Commercial Recommender Systems Research Articles

Deep Attentive Interest Collaborative Filtering for Recommender Systems

Collaborative filtering (CF) is a pivotal building block in commercial recommender systems due to its strength and utility in user interest modeling. Recently, many researchers have turned to deep learning as a way to capture richer collaborative signals from user-item feature interactions. However, most deep-based methods only consider nonlinear, high-order interactions while ignoring the explicit collaborative signals in low-order interactions. They also typically ignore the quality of the user and item profiles. These are cornerstones in item recommendation that, we argue, must be considered for high-quality recommendations. Hence, we propose Deep Attentive Interest Collaborative Filtering (DAICF) to overcome these limitations. DAICF profiles users based on their interactive items, i.e., user neighborhood information. Similarly, item profiles are based on users who had interacted with it, i.e., item neighborhood information. Given a user's profile varies over different items, DAICF accurately models his attentive interests based on the specific target item. Low-order collaborative signals are captured by a shallow component, and high-order collaborative signals are captured by a deep component. These two complementary collaborative signals are then fused to provide rich recommendations that cut through today's information overload. By designing a personalized feature extraction method based on bilateral neighborhood information to solve the data sparsity problem in recommender systems, DAICF can dynamically distinguish the importance of a user's historical interaction items for predicting user preferences for a specific target item. A set of experiments against four real-world datasets validate that DAICF outperforms the most recent state-of-the-art recommendation algorithms and justifies the effectiveness and interpretability of our method.

Click-through rate prediction using transfer learning with fine-tuned parameters

In real business platforms, recommendation systems usually need to predict the CTR of multiple business. Since different scenarios may have common feature interactions, knowledge transferring based methods are often used by re-optimizing the pre-trained CTR model from source scenarios to a new target domain. In addition to knowledge transfer, it is noteworthy that generalizing target domain data outside of the CTR model accurately is also important when re-training all of the fine-tuned parameters. Generally, the pre-trained model trained on large source domains can represent the characteristics of different instances and capture typical feature interactions. It would be useful to directly reuse fine-tuned parameters from source domains to serve the target domain. However, different instances of the target domain may need different amounts of source information to fine-tune the model parameters, and these decisions of freezing or re-optimizing model parameters, which highly depend on the fine-tuned model and target instances, may require much manual effort. In this paper, we propose an end-to-end transfer learning framework with fine-tuned parameters for CTR prediction, called Automatic Fine-Tuning (AutoFT). The principal component of AutoFT is a set of learnable transfer policies that independently determine how the specific instance-based fine-tuning policies should be trained, which decide the routing in the embedding representations and the high-order feature representations layer by layer in deep CTR model. Extensive tests on two benchmarks and one real commercial recommender system deployed in Huawei's App Store show that AutoFT can greatly increase CTR prediction performance when compared to current transferring methodologies.

Promoting active sports tourism through technology and evaluating its economic impact: experiences from Cyprus

ABSTRACT While the economic impact of major passive sport events is well documented, the contribution of active sports events is scarcely examined. This reflective practice paper contributes towards bridging the gap by examining, in a case study, the economic impact of one of the first big active sport events in Cyprus, organised using the state-of-the-art technology. The paper presents the first commercial web platform and recommender system dedicated to active sports events, which provide all necessary services to customers and thus simplify, and ultimately increase participation in the event. We find strong evidence in support of the idea that active sport events have a strong positive impact on the local and countrywide economy, while the technology can contribute and enable further promotion and better organisation of such events.

Recommending news in traditional media companies

AbstractThe adoption of recommender systems in online news personalization has made it possible to tailor the news stream to the individual interests of each reader. Previous research on commercial recommender systems has emphasized their use in large‐scale media houses and technology companies, and real‐world experiments indicate substantial improvements of click rates and user satisfaction. It is less understood how smaller media houses are coping with this new technology, how the technology affects their business models, their editorial processes, and their news production in general. Here we report on the experiences from numerous Scandinavian media houses that have experimented with various recommender strategies and streamlined their news production to provide personalized news experiences. In addition to influencing the content and style of news stories and the working environment of journalists, the news recommender systems have been part of a profound digital transformation of the whole media industry. Interestingly, many media houses have found it undesirable to automate the entire recommendation process and look for approaches that combine automatic recommendations with editorial choices.

