Search results

Patent trade recommendations necessitate recommendation interpretability in addition to recommendation accuracy because of patent transaction risks and the technological complexity of patents. This study designs an interpretable knowledge-aware patent recommendation model (IKPRM) for patent trading. IKPRM first creates a patent knowledge graph (PKG) for patent trade recommendations and then leverages paths in the PKG to achieve recommendation interpretability.

First, we construct a PKG to integrate online company behaviors and patent information using natural language processing techniques. Second, a bidirectional long short-term memory network (BiLSTM) is utilized with an attention mechanism to establish the connecting paths of a company — patent pair in PKG. Finally, the prediction score of a company — patent pair is calculated by assigning different weights to their connecting paths. The semantic relationships in connecting paths help explain why a candidate patent is recommended.

Experiments on a real dataset from a patent trading platform verify that IKPRM significantly outperforms baseline methods in terms of hit ratio and normalized discounted cumulative gain (nDCG). The analysis of an online user study verified the interpretability of our recommendations.

A meta-path-based recommendation can achieve certain explainability but suffers from low flexibility when reasoning on heterogeneous information. To bridge this gap, we propose the IKPRM to explain the full paths in the knowledge graph. IKPRM demonstrates good performance and transparency and is a solid foundation for integrating interpretable artificial intelligence into complex tasks such as intelligent recommendations.

Tourism spot or point-of-interest (POI) recommendation has become a common service in people's daily life. The purpose of this paper is to model users' check-in history in order to predict a set of locations that a user may soon visit.

The authors proposed a novel learning-based method, the pattern-based dual learning POI recommendation system as a solution to consider users' interests and the uniformity of popular POI patterns when making recommendations. Differing from traditional long short-term memory (LSTM), a new users’ regularity–POIs’ popularity patterns long short-term memory (UP-LSTM) model was developed to concurrently combine the behaviors of a specific user and common users.

The authors introduced the concept of dual learning for POI recommendation. Several performance evaluations were conducted on real-life mobility data sets to demonstrate the effectiveness and practicability of POI recommendations. The metrics such as hit rate, precision, recall and F-measure were used to measure the capability of ranking and precise prediction of the proposed model over all baselines. The experimental results indicated that the proposed UP-LSTM model consistently outperformed the state-of-the-art models in all metrics by a large margin.

This study contributes to the existing literature by incorporating a novel pattern–based technique to analyze how the popularity of POIs affects the next move of a particular user. Also, the authors have proposed an effective fusing scheme to boost the prediction performance in the proposed UP-LSTM model. The experimental results and discussions indicate that the combination of the user's regularity and the POIs’ popularity patterns in PDLRec could significantly enhance the performance of POI recommendation.

The purpose of this paper is to introduce an improved nearest‐neighbor collaborative filtering algorithm based on rough set theory to alleviate the sparsity problem of collaborative filtering. With experimentations, the new algorithm is thereafter evaluated.

Nearest‐neighbor algorithm is the earliest proposed and the main collaborative filtering recommendation algorithm, and its recommendation quality is seriously influenced by the sparsity of user ratings. By using rough set theory, the nearest‐neighbor collaborative filtering algorithm can be improved in the sparsity data situation. The union of user rating items is used as the basis of similarity computing among users, and then a rating predicting method based on rough set theory is proposed to estimate missing values in the union of user rating items for decreasing sparsity.

The sparsity problem of collaborative filtering can be alleviated by using the union of user rating items and estimating missing values based on rough set theory. The experimental results show that the new algorithm can efficiently improve recommendation quality of collaborative filtering.

The union of user rating items was used as the basis of similarity computing among users. A rating prediction method based on rough set theory with an assistant method was proposed to complete the missing values in the union of user rating items. Orthogonal list was used to storage user‐item ratings matrix.

Aspects extracted from the user’s historical records are widely used to define user’s fine-grained preferences for building interpretable recommendation systems. As the aspects were extracted from the historical records, the aspects that represent user’s negative preferences cannot be identified because of their absence from the records. However, these latent aspects are also as important as those aspects representing user’s positive preferences for building a recommendation system. This paper aims to identify the user’s positive preferences and negative preferences for building an interpretable recommendation.

