Recommender Systems In Order Research Articles

START Foundation: Coping with Bias and Fairness when Implementing and Using an AI System

START Foundation annually supports nearly 200 young people with a migratory background through an individualized scholarship program aimed at personal development and identity building. In 2022, START Foundation expanded its reach by launching a digital learning platform, which faced challenges such as information overload and navigation difficulties due to a growing number of courses and a lack of individual support. To improve user experience on the digital learning platform, START Foundation explored implementing a recommender system for personalized course navigation based on methods from the field of artificial intelligence (AI). However, this exploration raised ethical concerns, particularly regarding bias and fairness in AI systems. In light of these concerns, the following questions emerged: Should START Foundation integrate the AI-based recommender system into its digital learning platform? If so, what factors should START Foundation consider regarding bias and fairness in its AI-based recommender system in order to prevent negative consequences for its scholars? Readers are encouraged to explore the interactive learning module on bias and fairness in AI systems, available at https://bias-and-fairness-in-ai-systems.de/en/home/.

An intelligent recommender system for people who are prone to fatty liver disease

IntroductionConsidering the growth of the outbreak of fatty liver disease, it seems necessary and needed to design and develop an intelligent recommender system in order to provide clinical recommendations and a healthy lifestyle and also to provide a suitable educational and counseling platform. In this regard, the present research was conducted in order to design, develop and evaluate an intelligent recommender system for people who are prone to non-alcoholic fatty liver disease. Materials and methodsThe current research was of the applied-developmental type which was carried out in order to create a non-alcoholic fatty liver intelligent recommender system. In the stage of the knowledge base designing, the selected data elements were placed in the form of tables in the SQL software. In the stage of the system designing, first the physical and functional features of the system were drawn by UML diagram, then coding was done using C# programming language. At the final stage, the usability evaluation of the designed system was assessed using the System Usability Scale (SUS). ResultsThe data requirements and technical capabilities were identified in three areas: demographic, clinical and technical requirements. In order to create a recommender system, the Microsoft SQL Server database and C# programming languages were used in the asp.net environment in a reactive manner. The fatty liver intelligent recommender system designed in this study had an average score of 74.625 in the evaluation of SUS applicability. Therefore, the usability of the fatty liver intelligent recommender system was approved by experts. ConclusionIn this study, through the designed system, the researchers provided a platform to remotely communicate between the patient and the specialist and receive unlimited self-care recommendations. Investigating the effectiveness of using the current recommender system to advance treatment goals and follow up on the patient's condition is suggested for future studies.

A Big-Data Oriented Recommendation Method in E-learning Environment

E-learning is increasingly gaining popularity in organizational and institutional learning for its several benefits to learn anywhere, anytime, and anyplace. Therefore, explosive growth of E-learning has caused difficulty of locating appropriate learning activities for learner in this context, and it becomes relatively widespread learning method for learner. Several research in e-learning mainly focused on improving learner achievements based on recommendation technique. An ideal recommender system in e-learning environment should be built with both accurate and learning goals. To address this challenge, we propose a recommendation method based on machine learning technique. Based on this tool, a learning approach is designed to achieve personalized learning experiences by selecting the most appropriate learning activities. Moreover, some experiments were conducted to evaluate the performance of our approach. The results demonstrate that our method outperforms other state-of-the-art methods and reveals suitability of using recommender system in order to support online learning activities to enhance learning.

Context-Aware Service Recommendation Based on Knowledge Graph Embedding

Over two decades, context awareness has been incorporated into recommender systems in order to provide, not only the top-rated items to consumers but also the ones that are suitable to the user context. As a class of context-aware systems, context-aware service recommendation (CASR) aims to bind high-quality services to users, while taking into account their context requirements, including invocation time, location, social profiles, connectivity, and so on. However, current CASR approaches are not scalable with the huge amount of service data (QoS and context information, users reviews and feedbacks). In addition, they lack a rich representation of contextual information, as they adopt a simple matrix view. Moreover, current CASR approaches adopt the traditional user-service relation and they do not allow for multi-relational interactions between users and services in different contexts. To offer a scalable and context-sensitive service recommendation with great analysis and learning capabilities, we provide a rich and multi-relational representation of the CASR knowledge, based on the concept of knowledge graph. The constructed <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">context-aware service knowledge graph</i> (C-SKG) is, then, transformed into a low-dimensional vector space to facilitate its processing. For this purpose, we adopt Dilated Recurrent Neural Networks to propose a context-aware knowledge graph embedding, based on the principles of first-order and subgraph-aware proximity. Finally, a recommendation algorithm is defined to deliver the top-rated services according to the target user's context. Experiments have proved the accuracy and scalability of our solution, compared to state-of-the-art CASR approaches.

