Recommender System Algorithms Research Articles

Opening the Black Box: Explaining the Process of Basing a Health Recommender System on the I-Change Behavioral Change Model

Recommender systems are gaining traction in healthcare because they can tailor recommendations based on users' feedback concerning their appreciation of previous health-related messages. However, recommender systems are often not grounded in behavioral change theories, which may further increase the effectiveness of their recommendations. This paper's objective is to describe principles for designing and developing a health recommender system grounded in the I-Change behavioral change model that shall be implemented through a mobile app for a smoking cessation support clinical trial. We built upon an existing smoking cessation health recommender system that delivered motivational messages through a mobile app. A group of experts assessed how the system may be improved to address the behavioral change determinants of the I-Change behavioral change model. The resulting system features a hybrid recommender algorithm for computer tailoring smoking cessation messages. A total of 331 different motivational messages were designed using 10 health communication methods. The algorithm was designed to match 58 message characteristics to each user profile by following the principles of the I-Change model and maintaining the benefits of the recommender system algorithms. The mobile app resulted in a streamlined version that aimed to improve the user experience, and this system's design bridges the gap between health recommender systems and the use of behavioral change theories. This article presents a novel approach integrating recommender system technology, health behavior technology, and computer-tailored technology. Future researchers will be able to build upon the principles applied in this case study.

Augmenting content-based rating prediction with link stream features

While graph-based collaborative filtering recommender systems have been introduced several years ago, there are still several shortcomings to deal with, the temporal information being one of the most important. The new link stream paradigm is aiming at extending graphs for correctly modelling the graph dynamics, without losing crucial information. We investigate the impact of such link stream features for recommender systems. We design link stream features, that capture the intrinsic structure and dynamics of the data. We show that such features encode a fine-grained and subtle description of the underlying system. We focused on a traditional recommender system context, the rating prediction on the MovieLens20M movie dataset and the Goodreads book dataset. We input link stream features along with some content-based ones into a gradient boosting machine (XGBoost) and show that itoutperforms significantly a sole content-based solution. These encouraging results call for further exploration of this original modelling and its integration to complete state-of-the-art recommender systems algorithms. Link streams and graphs, as natural visualizations of recommender systems, may offer more interpretability in a time when algorithm transparency is an increasingly important topic of discussion. We also hope that these results will sparkle interesting discussions in the community about the connections between link streams and traditional methods (matrix factorization, deep learning).

Cultural Distance awared Collaborative Filtering Algorithm in Recommendation System

Recently most of the existing studies on recommendation system are based on historical rating matrix and specific characteristics, such as location, season and weather. Different from these studies, we introduce an abstract feature about cultural backgrounds and values, called cultural distance, into the recommendation system to understand user intent better and improve the precision of recommendation results. We design a novel similarity representation which combine the item-based collaborative filtering and cultural distance to recommend items for users. We also propose a collaborative filtering-based missing cultural distance prediction algorithm to improve the precision of recommendation further. To evaluate the performance of our proposed algorithm, we execute experiments based on a large-scale real-world dataset, the results show that our algorithm can improve the precision by 10% accurate compared to existing recommendation approaches.

Molecular Characterization of Acute Myeloid Leukemia Patients with Normal Karyotype

