Manipulation-resistant collaborative filtering systems

Abstract—This study focuses on developing the multicriteria collaborative filtering algorithm for improving the prediction accuracy. The approaches applied were user-item multirating matrix decomposition, the measurement of user similarity using cosine formula and multidimensional distance, individual criteria weight calculation, and rating prediction for the overall criteria by a combination approach. Results of the study show variation in multicriteria collaborative filtering algorithm, which was used for improving the document recommender system, with the two following characteristics- first, the rating prediction for four individual criteria using collaborative filtering algorithm by a cosine-based user similarity and a multidimensional distance-based user similarity; second, the rating prediction for the overall criteria using combination algorithms. Based on the results of testing, it can be concluded that a variety of models developed for the multicriteria collaborative filtering systems had much better prediction accuracy than for the classic collaborative filtering, which was characterized by the increasingly smaller values of Mean Absolute Error. The best accuracy was achieved by the multicriteria collaborative filtering system with multidimensional distance-based similarity. Keywords—Algorithm; multicriteria collaborative filtering; document; recommendation; system; similarity; multidimensional distance; decomposition; combination; prediction; accuracy I.

The concept of time in traditional distributed systems has been inherited in the Computer-Supported Collaborative Work (CSCW) literature. The following assumptions have generally been made: (1) Events are atomic and their durations do not matter. (2) Total ordering of events can be achieved by some mechanical algorithm. (3) The relationship between events is determined solely by time (causal relationship). However, we observe that these assumptions are not appropriate if the goal is to faithfully preserve user intentions in collaborative systems. In particular, we discuss why and how these assumptions should be relaxed or removed in the design of collaborative editing systems.

Cloud computing paradigm builds on the foundations of distributed computing, grid computing, virtualisation, service orientation, etc. Cloud storage, which is one of the most attractive cloud services, offers numerous benefits from both the technology and functionality perspectives such as increased availability, flexibility, and functionality. Therefore, Cloud-based platforms have become fundamental to collaborative and data management systems. However, the new Cloud environment creates major collaborative and data management challenges such as efficient and reliable remote storage, online data query processing, data integrity, outsourcing computation, and secure virtualisation, among others. Such issues need to be carefully studied and solved to enable the wide deployment and adoption of Cloud computing in businesses, industry, etc. The purpose of this special issue is to publish recent advances in Cloud-based collaborative systems and Cloud storage. The special issue comprises five high-quality papers, which are arranged as follows. De Maio et al in the first paper1 “Social Media Marketing through Time-Aware Collaborative Filtering,” define a time-aware collaborative filtering for estimating users' interest along the time in Twitter. Their approach uses text analysis services to semantically annotate tweets' content and to track concepts considering post frequencies along the time. A model-based approach implementing K-Nearest Neighbors is used to estimate user's similarity representing their profile by sampling user's interest with three different techniques: Vectorial Representation, Symbolic Aggregation Approximation, and Median. The authors present experimental results comparing these techniques and performing model training in different time windows. The second paper2 “SERNOTATE: An Automated Approach for Business Service Description Annotation for efficient Service Retrieval and Composition” by Chotipant et al proposes an automated approach, namely, SERNOTATE, for business service description annotation for efficient service retrieval and composition. To that end, the authors employ new semantic-based linking approaches, namely, Extended Case-based Reasoning, Vector-based and Classification-based, that automatically annotate business services to relevant service concepts. The experimental results test and validate the applicability of the proposed approaches to the automatic annotation of business service descriptions to service concepts on a real-world dataset. In the third paper,3 “Sensor Data Management in the Cloud: Data Storage, Data Ingestion, and Data Retrieval,” the authors address an increasing need to capture, store, and analyse the dynamic semistructured data from sensors. A similar growth of semistructured data in the modern Web has led to the creation of NoSQL data stores for scalability, availability, and performance, whereas large-scale data processing frameworks for parallel analysis. The authors study how sensor data management can benefit from MongoDB with Apache Spark technologies, specifically for ingesting high-velocity sensor data and parallel retrieval of high volume data. The performance of MongoDB sharding and no-sharding databases with Apache Spark are evaluated to identify the right software environment for sensor data management. Park et al in the fourth paper,4 “Study on the SDN-IP based Solution of Well-known Bottleneck Problems in Private Sector of National R&E Network for Big Data Transfer,” deal with issues arising in Software Defined Networking (SDN). The key feature of SDN networks is the provisioning of dynamic service architecture. Therefore, it is more appropriate for the big data science computing trends, for handling today's big data science, which requires massive parallel processing and a constant demand for additional capacity and on demand connectivity. The authors bring the experience from KREONET (Korea Research Environment Open Network) and Korea National R&E Network, which adopt SDN as a platform service for big data scientific application. The big data application for KREONET SDN service comprises high-energy physics, astrophysics, biology/genetics, meteorology, artificial satellite data, etc. The last paper5 “Improved Outsourced Private Set Intersection Protocol Based on Polynomial Interpolation” by Yang et al investigates issues in private set intersection (PSI) protocols, which enable two parties to compute the intersection of their inputs without compromising anything about the datasets beyond the intersection. The authors present a variant of delegated private set intersection protocol secure in the semi-honest model under RSA assumption as well as an efficient and secure outsourcing computation algorithm for RSA crypto-system. Based on this algorithm, the authors transform a variant of delegated private set intersection protocol into an improved outsourced one, which enables the clients to perform only simple modular multiplication for computing during the execution of the protocol. Compared with the state of the art, the proposed protocol has a great advantage in efficiency. The editor of this special issue would like to thank all authors for their contributions and the reviewers for their constructive and useful feedback to authors. I would like to thank Prof Geoffrey Fox, the Editor-in-Chief of Concurrency and Computation: Practice and Experience, for the opportunity to edit this special issue and his support.