Contextual Information Retrieval within Recommender System: Case Study "E-learning System"

This paper focuses on monitoring and analyzing user activities on collaborative filtering -based recommender system in order to guess suitable and unsuitable items' context information using rating matrix which makes more efficient adaptation task. An ontology-based user profile and rules-based context modeling for reasoning about context information is proposed in this research work, in addition to an investigation to apply Semantic Web technologies in user modeling and context reasoning. This proposal is applied in education field in which we have designed an authoring tool for learning objects within ubiquitous environment. This system aims to improve the learning object production task (creation, review, edition…) on behalf of technologies offered by collaborative filtering systems as well as user behaviors monitoring to improve the recommendation process.

SCOL: similarity and credibility-based approach for opinion leaders detection in collaborative filtering-based recommender systems

Recommender systems (RSs) have recently gained significant attention from both research and industrial communities. These systems generate the recommendations of items in one of two ways, namely collaborative or content-based filtering. Collaborative filtering is a technique used by recommender systems in order to suggest to the user a set of items based on the opinions of other users who share with him the same preferences. One of the key issues in collaborative filtering systems (CFSs) is how to generate adequate recommendations for newcomers who rate only a small number of items, a problem known as cold start user. Another interesting problem is the cold start item when a new item is introduced in the system and cannot be recommended. In this paper, we present a clustering-based approach SCOL that aims to alleviate the cold start challenges; by identifying the most effective opinion leaders among the social network of the CFS. SCOL clustering focuses on the credibility and correlation similarity concepts.

SCOL: similarity and credibility-based approach for opinion leaders detection in collaborative filtering-based recommender systems

Recommender systems (RSs) have recently gained significant attention from both research and industrial communities. These systems generate the recommendations of items in one of two ways, namely collaborative or content-based filtering. Collaborative filtering is a technique used by recommender systems in order to suggest to the user a set of items based on the opinions of other users who share with him the same preferences. One of the key issues in collaborative filtering systems (CFSs) is how to generate adequate recommendations for newcomers who rate only a small number of items, a problem known as cold start user. Another interesting problem is the cold start item when a new item is introduced in the system and cannot be recommended. In this paper, we present a clustering-based approach SCOL that aims to alleviate the cold start challenges; by identifying the most effective opinion leaders among the social network of the CFS. SCOL clustering focuses on the credibility and correlation similarity concepts.

Robust Model-Based Reliability Approach to Tackle Shilling Attacks in Collaborative Filtering Recommender Systems

As the use of recommender systems becomes generalized in society, the interest in varying the orientation of their recommendations is increasing. There are shilling attacks' strategies that introduce malicious profiles in collaborative filtering recommender systems in order to promote the own products or services or to discredit those of the competition. Academic research against shilling attacks has been focused in statistical approaches to detect the unusual patterns in user ratings. Nowadays, there is a growing research area focused on the design of robust machine learning methods to neutralize the malicious profiles inserted into the system. This paper proposes an innovative robust method, based on matrix factorization, to neutralize the shilling attacks. Our method obtains the reliability value associated with each prediction of a user to an item. By monitoring the unusual reliability variations in the items prediction, we can avoid promoting the shilling predictions to the erroneous recommendations. This paper openly provides more than 13 000 individual experiments involving a wide range of attack strategies, both push, and nuke, in order to test the proposed approach. The results show that the proposed method is able to neutralize most of the existing attacks; its performance only decreases in the not relevant situations: when the attack size is not large enough to effectively affect the recommendations provided by the system.