The Public Service Approach to Recommender Systems: Filtering to Cultivate

Online media consumption has been radically transformed by how media companies algorithmically recommend content to their users. Public service media (PSM) have also realized the potential of recommender systems and are increasingly using these technologies to personalize their online offering. PSM are on the other hand required to disseminate diverse content, which can be incompatible with the logics of commercial recommender systems that primarily seek to drive up media consumption. Drawing on previous research on selective exposure and media diversity, this study presents the results from interviews with ten PSM informants across Europe, revealing that data scientists within these organizations are highly aware of the effects recommendations have on media consumption, and design the PSM online services accordingly. This study contributes with in-depth knowledge of how diversity has been interpreted at operational levels in PSM and how recommender systems are being adapted to a non-commercial setting.

An Attention-Based Latent Information Extraction Network (ALIEN) for High-Order Feature Interactions

One of the primary tasks for commercial recommender systems is to predict the probabilities of users clicking items, e.g., advertisements, music and products. This is because such predictions have a decisive impact on profitability. The classic recommendation algorithm, collaborative filtering (CF), still plays a vital role in many industrial recommender systems. However, although straight CF is good at capturing similar users’ preferences for items based on their past interactions, it lacks regarding (1) modeling the influences of users’ sequential patterns from their individual history interaction sequences and (2) the relevance of users’ and items’ attributes. In this work, we developed an attention-based latent information extraction network (ALIEN) for click-through rate prediction, to integrate (1) implicit user similarity in terms of click patterns (analogous to CF), and (2) modeling the low and high-order feature interactions and (3) historical sequence information. The new model is based on the deep learning, which goes beyond the capabilities of econometric approaches, such as matrix factorization (MF) and k-means. In addition, the approach provides explainability to the recommendation by interpreting the contributions of different features and historical interactions. We have conducted experiments on real-world datasets that demonstrate considerable improvements over strong baselines.

Genomic Biomarkers to Predict Resistance to Hypomethylating Agents in Patients With Myelodysplastic Syndromes Using Artificial Intelligence.

PURPOSEWe developed an unbiased framework to study the association of several mutations in predicting resistance to hypomethylating agents (HMAs) in patients with myelodysplastic syndromes (MDS), analogous to consumer and commercial recommender systems in which customers who bought products A and B are likely to buy C: patients who have a mutation in gene A and gene B are likely to respond or not respond to HMAs.METHODSWe screened a cohort of 433 patients with MDS who received HMAs for the presence of common myeloid mutations in 29 genes that were obtained before the patients started therapy. The association between mutations and response was evaluated by the Apriori market basket analysis algorithm. Rules with the highest confidence (confidence that the association exists) and the highest lift (strength of the association) were chosen. We validated our biomarkers in samples from patients enrolled in the S1117 trial.RESULTSAmong 433 patients, 193 (45%) received azacitidine, 176 (40%) received decitabine, and 64 (15%) received HMA alone or in combination. The median age was 70 years (range, 31 to 100 years), and 28% were female. The median number of mutations per sample was three (range, zero to nine), and 176 patients (41%) had three or more mutations per sample. Association rules identified several genomic combinations as being highly associated with no response. These molecular signatures were present in 30% of patients with three or more mutations/sample with an accuracy rate of 87% in the training cohort and 93% in the validation cohort.CONCLUSIONGenomic biomarkers can identify, with high accuracy, approximately one third of patients with MDS who will not respond to HMAs. This study highlights the importance of machine learning technologies such as the recommender system algorithm in translating genomic data into useful clinical tools.

Using entropy for similarity measures in collaborative filtering

Collaborative filtering has been successfully implemented in many commercial recommender systems. These systems recommend items favored by other users with similar preference history to the current user. As finding similar users is critical to the performance of the system, various techniques have been suggested to develop similarity measures. However, there are still much to be improved, because existing similarity measures simply utilize additional heuristic information and seldom reflect the global rating behaviors on items. This paper aims to improve the previous similarity measures by employing the information entropy of user ratings so that the user’s global rating behavior on items can be reflected. The efficiency of the proposed method is examined through extensive experiments to demonstrate its superior performance over the previous similarity measures especially in small-scaled and sparse datasets.