Background Conventional cytogenetic classification remains one of the most important prognostic factors in acute myeloid leukemia (AML). Approximately 50-60 % of patients (pts) with AML have normal karyotype (NK). NK status has traditionally been associated with intermediate risk AML, but actually represents a heterogeneous group with variable outcomes. Adding mutational data such as NPM1, FLT3-ITD, and CEBPa can improve risk stratification for a subset of pts but does not reflect the genomic complexity and mutation interactions that may impact the overall outcome. In this study, we evaluated the association among several mutations and overall survival (OS) in pts with NK AML using unbiased advanced analytics approaches. Method Genomic and clinical data of 2793 primary AML (pAML) pts were analyzed. A panel of 35 genes that are commonly mutated in AML and myeloid malignancies was included. OS was calculated from the time of diagnosis to time of death or last follow up. To study the association of mutations and OS using an unbiased approach, we applied several machine learning algorithms that included random survival forest and recommender system algorithms (machine learning algorithm analogues to Amazon or Netflix recommender systems, in which a customer who buys A and B is likely to buy C; mutations A and B are determined to be likely associated with mutation C). Results Of 2793 pts with pAML, 1352 (48%) had NK and included in the final analysis. The median age of NK pts was 55 years (range, 18-93). Median WBC, hemoglobin, and platelet count at diagnosis were; 21.3 X 109(range, .2-600), 9.1 g/L (range, 2.7-17.6), and 61 X 109 (range, 5-950), respectively. The median number of mutations/sample was 3 (range, 0-7). The most commonly mutated genes were: NPM1 (49%), DNMT3A (37%), FLT3-ITD (24%), CEBPa (19%), TET2 (17%), IDH2 (17%), and RUNX1 (15%). In univariate Cox regression analysis, mutations in NPM1 (HR .81, median OS 58.3 months[m], p= .008), and CEBPa (single mutant, HR, .8, median OS 91.4 m, and double mutant, HR .69, median OS not reached, p These findings suggest that the prognostic impact of molecular data on OS in AML pts with NK is limited when using only one or two mutations (in the exception of TP53 mutations which are present in ~ 1% of NK AML pts), Figure 1. For example, pts with NPM1 mutations can have variable OS depending on presence or absence of mutations in other genes. The median OS for pts with NMP1Mut/FLT3-ITDWt/DNMT3AWt(Mut = mutated, Wt = wild type) was 99.1 m compared to 13.4 m for pts with NPM1Mut/FLT3-ITDMut/DNMT3AMut (triple positive), p Conclusions We developed genomic combinations that can improve the risk stratification of AML pts with NK using unbiased advanced analytic approaches. These combinations can divide patients into 4 risk categories that may aid physicians in treatment decisions. Such approaches account for the impact of individual mutations and the complexity of genomic interactions on outcomes. Disclosures Nazha:MEI: Consultancy. Meggendorfer:MLL Munich Leukemia Laboratory: Employment. Walter:MLL Munich Leukemia Laboratory: Employment. Hutter:MLL Munich Leukemia Laboratory: Employment. Carraway:Balaxa: Membership on an entity9s Board of Directors or advisory committees, Speakers Bureau; Agios: Consultancy, Speakers Bureau; Amgen: Membership on an entity9s Board of Directors or advisory committees; Celgene: Membership on an entity9s Board of Directors or advisory committees, Research Funding, Speakers Bureau; FibroGen: Consultancy; Jazz: Speakers Bureau; Novartis: Speakers Bureau. Maciejewski:Ra Pharmaceuticals, Inc: Consultancy; Alexion Pharmaceuticals, Inc.: Consultancy, Membership on an entity9s Board of Directors or advisory committees, Speakers Bureau; Apellis Pharmaceuticals: Consultancy; Ra Pharmaceuticals, Inc: Consultancy; Apellis Pharmaceuticals: Consultancy; Alexion Pharmaceuticals, Inc.: Consultancy, Membership on an entity9s Board of Directors or advisory committees, Speakers Bureau. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Sekeres:Opsona: Membership on an entity9s Board of Directors or advisory committees; Celgene: Membership on an entity9s Board of Directors or advisory committees; Celgene: Membership on an entity9s Board of Directors or advisory committees; Opsona: Membership on an entity9s Board of Directors or advisory committees. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership.

Constructing a Hybrid Recommender System

Recommender systems provide recommendations for users. In this paper, we train and test some algorithms for recommender systems on a certain dataset, as well as analyze their strengths and weaknesses. Based on this information, we construct a system that yields a better solution.

An evaluation of clinical order patterns machine-learned from clinician cohorts stratified by patient mortality outcomes.

Evaluate the quality of clinical order practice patterns machine-learned from clinician cohorts stratified by patient mortality outcomes. Inpatient electronic health records from 2010 to 2013 were extracted from a tertiary academic hospital. Clinicians (n = 1822) were stratified into low-mortality (21.8%, n = 397) and high-mortality (6.0%, n = 110) extremes using a two-sided P-value score quantifying deviation of observed vs. expected 30-day patient mortality rates. Three patient cohorts were assembled: patients seen by low-mortality clinicians, high-mortality clinicians, and an unfiltered crowd of all clinicians (n = 1046, 1046, and 5230 post-propensity score matching, respectively). Predicted order lists were automatically generated from recommender system algorithms trained on each patient cohort and evaluated against (i) real-world practice patterns reflected in patient cases with better-than-expected mortality outcomes and (ii) reference standards derived from clinical practice guidelines. Across six common admission diagnoses, order lists learned from the crowd demonstrated the greatest alignment with guideline references (AUROC range = 0.86-0.91), performing on par or better than those learned from low-mortality clinicians (0.79-0.84, P < 10-5) or manually-authored hospital order sets (0.65-0.77, P < 10-3). The same trend was observed in evaluating model predictions against better-than-expected patient cases, with the crowd model (AUROC mean = 0.91) outperforming the low-mortality model (0.87, P < 10-16) and order set benchmarks (0.78, P < 10-35). Whether machine-learning models are trained on all clinicians or a subset of experts illustrates a bias-variance tradeoff in data usage. Defining robust metrics to assess quality based on internal (e.g. practice patterns from better-than-expected patient cases) or external reference standards (e.g. clinical practice guidelines) is critical to assess decision support content. Learning relevant decision support content from all clinicians is as, if not more, robust than learning from a select subgroup of clinicians favored by patient outcomes.