First, high-frequency tags are selected as aspects to describe user preferences in aspect-level. Second, user positive and negative preferences are calculated according to the positive and negative preference model, and the interaction between similar aspects is adopted to address the aspect sparsity problem. Finally, an experiment is designed to evaluate the effectiveness of the model. The code and the experiment data link is: https://github.com/shiyu108/Recommendation-system

Experimental results show the proposed approach outperformed the state-of-the-art methods in widely used public data sets. These latent aspects are also as important as those aspects representing the user’s positive preferences for building a recommendation system.

This paper provides a new approach that identifies and uses not only users’ positive preferences but also negative preferences, which can capture user preference precisely. Besides, the proposed model provides good interpretability.

Due to the rapidly increasing volume of users and products in virtual worlds, recommender systems are an important feature in virtual worlds; they can help solve information overload problems. Virtual world users are able to perform several actions that promote the enjoyment of their virtual life, including interacting with others, visiting virtual houses and shopping for virtual products. This study aims to concentrate on the following two important factors: the social neighbors’ influences and the virtual house bandwagon phenomenon, which affects users’ preferences during their virtual house visits and purchasing processes.

The authors determine social influence by considering the interactions between the target user and social circle neighbors. The degree of influence of the virtual house bandwagon effect is derived by analyzing the preferences of the virtual house hosts who have been visited by target users during their successive visits. A novel hybrid recommendation method is proposed herein to predict users’ preferences by combining the analyses of both factors.

The recommendation performance of the proposed method is evaluated by conducting experiments with a data set collected from a virtual world platform. The experimental results show that the proposed method outperforms the conventional recommendation methods, and they also exhibit the effectiveness of considering both the social influence and the virtual house bandwagon effect for making effective recommendations.

Existing studies on recommendation methods did not investigate the virtual house bandwagon effects that are unique to the virtual worlds. The novel idea of the virtual house bandwagon effect is proposed and analyzed for predicting users’ preferences. Moreover, a novel hybrid recommendation approach is proposed herein for generating virtual product recommendations. The proposed approach is able to improve the accuracy of preference predictions and enhance the innovative value of recommender systems for virtual worlds.

The purpose of this paper is to alleviate the problem of poor robustness and over-fitting caused by large-scale data in collaborative filtering recommendation algorithms.

Interpreting user behavior from the probabilistic perspective of hidden variables is helpful to improve robustness and over-fitting problems. Constructing a recommendation network by variational inference can effectively solve the complex distribution calculation in the probabilistic recommendation model. Based on the aforementioned analysis, this paper uses variational auto-encoder to construct a generating network, which can restore user-rating data to solve the problem of poor robustness and over-fitting caused by large-scale data. Meanwhile, for the existing KL-vanishing problem in the variational inference deep learning model, this paper optimizes the model by the KL annealing and Free Bits methods.

The effect of the basic model is considerably improved after using the KL annealing or Free Bits method to solve KL vanishing. The proposed models evidently perform worse than competitors on small data sets, such as MovieLens 1 M. By contrast, they have better effects on large data sets such as MovieLens 10 M and MovieLens 20 M.

This paper presents the usage of the variational inference model for collaborative filtering recommendation and introduces the KL annealing and Free Bits methods to improve the basic model effect. Because the variational inference training denotes the probability distribution of the hidden vector, the problem of poor robustness and overfitting is alleviated. When the amount of data is relatively large in the actual application scenario, the probability distribution of the fitted actual data can better represent the user and the item. Therefore, using variational inference for collaborative filtering recommendation is of practical value.

In a multi-stakeholder recommender system (MSRS), stakeholders are the multiple entities (consumer, producer, system, etc.) benefited by the generated recommendations. Traditionally, the exclusive focus on only a single stakeholders' (for example, only consumer or end-user) preferences obscured the welfare of the others. Two major challenges are encountered while incorporating the multiple stakeholders' perspectives in MSRS: designing a dedicated utility function for each stakeholder and optimizing their utility without hurting others. This paper proposes multiple utility functions for different stakeholders and optimizes these functions for generating balanced, personalized recommendations for each stakeholder.

