Cold Start Issues Research Articles

A Hierarchical Attention Recommender System Based on Cross-Domain Social Networks

Search engines and recommendation systems are an essential means of solving information overload, and recommendation algorithms are the core of recommendation systems. Recently, the recommendation algorithm of graph neural network based on social network has greatly improved the quality of the recommendation system. However, these methods paid far too little attention to the heterogeneity of social networks. Indeed, ignoring the heterogeneity of connections between users and interactions between users and items may seriously affect user representation. In this paper, we propose a hierarchical attention recommendation system (HA-RS) based on mask social network, combining social network information and user behavior information, which improves not only the accuracy of recommendation but also the flexibility of the network. First, learning the node representation in the item domain through the proposed Context-NE model and then the feature information of neighbor nodes in social domain is aggregated through the hierarchical attention network. It can fuse the information in the heterogeneous network (social domain and item domain) through the above two steps. We propose the mask mechanism to solve the cold-start issues for users and items by randomly masking some nodes in the item domain and in the social domain during the training process. Comprehensive experiments on four real-world datasets show the effectiveness of the proposed method.

Heterogeneous teaching evaluation network based offline course recommendation with graph learning and tensor factorization

Course recommendation systems are applied to help students with different needs select courses in a large range of course resources. However, a student’s needs are not always determined by their personal interests, they are also influenced by teachers, peers etc. Unlike online courses, user behavior and user satisfaction of offline courses often have serious sparse and cold start issues, which cause overfitting problems in previous neural network and matrix factorization (MF) models. Additionally, the interpersonal relations, evaluation text and existing “user-item” formatted rating matrix constitute a multi-source and multi-modal data structure, so a systematic data fusion method is needed to establish recommendations based on these heterogeneous characteristics. Therefore, a hybrid recommendation model by fusing network structured feature with graph neural networks and user interactive activities with tensor factorization was proposed in this paper. First, a graph structured teaching evaluation network is proposed to describe students, courses, and other entities by using the students’ rating, commentary text, grading and interpersonal relations. Then, a random walk based neural network is employed to generate the vectorized representation of students by learning their own relational structure. Finally, by recognizing these personalization features as the third dimension of the rating tensor, a Bayesian Probabilistic Tensor Factorization-based tensor factorization is applied to learn and predict students’ ratings for classes they have not taken. Experiments on a real-world evaluation of teaching system including 532 participants with 7,453 rating records show that the proposed method outperforms other existing neural network and matrix factorization models including xSVD++, RTTF and DSE with a smaller predictive error as well as better recommendation accuracy.

SoftRec: Multi-Relationship Fused Software Developer Recommendation

Collaboration efficiency is of primary importance in software development. It is widely recognized that choosing suitable developers is an efficient and effective practice for improving the efficiency of software development and collaboration. Recommending suitable developers is complex and time-consuming due to the difficulty of learning developers’ expertise and willingness. Existing works focus on learning developers’ expertise and interactions from their explicit historical information and matching them to specific task. However, such procedures may suffer low accuracy because they ignore implicit information, such as (1) developer–developer collaboration relationships, (2) developer–task implicit interaction relationships, and (3) task–task association relationships, etc. To that end, this paper proposes a multi-relationship fused approach for software developer recommendation (termed SoftRec). First, in addition to explicit developer–task interactions, it considers multivariate implicit relationships, including the three types mentioned above. Second, it integrates these relationships based on joint matrix factorization and generates forecast results upon the architecture of deep neural network. Furthermore, we propose a fast update method to address the cold start issue by making online recommendations for new developers and new tasks. Extensive experiments are conducted on two real-world datasets, and a user study is conducted in a well-known software company. The results demonstrate that SoftRec outperforms four state-of-the-art works.

Alleviating New User Cold-Start in User-Based Collaborative Filtering via Bipartite Network

The recommender system (RS) can help us extract valuable data from a huge amount of raw information. User-based collaborative filtering (UBCF) is widely employed in practical RSs due to its outstanding performance. However, the traditional UBCF is subject to the new user cold-start issue because a new user is often extreme lack of available rating information. In this article, we develop a novel approach that incorporates a bipartite network into UBCF for enhancing the recommendation quality of new users. First, through the statistic and analysis of new users’ rating characteristics, we collect niche items and map the corresponding rating matrix to a weighted bipartite network. Furthermore, a new weighted bipartite modularity index merging normalized rating information is present to conduct the community partition that realizes coclustering of users and items. Finally, for each individual clustering that is much smaller than the original rating matrix, a localized low-rank matrix factorization is executed to predict rating scores for unrated items. Items with the highest predicted rating scores are recommended to a new user. Experimental results from two real-world data sets suggest that without requiring additional complex information, the proposed approach is superior in terms of both recommendation accuracy and diversity and can alleviate the new user cold-start issue of UBCF effectively.