HybRecSys: Content-based contextual hybrid venue recommender system

The popularity of location-based social networks has prompted researchers to study recommendation systems for location-based services. When used separately, each existing venue recommendation system algorithm has its own drawbacks (e.g. cold start, data sparsity, scalability). Another issue is that critical information about context is not commonly used in venue recommendation systems. This article proposes a hybrid recommendation model that combines contextual information, user-based and item-based collaborative filtering and content-based filtering. For this purpose, we collected user visit histories, venue-related information (distance, category, popularity and price) and contextual information (weather, season, date and time of visits) related to individual user visits from Twitter, Foursquare and Weather Underground. Experimental evaluation of the proposed hybrid system (HybRecSys) using a real-world dataset shows better results than baseline approaches.

Review Aware Recommender System

Context aware recommender systems (CARS) are recommender systems (RS) that provide recommendations according to user contexts. The first challenge for building such a system is to get the contextual information. Some works tried to get this information from reviews provided by users in addition to their ratings. However, all of these works perform important feature engineering in order to infer the context. In this article, the authors present a new CARS architecture that allows to automatically use contextual information from reviews without requiring any feature engineering. Moreover, they develop a new CARS algorithm that is tailored to textual contexts, that they call Textual Context Aware Factorization Machines (TCAFM). An empirical evaluation shows that the proposed architecture allows to significantly improve recommendation accuracy using state of the art RS and CARS algorithms, whereas TCAFM leads to additional improvements.

Trustor clustering with an improved recommender system based on social relationships

As we face a deluge of information in the modern world, the importance of recommender systems (RSs) that recommend relevant items to users has increased. The majority of existing RS schemes observe the prior ratings history of consumers to identify preferred items. However, current RSs suffer from the cold start problem, and their performance is dismal when new users or items appear. In order to address the cold start problem, a new type of solution that exploits social network features has been proposed. Many such social RSs analyze trustor–trustee relationships to discover latent social features shared between trustor and trustee. Since social relationships between trustors and trustees are directed, but not reciprocal, it is not guaranteed that a trustee has features in common with its trustors. Moreover, existing schemes are based on the assumption of independence between trustors who follow the same trustee, and therefore fail to recognize quintessential factors shared by the trustors.We posit that trustors who follow the same trustee have features in common. Based on the assumption that trustors who endorse the same trustee share similar tastes, we propose a new latent feature called Matrix S, and develop two novel RS algorithms that learn these latent features. We conduct an extensive performance evaluation using large scale real-world datasets, and observe that our proposed methods are not only more accurate than existing schemes but also show potential extensibility.

Tailoring motivational health messages for smoking cessation using an mHealth recommender system integrated with an electronic health record: a study protocol

BackgroundSmoking is one of the most avoidable health risk factors, and yet the quitting success rates are low. The usage of tailored health messages to support quitting has been proved to increase quitting success rates. Technology can provide convenient means to deliver tailored health messages. Health recommender systems are information-filtering algorithms that can choose the most relevant health-related items—for instance, motivational messages aimed at smoking cessation—for each user based on his or her profile. The goals of this study are to analyze the perceived quality of an mHealth recommender system aimed at smoking cessation, and to assess the level of engagement with the messages delivered to users via this medium.MethodsPatients participating in a smoking cessation program will be provided with a mobile app to receive tailored motivational health messages selected by a health recommender system, based on their profile retrieved from an electronic health record as the initial knowledge source. Patients’ feedback on the messages and their interactions with the app will be analyzed and evaluated following an observational prospective methodology to a) assess the perceived quality of the mobile-based health recommender system and the messages, using the precision and time-to-read metrics and an 18-item questionnaire delivered to all patients who complete the program, and b) measure patient engagement with the mobile-based health recommender system using aggregated data analytic metrics like session frequency and, to determine the individual-level engagement, the rate of read messages for each user. This paper details the implementation and evaluation protocol that will be followed.DiscussionThis study will explore whether a health recommender system algorithm integrated with an electronic health record can predict which tailored motivational health messages patients would prefer and consider to be of a good quality, encouraging them to engage with the system. The outcomes of this study will help future researchers design better tailored motivational message-sending recommender systems for smoking cessation to increase patient engagement, reduce attrition, and, as a result, increase the rates of smoking cessation.Trial registrationThe trial was registered at clinicaltrials.org under the ClinicalTrials.gov identifier NCT03206619 on July 2nd 2017. Retrospectively registered.