Collaborative Filtering systems consider users' social environment to predict what each user may like to visit in a social network i.e. they collect and analyze a large amount of information on users' behavior, activities or preferences and then predict or make suggestions to users. These systems use ranks or tags each user assign to different resources to make predictions. Lately, social tagging systems, in which users can insert new contents, tag, organize, share and search for contents, are becoming more popular. These social tagging systems have a lot of valuable information, but the data expansion in them is very fast and this has led to the need for recommender systems that will predict what each user may like or need make these suggestions to them. One of the problems in these systems is: “how much can we rely on the similar users, are they trustworthy?”. In this article we use trust metric, which we conclude from users' tagging behavior, beside similarities to give suggestions. Results show considering trust in a collaborative system can lead to better performance in generating suggestions.

Considering the natural tendency of people to follow direct or indirect cues of other people's activities, collaborative filtering-based recommender systems often predict the utility of an item for a particular user according to previous ratings by other similar users. Consequently, effective searching for the most related neighbors is critical for the success of the recommendations. In recent years, collaborative tagging systems with social bookmarking as their key component from the suite of Web 2.0 technologies allow users to freely bookmark and assign semantic descriptions to various shared resources on the web. While the list of favorite web pages indicates the interests or taste of each user, the assigned tags can further provide useful hints about what a user thinks of the pages.In this paper, we propose a new collaborative filtering approach TBCF (Tag-based Collaborative Filtering) based on the semantic distance among tags assigned by different users to improve the effectiveness of neighbor selection. That is, two users could be considered similar not only if they rated the items similarly, but also if they have similar cognitions over these items. We tested TBCF on real-life datasets, and the experimental results show that our approach has significant improvement against the traditional cosine-based recommendation method while leveraging user input not explicitly targeting the recommendation system.

The team design becomes increasingly important to future manufacturing industry. This paper focus on the key issues of collaborative 3D design system through Internet or Intranet in CSCW(Computer Support for Cooperative Work). In the paper, we propose our solutions of concurrency control tactics, collaboration aware, multi-mode communications network management, synchronous real-time protection in a collaborative 3D design system. Based on open source projects-open CASCADE and Adaptive Communication Environment (ACE), we implement our collaborative 3D design system, and experiments show our system can meet the requirements of some 3D cooperative modeling design applications.