영화 추천을 위한 장르 흥미도를 이용한 새로운 협력 필터링 방식

협력 필터링은 상업적 추천 시스템에서 널리 사용되어 왔는데, 고객의 사회적 행태를 구현하여 사용자의 흥미에 부합하는 항목들을 제안하기 때문이다. 현재까지 적절한 항목을 추천하기 위한 가장 보편적인 방법은 유사한 사용자들을 찾아 그들의 평가치를 참조하는 방법이다. 본 논문은 영화를 추천하기 위해서 장르 흥미도를 기반으로 하는 새로운 유사도 공식을 제안하는데, 이는 기존 공식에서 사용자들의 평가등급 차이를 기반으로 하는 것과 대비된다. 광범위한 실험 결과에 따르면, 제안한 공식은 정확도와 추천의 질에 있어서 전통적인 유사도 공식의 성능을 크게 향상시키는 것으로 확인되었다. Collaborative filtering has been popular in commercial recommender systems, as it successfully implements social behavior of customers by suggesting items that might fit to the interests of a user. So far, most common method to find proper items for recommendation is by searching for similar users and consulting their ratings. This paper suggests a new similarity measure for movie recommendation that is based on genre interest, instead of differences between ratings made by two users as in previous similarity measures. From extensive experiments, the proposed measure is proved to perform significantly better than classic similarity measures in terms of both prediction and recommendation qualities.

A hybrid knowledge-based approach to collaborative filtering for improved recommendations

Collaborative filtering (CF) is one of the most successful and effective recommendation techniques for personalized information access. This method makes recommendations based on past transactions and feedback from users sharing similar interests. However, many commercial recommender systems are widely adopting the CF algorithms; these methods are required to have the ability to deal with sparsity in data and to scale with the increasing number of users and items. The proposed approach addresses the problems of sparsity and scalability by first clustering users based on their rating patterns and then inferring clusters (neighborhoods) by applying two knowledge-based techniques: rule-based reasoning (RBR) and case-based reasoning (CBR) individually. Further to improve accuracy of the system, HRC (hybridization of RBR and CBR) procedure is employed to generate an optimal neighborhood for an active user. The proposed three neighborhood generation procedures are then combined with CF to develop RBR/CF, CBR/CF, and HBR/CF schemes for recommendations. An empirical study reveals that the RBR/CF and CBR/CF perform better than other state-of-the-art CF algorithms, whereas HRC/CF clearly outperforms the rest of the schemes.

A unified music recommender system using listening habits and semantics of tags

In this paper, we propose a unified music recommender system using both listening habits and semantics of tags in a social music site. Most commercial music recommender systems recommend music items based on the number of plays, or explicit ratings of a song. However, these approaches have some difficulties in recommending new items with only a few ratings, or recommending items to new users with little information. To resolve the problem, UniTag ontology is developed, which defines the meaning and the weighted score of tags. User profiles are created by combining the score of tags and the number of plays, and a collaborative filtering algorithm is executed. For performance evaluation, precisions, recalls, and F-measures are measured using the listening habits-based recommendation, the tag score-based recommendation, and the unified recommendation, respectively. Our experiments show that the proposed approach outperforms the other two approaches in terms of all the evaluation metrics.

Multidirectional Product Support System for Decision Making in Textile Industry using Collaborative Filtering Methods

In the information technology ground people are using various tools and software for their official use and for their personal reasons. Nowadays people are worrying to choose data accessing tools and software's at the time of buying and selling the products and they are also worrying about various constraints such as cost, life time of the product, color and size of the product etc. In this paper we generated the solutions to the existing unsolved problems. Here we proposed the algorithm Multidirectional Rank Prediction (MDRP) decision making algorithm in order to take an effective decision at all the levels of data extraction, using the above technique and we analyzed the results at various datasets, finally the results were observed and compared with the existing methods such as PCC and VSS. The result accuracy was higher than the existing rank prediction methods. Collaborative techniques are used to filtering the noise data and give product recommendation to novice users. Collaborative filtering may involve in large data sets. Collaborative technique is used to predict user's interests on particular product or more than one product. In this Proposal Selling and buying are the two conventional activities for each seller and customer. One who sells the products is aims to gain the maximum earnings, and the customer has to get trustworthy product and it is extendable to intermediate level. These scenarios are identified and people who are used products are worried to discover the best product and in addition they suffer and face difficulty to draw the features of the item such as color, product size, availability and durability of the item. To prevail over these situations identified a problem learning multidirectional asymmetric similarity collaborative filtering via matrix factorization technique. The Collaborative filtering technique is a conventional recommender system which was used by different peoples in different circumstances, and data was extracted by this conventional method to various purposes. Still data extraction is a foremost problem in various disciplines, to handle seriously this problem and provide remedy to this issue. Solving any problem using any one the available technique is a common method, but the difficulty is to identify a best method or technique for long- term solution to the existing problem. In the exiting problem is providing the solution not up to the expected level of the customers. Customers are suffering sparsity and scalability problems. Most of the commercial recommender systems are associated with large data sets. The user-item matrix used for collaborative filtering could be tremendously large and sparse, which carries out the challenges in the performances of the recommendation. 1.1 Data Sparsity Problem One classic problem in data sparsity is the cold start problem. As collaborative filtering methods recommend items based on users' past preferences, new users must need to rate sufficient number of items to enable the system to capture their likings exactly and to provide reliable recommendations. Similarly, users when rating the new items also face the same problem. When new items are added to system, they need to be rated by large number of users before they could be recommended to users who have similar tastes with the ones rated them. The new item problem does not limit the content-based recommendation, because the recommendation of an item is based on its discrete set of descriptive qualities rather than its ratings. 1.2 Scalability Problem The Secondly Scalability is another problem in our traditional CF algorithms will face serious scalability problems. If the customer data set, no of items are high a normal CF algorithm time complexity was too large and many online version systems have to respond immediately and provide recommendations of their purchases and ratings history, which demands a higher scalability of the Collaborative filtering system.

