivation for action significantly increased post-intervention. Anthropogenicity, personal impact, and human efficacy beliefs were predictive of personal motivation. Those who prioritized climate change as a safety issue and those driven by a desire to protect current and future generations had higher levels of personal motivation, post-intervention. Knowledge of climate change increased, psychosocial factors as barriers to climate action decreased, and preferences for personal mitigating actions shifted towards more impactful choices post-intervention. Holding Egalitarian worldviews significantly predicted climate action motivation. Presenting climate change and climate action strategies via a traditional health and safety risk management context was effective in increasing personal motivation for climate action. This study contributes to the literature on climate change communication and climate action motivation.

A Ranking Recommendation Algorithm Based on Dynamic User Preference.

In recent years, hybrid recommendation techniques based on feature fusion have gained extensive attention in the field of list ranking. Most of them fuse linear and nonlinear models to simultaneously learn the linear and nonlinear features of entities and jointly fit user-item interactions. These methods are based on implicit feedback, which can reduce the difficulty of data collection and the time of data preprocessing, but will lead to the lack of entity interaction depth information due to the lack of user satisfaction. This is equivalent to artificially reducing the entity interaction features, limiting the overall performance of the model. To address this problem, we propose a two-stage recommendation model named A-DNR, short for Attention-based Deep Neural Ranking. In the first stage, user short-term preferences are modeled through an attention mechanism network. Then the user short-term preferences and user long-term preferences are fused into dynamic user preferences. In the second stage, the high-order and low-order feature interactions are modeled by a matrix factorization (MF) model and a multi-layer perceptron (MLP) model, respectively. Then, the features are fused through a fully connected layer, and the vectors are mapped to scores. Finally, a ranking list is output through the scores. Experiments on three real-world datasets (Movielens100K, Movielens1M and Yahoo Movies) show that our proposed model achieves significant improvements compared to existing methods.

TCSA-Net: A Temporal-Context-Based Self-Attention Network for Next Location Prediction

Next location prediction aims to find the location that the user will visit next. It plays a fundamental role for location-based applications. However, the heterogeneity and sparsity of the trajectory data pose great challenges to the task. Recently, RNN-based methods have shown promising performance in learining the spatio-temporal characteristics of the trajectory. While the effectiveness of location prediction has been improved, the computational efficiency and the long-term preferences still leave space for further research. The self-attention mechanism is viewed as a promising solution for parallel computation and exploiting sequential regularities from sparse data. But the huge memory cost and the neglect of temporal information make it infeasible to directly modeling human mobility regularities. In this paper, we propose a temporal-context-based self-attention network named TCSA-Net, which can simultaneously exploit long- and short-term mvoement preferences from sparse and long trajectories. In particular, we design a novel two-stage self-attention architecture that can learn long-term dependency under constrained memory budget. Further, we propose a multi-modal embedding layer to model two complementary temporal contexts and provide more abundant temporal and sequential information. Extensive experiments on two real-life datasets show that the TCSA-Net significantly outperforms the state-of-the-art methods in terms of standard evaluation metrics.

Interval-enhanced Graph Transformer solution for session-based recommendation

In many online recommendation services (e.g., multimedia streaming, e-commerce), predicting user’s next behavior based on anonymous sessions remains a challenging problem, mainly due to the lack of basic user information and limited behavioral information. The existing typical methods either model user behavior sequences based on RNN or capture potential relationships among items based on GNN. However, these pioneers ignore the importance of different time intervals in the behavior sequence, which implies the user preferences and makes the session sequence more distinguishable. Towards this end, we contribute an Interval-enhanced Graph Transformer (IGT) solution for the session-based recommendation, which takes both item relations and corresponding time intervals into consideration. Specifically, IGT consists of three modules: (i) Interval-enhanced session graph, which constructs all session sequences as session graphs with time intervals; (ii) Graph Transformer, which is embedded with time intervals is adopted to learn the complex interaction information among items. Among them, we design various time interval embedding functions, which can be flexibly injected into the framework; (iii) Preference representation and prediction, which uses an attention network to fuse the user’s long-term preferences and short-term preferences to predict the next click. By conducting extensive experiments on the DIGINETICA, YOOCHOOSE and Last.FM three real-world datasets, we validate that IGT outperforms state-of-the-art solutions.

A personalized recommendation method under the cloud platform based on users’ long-term preferences and instant interests

Rich consumer online text data are embedded in the cloud platform. Using new technologies has become a central issue for acquiring consumer preference, analyzing consumer demand, and performing personalized recommendation services. In order to recommend the cloud platform services efficiently and accurately, this paper proposes a personalized recommendation model referred to as Residual bi-directional Recurrent Neural Network with Dual Attentive mechanism (BiRDA) for the service recommend to cloud platforms, by combining users’ long-term preferences with instant interest. The proposed recommender prototype is summarized as follows. (1) Analyzing the relationship between long-term preferences and instant interests based on co-opetition theory. (2) Extracting users’ online text data from the cloud platform. (3) Deriving the product attribute words of user preference using an analysis of online text data. (4) Product attribute words are transformed into the form of word vectors. (5) The word vector is input into the Residual bi-directional Recurrent Neural Network (Res-BiRNN) to make the prediction. On the one hand, the long-term preference is expressed by the user's field of expertise (i.e., answer content). On the other hand, the even interest is expressed by the user's changing interest (i.e., question data). (6) Assigning different weights to long-term preferences and instant interest using the dual attention mechanism to output predictions. (7) Generating recommendation lists for users based on the predicted values. Accordingly, BiRDA is compared with five state-of-the-art recommendation methods (i.e., DREAM, BINN, SHAN, Caser, and DeepMove), as well as six variants of the BiRDA model, Using users’ Q&A datasets from NiorcngeCDS cloud platform, XMAKE cloud platform, and Asksubarme cloud platform as examples. The experiments show that the proposed method is more efficient and accurate than the other models. Therefore, the study offers some important insights into allowing a large number of resources under the cloud platform to be fully utilized and provides a novel idea for the construction of the cloud platform front-end.

Multisource Information Fusion Algorithm for Personalized Tourism Destination Recommendation

In this paper, the existing scenic spot recommendation algorithms ignore the implicit trust and trust transmission of users when dealing with user relationships, and the lack of historical browsing behavior data of users in new city scenes leads to an inaccurate recommendation. In this paper, a personalized scenic spot recommendation method combining user trust relationship and tag preference is proposed. Firstly, the trust degree is introduced when the recommendation quality is poor only considering the similarity of users. By mining the implicit trust relationship of users, the problem that the existing research cannot make recommendations when the direct trust is difficult to obtain is solved, and the data sparsity and cold start problems are effectively alleviated. Secondly, in the process of user interest analysis, the relationship between scenic spots and tags is extended to the relationship among users, scenic spots and tags, and users’ interest preferences are decomposed into long-term preferences for different scenic spots tags, which effectively alleviates the problem of poor recommendation quality when users’ historical tour records are lacking. The personalized tourism recommendation method proposed in this paper effectively integrates many features of social networks and effectively alleviates the problems of data sparseness and feature learning in tourism recommendation based on social networks by using vectorization and deep learning technology. Its research has very important usage scenarios and commercial value in the tourism industry. This model can efficiently mine the association rules between scenic spots in multisource information data. The experimental results show that mining the correlation between the scenic spots selected by tourists can provide effective information for tourism decision-making.