Collaborative filtering recommender systems provide their users with relevant items based on information from other similar users. Popular collaborative filtering approaches such as Pearson correlation coefficient and cosine similarity, compute the similarity between users based on the set of their co-rated items. However, similarities are commonly computed without taking the popularity of the set of two users' co-rated items into consideration, e.g. an item rated by very many users should have less impact on the similarity measure, and analogously an item rated by few should have a larger impact on the similarity score of two users. In this paper, we investigate the effects of common weighting schemes on different types of users, i.e. new users with few ratings (so-called cold start users), post cold start users, and power users. Empirical studies over two datasets have shown in which of these cases weighting schemes are beneficial in terms of recommendation quality.

Recommendation systems (RS) are decision support tools created to deal with information overload, which is the main challenge of the modern digital world. The aim of RS is to provide users with interesting items based on their preferences. Collaborative filtering (CF) is the most implemented recommendation technique, it is based on the idea that similar users have similar preferences. Evidential CF is a subclass of classical CF handling uncertainty using the framework of Dempster–Shafer Theory (DST). Evidential CF recommenders (ECFRS) are suitable for critical domains such as healthcare and threat assessment, where uncertainty management remains a major challenge. In this paper, we developed a user-based evidential CF system, where the number of the co-rated items is considered in predictions generation. The proposed approach is based on Evidential KNN where the Jaccard factor is used in the neighborhood selection. Our approach is tested using Movielens dataset. Experimental results show the importance of introducing a co-rating factor in improving the recommendation quality of traditional ECFRS.KeywordsRecommender systemUser-based CFCo-ratingDempster–Shafer theoryEvidential KNN

This paper gives an overview of a collaborative design system (CDS) for electromechanical products. To reduce design costs and to manufacture high-quality products, it is well known that concurrent engineering (CE) is a very efficient approach. Three-dimensional (3D) CAD system and engineering database system are essential components of CE. The CDS is an environment to realize CE. By creating 3D models in a computer and performing some simulations such as mechanical, electronic, software simulation and integrated simulations, it is possible to estimate functions, assemblability, manufacturability and so on, before making prototype models. In this paper, we outline the CDS and mainly discuss the total information management system (TIMS) which makes an important role of the CDS. This paper describes the implementation experience of some functions of the TIMS.

In order to facilitate product realisation processes, presently, research is actively being carried out to develop methodologies and technologies to support geographically dispersed teams to organise collaborative design based on the quickly evolving information technologies. A number of research works and commercial systems have appeared to provide solutions for collaborative and distributed product development, and the practical applications are getting more pervasive and mature. In this paper, the recently related works are summarised from three aspects—visualisation-based collaborative systems, co-design collaborative systems and CE (concurrent engineering)-based collaborative systems. Around these aspects, about 100 papers and 30 commercial systems/international standards published or launched recently are discussed. The current research and development statuses and issues, underlying algorithms, mechanisms and system architectures, and the future trends and challenges are explained and compared in detail.

Collaborative filtering (CF) approach is successfully applied in the rating prediction of personal recommendation. But individual information source is leveraged in many of them, i.e., the information derived from single perspective is used in the user-item matrix for recommendation, such as user-based CF method mainly utilizing the information of user view, item-based CF method mainly exploiting the information of item view. In this paper, in order to take full advantage of multiple information sources embedded in user-item rating matrix, we proposed a rating-based integrated recommendation framework of CF approaches to improve the rating prediction accuracy. Firstly, as for the sparsity of the conventional item-based CF method, we improved it by fusing the inner similarity and outer similarity based on the local sparsity factor. Meanwhile, we also proposed the improved user-based CF method in line with the user-item-interest model (UIIM) by preliminary rating. Second, we put forward a background method called user-item-based improved CF (UIBCF-I), which utilizes the information source of both similar items and similar users, to smooth itembased and user-based CF methods. Lastly, we leveraged the three information sources and fused their corresponding ratings into an Integrated CF model (INTE-CF). Experiments demonstrate that the proposed rating-based INTE-CF indeed improves the prediction accuracy and has strong robustness and low sensitivity to sparsity of dataset by comparisons to other mainstream CF approaches.