BlindDate recommender: A context-aware ontology-based dating recommendation platform

Online dating sites have become popular platforms for those individuals who utilise the Internet to develop a personal or romantic relationship. Unlike typical recommenders systems, which attempt to suggest items such as films, songs, books and so on. According to a user’s interests, dating recommender systems provide services that people can use to find potential romantic partners. Since these services have a higher expectancy of users, online dating sites are considering the introduction of recommender systems in order to build an improved dating network. Different kinds of techniques based on content-based, collaborative filtering or hybrid techniques exist. In this article, we introduce BlindDate recommender, a context-based platform that utilises semantic technologies to describe users’ preferences more precisely. We utilise DBPedia repositories to obtain information that is subsequently used to enrich a previously generated ontology model. The instances inserted into the ontology enable the matching algorithms that we have generated to identify potential matches between users. In order to validate the performance of the platform, we utilise a real-world data set that has produced relevant results enhancing the accuracy compared with other well-known approaches and identifying the discriminant parameters used in the dating domain. More specifically, the proposed approach attains 0.79, 0.8 and 0.55 in the I-Precision, I-Recall and I-F-measure, respectively, when employed in separate topics.

A Recommender System for Library Based on Hadoop Ecosystem

With the rapid development of network and information technology, big data already exists in various fields. Its main features include large amount of data (Volume), speed (Velocity) and diversity (Variety). The information system of library also has the characteristics of big data which have a large number of collected books, readers, and diverse resources. Because of the era of big data, libraries need to assist the reader to rapidly get the appropriate information from a large number of collected resources in the library information system. It is an important issue to establish the suitable recommender system in order to recommend readers to the related library resources and to enhance the usage of library resources. In this paper, a recommender system for library based on Hadoop Ecosystem is proposed. First, it uses Hadoop ecosystem to import data from the information system of library. Furthermore, the association rule and parallel CLARANS are implemented in Hadoop platform for the library recommender system. The proposed approach can assist readers to find the suitable resources in library.

A Recommender Model in E-learning Environment

Various researches in E-learning mainly focused on improving learner achievements based on learner profile. Explosive growth of distance learning has caused difficulty of locating appropriate learning objects for learner in this environment, and it becomes relatively widespread learning method for learner. In this paper, an innovative learning approach is proposed by using recommender system to address this challenge. Based on this tool, a learning model is designed to achieve personalized learning experiences by selecting and sequencing the most appropriate learning objects. Moreover, some experiments were conducted to evaluate the performance of our approach. The result reveals suitability of using recommender system in order to support online learning activities to enhance learning.

An Architecture for Developing Educational Recommender Systems

The great amount of educational resources available on educational repositories enriches the learning process. However, it raises a new challenge: the need to provide support to the location of those resources that meet the needs, goals and preferences of each student. The location of useful educational resources to support the learning process is addressed by using recommender systems. Recommender systems are a tool to help student find information quickly and recommend new items of interest to the active student based on their preferences. In this paper, we propose a generic architecture for developing educational recommender systems independent of the type of recommendation generated. Also, we identify main features of an educational recommender system. These features are important components to achieve the objectives that have educational recommender systems in order to provide accurate information to students according to their preferences, user profile and learning objectives.

Implications of Perceived Utility on Individual Choice and Preferences: A New Framework for Designing Recommender System