Addressing the Item Cold-Start Problem by Attribute-Driven Active Learning

In recommender systems, cold-start issues are situations where no previous events, e.g. ratings, are known for certain users or items. In this paper, we focus on the item cold-start problem. Both content information (e.g. item attributes) and initial user ratings are valuable for seizing users' preferences on a new item. However, previous methods for the item cold-start problem either 1) incorporate content information into collaborative filtering to perform hybrid recommendation, or 2) actively select users to rate the new item without considering content information and then do collaborative filtering. In this paper, we propose a novel recommendation scheme for the item cold-start problem by leverage both active learning and items' attribute information. Specifically, we design useful user selection criteria based on items' attributes and users' rating history, and combine the criteria in an optimization framework for selecting users. By exploiting the feedback ratings, users' previous ratings and items' attributes, we then generate accurate rating predictions for the other unselected users. Experimental results on two real-world datasets show the superiority of our proposed method over traditional methods.

Genetic Algorithm Influenced Top-N Recommender System to Alleviate New User Cold Start Problem

Most recommender systems are based on the familiar collaborative filtering algorithm to suggest items. Quite often, collaborative filtering algorithm fails in generating recommendations due to the lack of adequate user information resulting in new user cold start problem. The cold start problem is one among the prevailing issue in recommendation system where the system fails to render recommendations. To overcome the new user cold start issue, demographical information of the user is utilised as the user information source. Among the demographical information, the impact of the user gender is less explored when compared with other information like age, profession, region, etc. In this work, a genetic algorithm-influenced gender-based top-n recommender algorithm is proposed to address the new user cold start problem. The algorithm utilises the evolution concepts of the genetic algorithm to render top-n recommendations to a new user. The evaluation of the proposed algorithm using real world datasets proved that the algorithm has a better efficiency than the state of art approaches.

Deep Representation Learning for Location-Based Recommendation

Location-based recommendation has recently received a lot of attention in the communities of information service and mobile application. Its task is to provide personalized recommendations of points of interest (POIs) to users at a certain time and location. However, existing location-based recommendation models have at least two main drawbacks: first they cannot adequately capture semantic features of POIs and users, which may lead to unsatisfactory recommendations and second they cannot effectively address the cold-start problem. To address the above drawbacks, in this article, we first propose a novel deep representation learning-based model (DRLM) for improving the recommendation accuracy. In DRLM, we mainly focus on learning to accurately represent semantic features of POIs and users. Specifically, four co-occurrence matrices are constructed to produce four different original features for each POI, and a principal component analysis (PCA) algorithm is utilized to generate a semantic feature of each POI from its four original features. On the other hand, a three-modal simple recurrent unit (TMSRU) network is given to constructed semantic features of users using semantic features of POIs, times, and locations. We further propose minimum description length (MDL)-based and skyline-based strategies to address the cold-start issues for new users and new POIs, respectively. Through experiments on two real-world data sets, we show that compared with the state-of-the-art approaches, the proposed model DRLM can achieve the superior performance in terms of high recommendation accuracy and effectiveness in handling the cold-start problem.

A personalized search model using online social network data based on a holonic multiagent system

Personalized search utilizes user preferences to optimize search results, and most existing studies obtain user preferences by analyzing user behaviors in search engines that provide click-through data. However, the behavioral data are noisy because users often clicked some irrelevant documents to find their required information, and the new user cold start issue represents a serious problem, greatly reducing the performance of personalized search. This paper attempts to utilize online social network data to obtain user preferences that can be used to personalize search results, mine the knowledge of user interests, user influence and user relationships from online social networks, and use this knowledge to optimize the results returned by search engines. The proposed model is based on a holonic multiagent system that improves the adaptability and scalability of the model. The experimental results show that utilizing online social network data to implement personalized search is feasible and that online social network data are significant for personalized search.