An improved hybrid collaborative filtering algorithm based on tags and time factor

The Collaborative Filtering (CF) recommendation algorithm, one of the most popular algorithms in Recommendation Systems (RS), mainly includes memory-based and model-based methods. When performing rating prediction using a memory-based method, the approach used to measure the similarity between users or items can significantly influence the recommendation performance. Traditional CFs suffer from data sparsity when making recommendations based on a rating matrix, and cannot effectively capture changes in user interest. In this paper, we propose an improved hybrid collaborative filtering algorithm based on tags and a time factor (TT-HybridCF), which fully utilizes tag information that characterizes users and items. This algorithm utilizes both tag and rating information to calculate the similarity between users or items. In addition, we introduce a time weighting factor to measure user interest, which changes overtime. Our experimental results show that our method alleviates the sparsity problem and demonstrates promising prediction accuracy.

Social learning strategies for matters of taste.

Most choices people make are about 'matters of taste', on which there is no universal, objective truth. Nevertheless, people can learn from the experiences of individuals with similar tastes who have already evaluated the available options-a potential harnessed by recommender systems. We mapped recommender system algorithms to models of human judgement and decision-making about 'matters of fact' and recast the latter as social learning strategies for matters of taste. Using computer simulations on a large-scale, empirical dataset, we studied how people could leverage the experiences of others to make better decisions. Our simulations showed that experienced individuals can benefit from relying mostly on the opinions of seemingly similar people; by contrast, inexperienced individuals cannot reliably estimate similarity and are better off picking the mainstream option despite differences in taste. Crucially, the level of experience beyond which people should switch to similarity-heavy strategies varies substantially across individuals and depends on how mainstream (or alternative) an individual's tastes are and the level of dispersion in taste similarity with the other people in the group.

Category Preferred Canopy–K-means based Collaborative Filtering algorithm

It is the era of information explosion and overload. The recommender systems can help people quickly get the expected information when facing the enormous data flood. Therefore, researchers in both industry and academia are also paying more attention to this area. The Collaborative Filtering Algorithm (CF) is one of the most widely used algorithms in recommender systems. However, it has difficulty in dealing with the problems of sparsity and scalability of data. This paper presents Category Preferred Canopy–K-means based Collaborative Filtering Algorithm (CPCKCF) to solve the challenges of sparsity and scalability of data. In particular, CPCKCF proposes the definition of the User–Item Category Preferred Ratio (UICPR), and use it to compute the UICPR matrix. The results can be applied to cluster the user data and find the nearest users to obtain prediction ratings. Our experimentation results performed using the MovieLens data set demonstrates that compared with traditional user-based Collaborative Filtering algorithm, the proposed CPCKCF algorithm proposed in this paper improved computational efficiency and recommendation accuracy by 2.81%.

Evaluating recommendation and search in the labor market

This study evaluates the most popular recommender system algorithms for use on both sides of the labor market: job recommendation and job seeker recommendation. Recent research shows the drawbacks of focusing solely on predictive power when evaluating recommender systems, which become especially prominent in job- and job seeker recommendation, where aspects such as reciprocity and item spread are two other vital performance metrics for the quality of recommendations. Besides evaluating using these extra metrics, we compare recommendation with search using free text search engines. We measure what is gained, and what is lost when consuming items (jobs and job seekers) retrieved using search versus items presented via a recommender system. Based on insights in date recommendation literature, we propose changes to rating matrix construction aimed at mitigating the drawbacks of recommendation in the labor market. Our results, obtained from extensive experimentation on three datasets gathered from the Flemish public employment services, show that popular recommender algorithms perform significantly worse than user search in terms of reciprocity. Furthermore, we show that by swapping the rating matrices between two sides of a reciprocal recommender context, we can outperform user search in terms of reciprocity with limited trade off in predictive power. The insights from this research can help actors in the labor market to better understand the positioning of recommendation versus search, and to provide better job recommendations and job seeker recommendations.