Multi-LLM collaborative systems have attracted significant attention as a promising solution for complex tasks, enabling multiple large language models (LLMs) with different domains to work together toward a common goal. Different collaborative structures (e.g., Centralized, Horizontal, and Joint Interaction) and communication methods (e.g., direct, summary, and vote) give the system with enhanced flexibility and reasoning capability. However, these same mechanisms also introduce potential security and privacy risks, such as the generation of incorrect responses and the leakage of sensitive information. Based on the above unique characteristics of collaborative systems, we propose three attack methods (named Decision Poisoning Attack, Indirect Echoleak Attack and Information Collision Attack) that exploit the interactions between LLMs to achieve different objectives like system manipulation and privacy leakage. Extensive experiments demonstrate the effectiveness of the proposed attack on three structures and three communication methods, highlighting the security vulnerabilities and potential risks in Multi-LLM collaborative systems. We further discuss possible defense methods that can mitigate the attack performance. Our work show that (1) the key factor for a successful attack on collaborative systems is ensuring the malicious instruction persists and propagates throughout the inter-LLMs communication. (2) both the system architecture and the communication method can affect attack effectiveness, offering valuable insights for the design of more secure Multi LLM collaborative systems in the future.

A key challenge of the collaborative filtering (CF) information filtering is how to obtain the reliable and accurate results with the help of peers' recommendation. Since the similarities from small-degree users to large-degree users would be larger than the ones in opposite direction, the large-degree users' selections are recommended extensively by the traditional second-order CF algorithms. By considering the users' similarity direction and the second-order correlations to depress the influence of mainstream preferences, we present the directed second-order CF (HDCF) algorithm specifically to address the challenge of accuracy and diversity of the CF algorithm. The numerical results for two benchmark data sets, MovieLens and Netflix, show that the accuracy of the new algorithm outperforms the state-of-the-art CF algorithms. Comparing with the CF algorithm based on random walks proposed by Liu et al. (Int. J. Mod. Phys. C, 20 (2009) 285) the average ranking score could reach 0.0767 and 0.0402, which is enhanced by 27.3% and 19.1% for MovieLens and Netflix, respectively. In addition, the diversity, precision and recall are also enhanced greatly. Without relying on any context-specific information, tuning the similarity direction of CF algorithms could obtain accurate and diverse recommendations. This work suggests that the user similarity direction is an important factor to improve the personalized recommendation performance.

With the development of e-commerce, shopping on-line is becoming more and more popular. When we need to decide whether to purchase a product or not on line, the opinions of others become important. The convenience of new web technologies enables us to freely express our opinions and reviews for various products we have purchased which leads to a serious problem, information overloading. How to mine these review data to understand customers' preferences and make recommendations is crucial to merchants and researchers. Traditional collaborative filtering (CF) algorithm is one of the most successful recommendation system technologies. The core idea of CF algorithm is to recommend products based on other people who have similar tastes with target users. However, the ability of CF is limited by the sparsity problem, which is very common in reality. The reason derives from the fact that traditional CF method only takes users' ratings into account. In this paper, we propose a new personalized recommendation model, i.e. topic model based collaborative filtering (TMCF) utilizing users' reviews and ratings. We exploit extended LDA model to generate topic allocations for each review and then obtain each user's preference. Moreover, a new metric is designed to measure similarity between users alleviating the sparsity problem to a large extent. Finally, recommendations are made based on similar users' ratings. Experiments on seven data sets indicate better prediction accuracy than other traditional and state-of-the-art methods with substantial improvement in alleviating the sparsity problem.

Collaborative and content-based filtering are the majorly used approaches towards implementing recommender systems which can successfully predict the item(s) to be recommended to users based upon their preferences. Each of the methods has its own advantages and disadvantages that are specific to the situation they are used in. The hybrid approach aims to combine both techniques in multiple ways to optimize results and overcome shortcomings. In this paper, we propose a hybrid approach based recommendation engine to improve accuracy. While content based recommendations are made on the similarity of items’ attributes, collaborative recommendations are based on similarity of users. Collaborative filtering is achieved by matrix factorization technique. One of the most effective algorithms for matrix factorization: Singular Value Decomposition (SVD) has been used to perform collaborative filtering and combined with content based model to predict the item ratings per user. The paper aims to achieve the final result with minimum possible errors as per user preferences and recommended items.