Trust-embedded collaborative deep generative model for social recommendation

Social networks can provide massive amounts of information for communication among users and communities. The trust relationships in social networks can be utilized to reveal user preferences for improving the quality of social recommendation, which aims to mitigate information overload and provide users with the most attractive and relevant items or services. However, the data sparsity and cold-start issue degrade recommendation performance significantly. To address these issues, a novel trust-embedded collaborative deep generative model (TCDG) is proposed for exploiting multisource information (content, rating and trust) to predict ratings. TCDG employs deep generative model to unsupervisedly learn deep latent representations for item content through an inference network in latent space instead of observation space. Meanwhile, TCDG adopts probabilistic matrix factorization to map users into low-dimensional latent feature spaces by trust relationships, which can reflect users’ mutual influence on the formation of users’ opinions more accurately and learn better implicit relationships between items and users from content, rating and trust. In addition, an approach with an annealing parameter to calculate the maximum a posteriori estimates is proposed to learn model parameters. Experiments using four real-world datasets are conducted to evaluate the prediction and top-ranking performance of our model. The results indicate that TDCG has better accuracy and robustness than other methods for making recommendations.

Resolving data sparsity and cold start problem in collaborative filtering recommender system using Linked Open Data

The web contains a huge volume of data, and it's populating every moment to the point that human beings cannot deal with the vast amount of data manually or via traditional tools. Hence an advanced tool is required to filter such massive data and mine the valuable information. Recommender systems are among the most excellent tools for such a purpose in which collaborative filtering is widely used. Collaborative filtering (CF) has been extensively utilized to offer personalized recommendations in electronic business and social network websites. In that, matrix factorization is an efficient technique; however, it depends on past transactions of the users. Hence, there will be a data sparsity problem. Another issue with the collaborative filtering method is the cold start issue, which is due to the deficient information about new entities. A novel method is proposed to overcome the data sparsity and the cold start problem in CF. For cold start issue, Recommender System with Linked Open Data (RS-LOD) model is designed and for data sparsity problem, Matrix Factorization model with Linked Open Data is developed (MF-LOD). A LOD knowledge base “DBpedia” is used to find enough information about new entities for a cold start issue, and an improvement is made on the matrix factorization model to handle data sparsity. Experiments were done on Netflix and MovieLens datasets show that our proposed techniques are superior to other existing methods, which mean recommendation accuracy is improved.

Thermal stability analysis of cold start processes in PEM fuel cells

Cold-start issue of proton exchange membrane (PEM) fuel cells is one of the major factors hindering its commercialization. The key element involved is the competition between ice formation and melting, behind which is the thermal behavior accompanied by nonlinear effects. In this work, the steady-state multiplicity feature during cold-start has been discovered based on the thermal stability analysis approach. The nonlinear heat generation term Qg and linear removal term Qr are separated from conservation equations and constructed against ice fraction. The plot thus obtained could identify the existence of multiplicity which is the intersection of the above two functions. Results indicate that one or two steady states arise for different operating conditions. The startup with only one steady state, namely the “extinguished” state is the worst condition and bound to fail. The startup with two steady states also incorporates an unstable one which denotes not only the critical transition from ice formation to melting but also the limit of the operational domain. Based on the theoretical analysis, simple and explicit criteria are quantitatively developed for the prediction of startup feasibility. Moreover, in order to ensure a reasonable and efficient operation, the impacts of key parameters are further summarized and discussed on the cold-start operability. The proposed approach that develops analytical expressions for the steady-state points provides a systematic yet simple way to reveal the essential physics of cold start.

Multi-criteria clustering-based recommendation using Mahalanobis distance

There have been significant advances made in the research of recommender systems over the past decades and have been implemented in both industry and academia. Recently, multi-criteria ratings are being incorporated into traditional recommender systems to further improve their quality, especially to handle the data sparsity and cold start issues. However, incorporation of multi-criteria ratings have improved the performance of the recommendation, but at the same time, multi-dimensionality issue also arises. This paper presents a clustering-based recommendation approach which is used for dealing with the multi-dimensionality issue in multi-criteria recommender systems. Here, we cluster the users based on their individual criteria ratings using K-means clustering and the intra-cluster similarity is computed using Mahalanobis distance measure for neighbourhood set generation. This improves the recommendations quality and predictive accuracy of both traditional and clusteringbased collaborative recommendations. The Yahoo! Movies dataset was used for testing the approach and the experiment conducted shows promising results.