The proposed methodology considers four valid stakeholders user, producer, cast and recommender system from the multi-stakeholder recommender setting and builds dedicated utility functions. The utility function for users incorporates enhanced side-information-based similarity computation for utility count. Similarly, to improve the utility gain, the authors design new utility functions for producer, star-cast and system to incorporate long-tail and diverse items in the recommendation list. Next, to balance the utility gain and generate the trade-off recommendation solution, the authors perform the evolutionary optimization of the conflicting utility functions using NSGA-II. Experimental evaluation and comparison are conducted over three benchmark data sets.

The authors observed 19.70% of average enhancement in utility gain with improved mean precision, diversity and novelty. Exposure, hit, reach and target reach metrics are substantially improved.

A new approach considers four stakeholders simultaneously with their respective utility functions and establishes the trade-off recommendation solution between conflicting utilities of the stakeholders.

In a large‐scale digital library, it is essential to recommend a small number of useful and related articles to users. In this paper, a literature recommendation framework for digital libraries is proposed that dynamically provides recommendations to an active user when browsing a new article. This framework extends our previous work that considers only Web usage data by utilizing content information of articles when making recommendations. Methods that make use of pure content data, pure Web usage data, and both content and usage data are developed and compared using the data collected from our university's electronic thesis and dissertation (ETD) system. The experimental results demonstrate that content data and usage data are complements of each other and hybrid methods that take into account of both types of information tend to achieve more accurate recommendations.

The purpose of this paper is to provide adaptive access to learning resources in the digital library.

A novel method using ontology-based multi-attribute collaborative filtering is proposed. Digital libraries are those which are fully automated and all resources are in digital form and access to the information available is provided to a remote user as well as a conventional user electronically. To satisfy users' information needs, a humongous amount of newly created information is published electronically in digital libraries. While search applications are improving, it is still difficult for the majority of users to find relevant information. For better service, the framework should also be able to adapt queries to search domains and target learners.

This paper improves the accuracy and efficiency of predicting and recommending personalized learning resources in digital libraries. To facilitate a personalized digital learning environment, the authors propose a novel method using ontology-supported collaborative filtering (CF) recommendation system. The objective is to provide adaptive access to learning resources in the digital library. The proposed model is based on user-based CF which suggests learning resources for students based on their course registration, preferences for topics and digital libraries. Using ontological framework knowledge for semantic similarity and considering multiple attributes apart from learners' preferences for the learning resources improve the accuracy of the proposed model.

The results of this work majorly rely on the developed ontology. More experiments are to be conducted with other domain ontologies.

The proposed approach is integrated into Nucleus, a Learning Management System (https://nucleus.amcspsgtech.in). The results are of interest to learners, academicians, researchers and developers of digital libraries. This work also provides insights into the ontology for e-learning to improve personalized learning environments.

This paper computes learner similarity and learning resources similarity based on ontological knowledge, feedback and ratings on the learning resources. The predictions for the target learner are calculated and top N learning resources are generated by the recommendation engine using CF.

The purpose of this paper is to develop a novel and flexible recommender system based on usage patterns and keyword preferences using collaborative filtering (CF) and content‐based filtering (CBF).

The proposed system analyzes data captured from the navigational and behavioral patterns of users and estimates the popularity and similarity levels of a user's clicked content. Based on this information, content is recommended to each user using recommendation methods such as CF and CBF. To assess the effectiveness of the proposed approach, an empirical study was conducted by constructing an experimental news site.

The results of the experimental study clearly show that the proposed hybrid method is superior to conventional methods that use only CF or CBF.

The above findings are based on data captured from a relatively small experimental site, and they require further verification using various actual content sites. A promising area for future research may be the application of the proposed approach to making recommendations in user‐created content environments, such as blog sites and video upload sites, where users can actively participate as both writers and readers.

Unlike the most research on recommender systems, this is the first study to analyze user usage patterns and thereby determine appropriate recommendation algorithms for each user. The proposed recommender system provides greater prediction accuracy than conventional systems.