ABSTRACT Consumer psychology and consumer behaviour has constantly been a field of interest for the researchers. Volumes of researches are available in this area, still new theories and concepts keep emerging in response to change in context and environment. Consumer psychology has large cross disciplinary implications. In this paper, a relationship between consumer psychology and recommender system has been explained and how implementing a consumer psychology model can improve recommender system performance. This paper also gives first level analysis of various filters that can be developed along with the design of recommender system in order to generate a more refined and relevant choice sets. There are attraction effects observed among different items and the attractiveness of a product is codependent on attractiveness of other options available in a choice set. In the present paper, we have explored the utility of defender model in the design of recommender systems. Different effects like decoy and asymmetric dominance are also analyzed to achieve better efficiency and effectiveness in the design of recommender system. Keywords: Decoy, Recommender, Defender Model, Asymmetric Dominance INTRODUCTION Recommender systems are the category of information systems which supports searching and suggesting products and services according to user's requirement and preferences (Resnick et al., 1997). The product range may vary from consumer durables such as a book in Amazon or digital goods such as song on lastfm. The task of recommending products proactively is performed by recommender systems which work backstage on E-commerce websites. Many recommendation approaches and methods are applied to target internet based consumers. Collaborative filtering was earliest approach to create user to user similarity and then recommend similar products. Recommendation approach was simple. Positively rated items by one set of user were recommended to similar users. However, this 'user to user' correlation lead to scarce data and cold start problems (Ahn, 2008). This problem was targeted by designing an 'item to item' based correlation and this category of algorithms were termed as content based filtering as volume of data available for products was much denser than user data. There were some specialized variations also for example demographic or knowledge based recommender system which includes domain based knowledge to analyze user preference and predict user choices. As it is evident that recommender system directly interact with consumer and for establishing a better communication it is essential that recommender systems should be intelligent enough to understand consumer's behavioral traits(Pu et al., 2011). Understanding consumer psychology and its phenomenon will actually present meaningful insights to improve recommendations and effectiveness of these systems which in turn can be translated into economical benefits. The purpose of the recommender system is not only to elicit preference and produce accurate search results but also to improve the decision making in an ecommerce context (Xiao et al., 2007). A good amount of research in the recommender system focuses on developing complex algorithms and software programs which improves the efficiency of recommender systems (Cosley et al., 2003). However, there is a need to identify other significant aspects which attempts to maximize the value derived from a recommender system experience (Xiao et al., 2007). Factors such as trust on online medium (Yaobin et al., 2007) and personal choices (Ho et al., 2008) are often ignored in the design and implementation process. .In the present paper, we propose a conceptual framework based on established theories from the field of consumer psychology. The proposed framework allows the systems designer to incorporate the significance of derived value of a given product/service in order to achieve the single most important objective of any decision behaviour-maximizing utility. …

Evaluation and selection of group recommendation strategies for collaborative searching of learning objects

Nowadays, there is a wide variety of e-learning repositories that provide digital resources for education in the form of learning objects. Some of these systems provide recommender systems in order to help users in the search for and selection of the learning objects most appropriate to their individual needs. The search for and recommendation of learning objects are usually viewed as a solitary and individual task. However, a collaborative search can be more effective than an individual search in some situations – for example, when developing a digital course between a group of instructors. The problem of recommending learning objects to a group of users or instructors is much more difficult than the traditional problem of recommending to only one individual. To resolve this problem, this paper proposes a collaborative methodology for searching, selecting, rating and recommending learning objects. Additionally, voting aggregation strategies and meta-learning techniques are used in order to automatically obtain the final ratings without having to reach a consensus between all the instructors. A functional model has been implemented within the DELPHOS hybrid recommender system. Finally, various experiments have been carried out using 50 different groups in order to validate the proposed learning object group recommendation approach.

Ubiquitous recommender systems

Ubiquitous recommender systems combine characteristics from ubiquitous systems and recommender systems in order to provide personalized recommendations to users in ubiquitous environments. Although not a new research area, ubiquitous recommender systems research has not yet been reviewed and classified in terms of ubiquitous research and recommender systems research, in order to deeply comprehend its nature, characteristics, relevant issues and challenges. It is our belief that ubiquitous recommenders can nowadays take advantage of the progress mobile phone technology has made in identifying items around, as well as utilize the faster wireless connections and the endless capabilities of modern mobile devices in order to provide users with more personalized and context-aware recommendations on location to aid them with their task at hand. This work focuses on ubiquitous recommender systems, while a brief analysis of the two fundamental areas from which they emerged, ubiquitous computing and recommender systems research is also conducted. Related work is provided, followed by a classification schema and a discussion about the correlation of ubiquitous recommenders with classic ubiquitous systems and recommender systems: similarities inevitably exist, however their fundamental differences are crucial. The paper concludes by proposing UbiCARS: a new class of ubiquitous recommender systems that will combine characteristics from ubiquitous systems and context-aware recommender systems in order to utilize multidimensional context modeling techniques not previously met in ubiquitous recommender systems.

Data Mining Techniques and Preference Learning in Recommender Systems

The importance of implementing recommender systems has significantly increased during the last decade. The majority of available recommender systems do not offer clients the ability to make selections based on their choices or desires. This has motivated the development of a web based recommender system in order to recommend products to users and customers. The new system is an extension of an online application previously developed for online shopping under constraints and preferences. In this work, the system is enhanced by introducing a learning component to learn user preferences and suggests products based on them. More precisely, the new component learns from other customers’ preferences and makes a set of recommendations using data mining techiques including classification, association rules and cluster analysis techniques. The results of experimental tests, conducted to evaluate the performance of this component when compiling a list of recommendations, are very promising.