Multi-criteria clustering-based recommendation using Mahalanobis distance

There have been significant advances made in the research of recommender systems over the past decades and have been implemented in both industry and academia. Recently, multi-criteria ratings are being incorporated into traditional recommender systems to further improve their quality, especially to handle the data sparsity and cold start issues. However, incorporation of multi-criteria ratings have improved the performance of the recommendation, but at the same time, multi-dimensionality issue also arises. This paper presents a clustering-based recommendation approach which is used for dealing with the multi-dimensionality issue in multi-criteria recommender systems. Here, we cluster the users based on their individual criteria ratings using K-means clustering and the intra-cluster similarity is computed using Mahalanobis distance measure for neighbourhood set generation. This improves the recommendations quality and predictive accuracy of both traditional and clusteringbased collaborative recommendations. The Yahoo! Movies dataset was used for testing the approach and the experiment conducted shows promising results.

A hybrid model collaborative movie recommendation system using K-means clustering with ant colony optimisation

Movie recommendation system offers a mechanism to allocate the user to attain the famous film by getting an opinion from similar users or past rating by user. This produces recommender systems has a crucial part of website and e-commerce application. The primary objective of the system to prefer a recommender system by data clustering and computational intelligence. We proposed a hybrid model collaborative movie recommendation system that performs with a combination of K-means clustering with ant colony optimisation technique (ACO-KM) that has employed in movie dataset. The proposed system compared with existing works, and its efforts have been analysed. The evaluation process of movie recommendation system that offers improved result from ACO-KM collaborative movie recommender system based on precision, recall, mean square error (MSE), and accuracy. By comparison of speed (in seconds) of various approaches in Movielens dataset, our approach gives best result 42.24 compared with existing one 53.22. The outcome of this experiment from Movielens dataset that offers scalability and efficiency in a recommendation by decreasing cold start issues.

Joint Personalized Markov Chains with social network embedding for cold-start recommendation

The primary objective of recommender systems is to help users select their desired items, where a key challenge is providing high-quality recommendations to users in a “cold-start” situation. Recent advances in tackling this problem combine social relations and temporal information and integrate them into a unified framework. However, these methods suffer from a limitation that there not always exist links for the newcomers, thus these users are filtered in related studies. To break the boundary, in this paper, we propose a Joint Personalized Markov Chains (JPMC) model to address the cold-start issues for implicit feedback recommendation system. In our study, we first utilize user embedding to mine Network Neighbors, so that newcomers without relations can be represented by similar users, then we designed a two-level model based on Markov chains at both user level and user group level respectively to model user preferences dynamically. Experimental results on three real-world datasets have shown that our model can significantly outperform the state-of-the-art models.

Modeling hierarchical category transition for next POI recommendation with uncertain check-ins

Most existing next POI recommendation studies rely on users’ certain check-ins at individual POIs (e.g., Italian restaurant). In reality, users may leave some uncertain check-ins in the places (e.g., shopping mall), which are named as collective POIs. It indicates that we cannot always access users’ precise check-ins at collective POIs, thus existing approaches fail to work well. To this end, we propose a new research problem, that aims to recommend next individual POIs with uncertain check-ins at collective POIs. It is, however, difficult to learn complete and accurate users’ check-in transition patterns with uncertain check-ins. Besides, uncertain check-ins aggravate the cold start issue, as the individual POIs inside collective POIs cannot be observed in users’ historical check-ins. To tackle these challenges, we devise a novel hierarchical category transition (HCT) framework, which exploits category transitions at different layers to model users’ preference transition patterns in different granularity. By doing so, HCT predicts users’ preferred categories inside collective POIs. As bounded to specific categories, HCT further adopts hierarchical dependencies between POIs and categories to capture the semantic relatedness of POIs, thus easing the cold start issue. Empirical studies on multiple datasets show the superiority of HCT against state-of-the-art algorithms.