Investigating the Persuasion Potential of Recommender Systems from a Quality Perspective

Recommender Systems (RSs) help users search large amounts of digital contents and services by allowing them to identify the items that are likely to be more attractive or useful. RSs play an important persuasion role, as they can potentially augment the users’ trust towards in an application and orient their decisions or actions towards specific directions. This article explores the persuasiveness of RSs, presenting two vast empirical studies that address a number of research questions. First, we investigate if a design property of RSs, defined by the statistically measured quality of algorithms, is a reliable predictor of their potential for persuasion. This factor is measured in terms of perceived quality, defined by the overall satisfaction, as well as by how users judge the accuracy and novelty of recommendations. For our purposes, we designed an empirical study involving 210 subjects and implemented seven full-sized versions of a commercial RS, each one using the same interface and dataset (a subset of Netflix), but each with a different recommender algorithm. In each experimental configuration we computed the statistical quality (recall and F-measures) and collected data regarding the quality perceived by 30 users. The results show us that algorithmic attributes are less crucial than we might expect in determining the user’s perception of an RS’s quality, and suggest that the user’s judgment and attitude towards a recommender are likely to be more affected by factors related to the user experience. Second, we explore the persuasiveness of RSs in the context of large interactive TV services. We report a study aimed at assessing whether measurable persuasion effects (e.g., changes of shopping behavior) can be achieved through the introduction of a recommender. Our data, collected for more than one year, allow us to conclude that, (1) the adoption of an RS can affect both the lift factor and the conversion rate, determining an increased volume of sales and influencing the user’s decision to actually buy one of the recommended products, (2) the introduction of an RS tends to diversify purchases and orient users towards less obvious choices (the long tail), and (3) the perceived novelty of recommendations is likely to be more influential than their perceived accuracy. Overall, the results of these studies improve our understanding of the persuasion phenomena induced by RSs, and have implications that can be of interest to academic scholars, designers, and adopters of this class of systems.

Recommender Systems in Commercial Use

Commercial recommender systems are deployed by marketing teams to increase revenue or personalize user experience. Marketers evaluate recommender systems not on their algorithms but on how well the vendor's expertise and interfaces will support achieving business goals. Driven by a business model that pays based on recommendation success, vendors guide clients through continuous optimization of recommendations. While recommender technology is mature, the solutions and market are still young. As a result, solutions are not fully integrated with other business systems and technology platforms. While the market is retail‐focused today, interest and vendor offerings are rapidly expanding to other areas. Retail clients will drive social, location, and mobile enhancements.

Enhancing Product Recommender Systems on Sparse Binary Data

Commercial recommender systems use various data mining techniques to make appropriate recommendations to users during online, real-time sessions. Published algorithms focus more on the discrete use...

Item-based top-Nrecommendation algorithms

The explosive growth of the world-wide-web and the emergence of e-commerce has led to the development ofrecommender systems---a personalized information filtering technology used to identify a set of items that will be of interest to a certain user. User-based collaborative filtering is the most successful technology for building recommender systems to date and is extensively used in many commercial recommender systems. Unfortunately, the computational complexity of these methods grows linearly with the number of customers, which in typical commercial applications can be several millions. To address these scalability concerns model-based recommendation techniques have been developed. These techniques analyze the user--item matrix to discover relations between the different items and use these relations to compute the list of recommendations.In this article, we present one such class of model-based recommendation algorithms that first determines the similarities between the various items and then uses them to identify the set of items to be recommended. The key steps in this class of algorithms are (i) the method used to compute the similarity between the items, and (ii) the method used to combine these similarities in order to compute the similarity between abasketof items and a candidate recommender item. Our experimental evaluation on eight real datasets shows that theseitem-basedalgorithms are up to two orders of magnitude faster than the traditional user-neighborhood based recommender systems and provide recommendations with comparable or better quality.