Integrating high-level requirements in optimization problems: theory and applications

We mix Mathematical Programming (MP) and Systems Engineering (SE) modelling methods and integrate system-level requirements in optimization problems. We use MP, in the first phases of the design process, as a common methodology to model prototypes that fulfil high-level requirements, consistently with the holistic, top-down approach of SE. This approach is applied to three different kinds of system. (1)Information systems, i.e. the network of communication channels, hardware, software and trained people used within an enterprise to help planning and control, are required to provide a basis for numerous IT projects that are launched in order to respond to the needs of the business. The information system shall ensure profitable services and it shall be maintainable and extensible. We propose an operational model and a mathematical programming formulation expressing a generic global prioritization problem occurring in the context of a technological evolution of an information system. (2)Recommender systems (RS) are a kind of search engine and aim to provide personalized recommendations. We consider the problem of designing recommender systems in order to provide good, interesting and accurate suggestions. (3)The transportation of hazardous materials (hazmat) entails several issues due to the environmental consequences of possible accidents. The transportation system shall ensure safe disposal of hazardous waste in such a way that the risk of potential catastrophic accident is equitably distributed over the population. We consider and integrate in MP formulations two different ideas of equity.

Defending recommender systems by influence analysis

Collaborative filtering (CF) is a popular method for personalizing product recommendations for e-commerce applications. In order to recommend a product to a user and predict that user's preference, CF utilizes product evaluation ratings of like-minded users. The process of finding like-minded users forms a social network among all users and each link between two users represents an implicit connection between them. Users having more connections with others are the most influential users. Attacking recommender systems is a new issue for these systems. Here, an attacker tries to manipulate a recommender system in order to change the recommendation output according to her wish. If an attacker succeeds, her profile is used over and over again by the recommender system, making her an influential user. In this study, we applied the established attack detection methods to the influential users, instead of the whole user set, to improve their attack detection performance. Experiments were conducted using the same settings previously used to test the established methods. The results showed that the proposed influence-based method had better detection performance and improved the stability of a recommender system for most attack scenarios. It performed considerably better than established detection methods for attacks that inserted low numbers of attack profiles (20---25 %).

Collective intelligence as mechanism of medical diagnosis: The iPixel approach

Collective intelligence (CI) is an active field of research, which capitalizes the knowledge of human collectives in order to create, to innovate and to invent. There are two important mechanisms to implement CI: recommender and reputation systems. Recommender systems are used to provide filtered information from a large amount of elements. The recommendations are intended to provide interesting elements to users. Recommendation systems can be developed using different techniques and algorithms where the selection of these techniques depends on the area in which they will be applied. This work presents iPixel Recommender Engine, which is focused on the medical field. iPixel Recommendation Engine supports the process of differential diagnosis by recommending mammographic evaluations. Each mammogram is collectively tagged by the users’ community with a semantic sense; this feature allows iPixel acquires collective knowledge. iPixel can associate more than one feature with each mammogram. This work also presents a qualitative evaluation, where the basic features that a recommendation system should have in the medical field were obtained. Finally, a comparison was carried out with other similar recommender systems in order to know the Pixel advantages.

E-LORS: Uma Abordagem para Recomendação de Objetos de Aprendizagem

The use of recommender systems in order to improve human-computer interaction has increased significantly in applications as e-commerce and e-learning. Such systems aim at suggesting information and services that can be attractive to the users based on the automatic use of data that describe their preferences. In the educational area, the recommendation of relevant and interesting contents can be effective for catching the students' attention and driving higher motivation during the learning-teaching process. The ability to provide contents of their preferences considering the students' learning style may increase substantially the success of such effort. In this context, the goal of this work is to present the e-LORS system, an approach for recommendation of electronic content based on the relationship between learning styles and learning objects. In our methodology, the learning profiles are described according to discrete dimensions in order to support different perspectives concerning the students' preferences. Our recommendation methodology, in turn, uses these dimensions to filter the learning objects – described by the IEEE LOM standard – that are more suitable to the students.