A Collaborative Filtering Recommendation Algorithm with Improved Similarity Calculation

Abstract In order to improve the accuracy of the proposed algorithm in collaborative filtering recommendation system, an Improved Pearson collaborative filtering (IP-CF) algorithm is proposed in this paper. The algorithm uses the user portrait, item characteristics and data of user behavior to compute the baseline predictors model. Instead of the traditional algorithm’s similarity calculation, the prediction model is used to improve the accuracy of the recommendation algorithm. Experimental results on Moivelens dataset show that the IP-CF algorithm significantly improves the accuracy of the recommended results, and the RMSE and MAE evaluation results are better than the traditional algorithms.

Transfer Learning for Resolving Sparsity Problem in Recommender systems: Human Values Approach

With the rapid rise in popularity of ecommerce application, Recommender Systems are being widely used by them to predict the response that a user will give to a given item. This prediction helps in cross selling, upselling and to increase the loyalty of their customers. However due to lack of sufficient feedback data these systems suffer from sparsity problem which leads to decline in their prediction efficiency. In this work, we have proposed and empirically demonstrated how the Transfer Learning approach using five dimensions of basic human values can be successfully used to alleviate the sparsity problem and increase the efficiency of recommender system algorithms.

Social tagging strategy for enhancing e-learning experience

Success of e-learning systems depends on their capability to automatically retrieve and recommend relevant learning content according to the preferences of a specific learner. Learning experience and dynamic choice of educational material that is presented to learners can be enhanced using different recommendation techniques. As popularity of collaborative tagging systems grows, users’ tags could provide useful information to improve recommender system algorithms in e-learning environments. In this paper, we present an approach for implementation of collaborative tagging techniques into online tutoring system. The implemented approach combines social tagging and sequential patterns mining for generating recommendations of learning resources to learners. Several experiments were carried out in order to verify usability of the proposed hybrid method within e-learning environment and analyze selected social tagging techniques.

A survey on data mining techniques in recommender systems

Recommender systems have been regarded as gaining a more significant role with the emergence of the first research article on collaborative filtering (CF) in the mid-1990s. CF predicts the interests of an active user based on the opinions of users with similar interests. To extract information on the preference of users for a set of items and evaluate the performance of the recommender system's techniques and algorithms, a critical analysis can be conducted. This study therefore employs a critical analysis on 131 articles in CF area from 36 journals published between the years 2010 and 2016. This analysis seems to be the exclusive survey which supports and motivates the community of researchers and practitioners. It is done by using the applications of users' activities and intelligence computing and data mining techniques on CF recommendation systems. In addition, it provides a classification of the literature on academic database according to the benchmark recommendation databases, two users' feedbacks (explicit and implicit feedbacks) which reflect their activities and categories of intelligence computing and data mining techniques. Eventually, this study provides a road map to guide future direction on recommender systems research and facilitates the accumulated and derived knowledge on the application of intelligence computing and data mining techniques in CF recommendation systems.

Performance Comparison of Feed-Forward Neural Networks Trained with Different Learning Algorithms for Recommender Systems

Accuracy improvement is among the primary key research focuses in the area of recommender systems. Traditionally, recommender systems work on two sets of entities, Users and Items, to estimate a single rating that represents a user’s acceptance of an item. This technique was later extended to multi-criteria recommender systems that use an overall rating from multi-criteria ratings to estimate the degree of acceptance by users for items. The primary concern that is still open to the recommender systems community is to find suitable optimization algorithms that can explore the relationships between multiple ratings to compute an overall rating. One of the approaches for doing this is to assume that the overall rating as an aggregation of multiple criteria ratings. Given this assumption, this paper proposed using feed-forward neural networks to predict the overall rating. Five powerful training algorithms have been tested, and the results of their performance are analyzed and presented in this paper.

RETRACTED ARTICLE: A novel technique applied to the economic investigation of recommender system

Recommender system has emerged as a new research concept in the economic field, in which a new recommend algorithm such as stock data mining plays an important role in studying the level of economic development in a region. A novel recommends method of big data analysis method based on singular value decomposition is proposed. The proposed algorithm exploits the historical data of stocks in the western region, the regional leading stock average data and volatility of individual stocks data. Then volatility charts could be gotten from data mining. The stability of the western region stock could be drawn by comparison between leading stocks and common stocks. Money flow of stocks can also be calculated by new recommender system algorithm. The experimental results show that our approach has ability to forecast the economic development of the western region by the perspective of stock data mining. It could effectively recommend investors to identify the economic development of the western region, obtaining higher returns, and avoiding unnecessary losses.