Reinforcement Learning Based Query Routing Approach for P2P Systems

Peer-to-peer (P2P) systems have offered users an efficient way to share various resources and access diverse services over the Internet. In unstructured P2P systems, resource storage and indexation are fully distributed among participating peers. Therefore, locating peers sharing pertinent resources for a specific user query is a challenging issue. In fact, effective query routing requires smart decisions to select a certain number of peers with respect to their relevance for the query instead of choosing them at random. In this respect, we introduce here a new query-oriented approach, called the reinforcement learning-based query routing approach (RLQR). The main goal of RLQR is to reach high retrieval effectiveness as well as a lower search cost by reducing the number of exchanged messages and contacted peers. To achieve this, the RLQR relies on information gathered from previously sent queries to identify relevant peers for forthcoming queries. Indeed, we formulate the query routing issue as the reinforcement learning problem and introduce a fully distributed approach for addressing it. In addition, RLQR addresses the well-known cold-start issue during the training stage, which allows it to improve its retrieval effectiveness and search cost continuously, and, therefore, goes quickly through the cold-start phase. Performed simulations demonstrate that RLQR outperforms pioneering query routing approaches in terms of retrieval effectiveness and communications cost.

Integrating collaborative topic modeling and diversity for movie recommendations during news browsing

PurposeOnline news websites provide diverse article topics, such as fashion news, entertainment and movie information, to attract more users and create more benefits. Recommending movie information to users reading news online can enhance the impression of diverse information and may consequently improve benefits. Accordingly, providing online movie recommendations can improve users’ satisfactions with the website, and thus is an important trend for online news websites. This study aims to propose a novel online recommendation method for recommending movie information to users when they are browsing news articles.Design/methodology/approachAssociation rule mining is applied to users’ news and movie browsing to find latent associations between news and movies. A novel online recommendation approach is proposed based on latent Dirichlet allocation (LDA), enhanced collaborative topic modeling (ECTM) and the diversity of recommendations. The performance of proposed approach is evaluated via an online evaluation on a real news website.FindingsThe online evaluation results show that the click-through rate can be improved by the proposed hybrid method integrating recommendation diversity, LDA, ECTM and users’ online interests, which are adapted to the current browsing news. The experiment results also show that considering recommendation diversity can achieve better performance.Originality/valueExisting studies had not investigated the problem of recommending movie information to users while they are reading news online. To address this problem, a novel hybrid recommendation method is proposed for dealing with cross-type recommendation tasks and the cold-start issue. Moreover, the proposed method is implemented and evaluated online in a real world news website, while such online evaluation is rarely conducted in related research. This work contributes to deriving user’s online preferences for cross-type recommendations by integrating recommendation diversity, LDA, ECTM and adaptive online interests. The research findings also contribute to increasing the commercial value of the online news websites.

POLAR++: Active One-Shot Personalized Article Recommendation

We study the problem of personalized article recommendation, in particular when the user's preference data is missing or limited, which is knowns as the user cold-start issue in recommender systems. We propose POLAR++, an active recommendation framework that utilizes Bayesian neural networks to capture the uncertainty of user preference, actively selects articles to query the user for feedback, and adaptively learns user preference with one-shot learning. For the article recommendation, we design an attention-based CNN to quantify the similarity between user preference and recommended articles, which significantly improves the performance with only a few articles rated by the users. We evaluate the proposed POLAR++ on datasets of different scale and sources. Experimental results demonstrate the effectiveness of the proposed model. We have successfully deployed POLAR++ into AMiner as the recommendation engine for article recommendation, which further confirms the effectiveness of the proposed model.

Feasibility study on a vehicular thermoelectric generator coupled to an exhaust gas heater to improve aftertreatment’s efficiency in cold-starts

Low exhaust gas temperatures of internal combustion engines, resulting from cold-starts, low loads or low temperature ambient conditions, can compromise the performance of the automotive aftertreatment system. In this situation, the engine is running far from its design point, with a very low performance of the catalytic converter since it has not reached its optimal temperature. This effect can result in more than 20% of the total emissions for nitrogen oxides (NOx), hydrocarbons (HC) and carbon monoxide (CO) during a driving cycle.This study presents a new approach for vehicular thermoelectric generators to reduce NOx, HC and CO emissions and solve the cold-start issue. Authors propose the use of a thermoelectric generator coupled to an exhaust gas heater. This system, that operates autonomously, is designed to transform the waste heat into electric heat that can quickly raise the exhaust gas temperature to the optimal level.Results show that an improvement up to 94% of the mean catalyst’s efficiency can be achieved. In addition, the system could reduce up to 94%, 91%, and 97% the pollutant emissions of NOx, HC and CO, respectively, in a WLTC driving cycle with a threshold exhaust gas temperature of 350 °C.