Recommender Research Articles

SecK2 – A novel machine learning algorithm for detecting data poisoning attacks

Machine learning (ML) techniques play a crucial role in producing precise predictions without the use of explicit programming by utilizing representative and unbiased data. These methods, which are a subset of artificial intelligence (AI), are used in a variety of settings, including recommendation engines, spam filtering, malware detection, classification, and predictive maintenance. While ML algorithms improve results, they also present security and privacy threats, especially in the face of adversarial ML attacks such as data poisoning assaults that can undermine data modeling applications. This study introduces SecK2, a cutting-edge ML method developed to stop dangerous input from entering ML models. The scalability of SecK2 is proved through meticulous experimental research, revealing its astonishing capacity to identify data poisoning attacks at a previously unheard-of pace. As a result, SecK2 becomes a valuable tool for guaranteeing the reliability and security of ML models. Our suggested method produces outstanding results by a variety of criteria. Notably, it achieves a noteworthy 61% convergence rate and an exceptional 89% attack detection rate. Additionally, it offers a phenomenal 96% throughput while protecting data integrity at 53%. The technique also boasts impressive Validation accuracy of 96% and Training accuracy of 92%. The suggested technology offers a strong and reliable barrier against the rising danger of data poisoning attacks. ML practitioners can have more faith in their models, thanks to SecK2’s capabilities, protecting against potential adversarial assaults and preserving the dependability of ML-based applications.

Leveraging artificial intelligence and mutual authentication to optimize content caching in edge data centers

Available online Edge data centers are designed to meet the stringent QoE requirements of delay-sensitive and computationally intensive services in Content Delivery Network (CDN) and 5G networks. The primary purpose of this paper was to formulate and solve the problem of optimizing many control variables jointly: (i) what contents to store by taking into consideration edge capacity, and (ii) what contents to recommend to each Internet of Everything (IoE) item, based on identity and access management (IAM). In reactive caching policy, we proposed a new Two-Factor Authentication (2FA) scheme founded upon the Elliptic Curve Cryptography (ECC) and one-way hash function for access control. More interestingly, we use Non-negative Matrix Factorization (NMF), Fuzzy C-Means (FCM), Random Forest (RF) and Pearson Correlation (PC) to improve the accuracy and latency of traditional data filtering models. The intelligent recommendation engine we propose is designed to be implemented by cloud for caching and prefetching contents at the edge. The experimental results validate the theoretical guarantees of the proposed solution and its ability to achieve significant performance gains compared to common baseline models.

Text2Color Networks: Deep Learning Models for Color Generation from Compositional Color Descriptions

Color serves as an important cue in graphics, arts and in many computer vision applications. We reliably and effortlessly name colors and communicate them with their names. However, many applications such as graphics, color design, palette generation and color selection tools demand numerical values of colors. Predicting and communicating colors by their numerical values is less intuitive and difficult task as it is a mapping of millions of colors for a specific color space. To bridge the gap between linguistic color names and numerical color values, in this paper, we present neural network architectures that predict a point in color space for a given color name. Proposed models provide user interface between colors and names by learning the language semantics and mimic human level comprehension of color descriptions to predict colors and modify them with respect to linguistic adjectives of color names. We consider color prediction as a regression problem and solve it as a language modeling task. Color descriptions are taken as text sequences and each sentence is represented with word-level tokenization. Each token is transformed into a word vector in the latent space using CBOW word embeddings model. Word vectors representing color names are fed as input to neural networks and trained with normalized R, G, B values as supervision information. Trained models are capable of predicting color for a given color name and modify colors for different nouns and adjectives associated with color names. We also built color generation models based on pre-trained word embeddings to overcome the limited availability of large linguistic color name datasets. These pre-trained models perform well with datasets containing few thousand color names. We then present two recommendation engines that suggest similar color palette to user given color name. These recommendation engines enhance the color vocabulary and assist users in the color selection process.

Offloading Machine Learning to Programmable Data Planes: A Systematic Survey

The demand for machine learning (ML) has increased significantly in recent decades, enabling several applications, such as speech recognition, computer vision, and recommendation engines. As applications become more sophisticated, the models trained become more complex while also increasing the amount of data used for training. Several domain-specific techniques can be helpful to scale machine learning to large amounts of data and more complex models. Among the methods employed, of particular interest is offloading machine learning functionality to the network infrastructure, which is enabled by the use of emerging programmable data plane hardware, such as SmartNICs and programmable switches. As such, offloading machine learning to programmable network hardware has attracted considerable attention from the research community in the last few years. This survey presents a study of programmable data planes applied to machine learning, also highlighting how in-network computing is helping to speed up machine learning applications. In this article, we provide various concepts and propose a taxonomy to classify existing research. Next, we systematically review the literature that offloads machine learning functionality to programmable data plane devices, classifying it based on our proposed taxonomy. Finally, we discuss open challenges in the field and suggest directions for future research.

Graph-Based Extractive Text Summarization Sentence Scoring Scheme for Big Data Applications

The recent advancements in big data and natural language processing (NLP) have necessitated proficient text mining (TM) schemes that can interpret and analyze voluminous textual data. Text summarization (TS) acts as an essential pillar within recommendation engines. Despite the prevalent use of abstractive techniques in TS, an anticipated shift towards a graph-based extractive TS (ETS) scheme is becoming apparent. The models, although simpler and less resource-intensive, are key in assessing reviews and feedback on products or services. Nonetheless, current methodologies have not fully resolved concerns surrounding complexity, adaptability, and computational demands. Thus, we propose our scheme, GETS, utilizing a graph-based model to forge connections among words and sentences through statistical procedures. The structure encompasses a post-processing stage that includes graph-based sentence clustering. Employing the Apache Spark framework, the scheme is designed for parallel execution, making it adaptable to real-world applications. For evaluation, we selected 500 documents from the WikiHow and Opinosis datasets, categorized them into five classes, and applied the recall-oriented understudying gisting evaluation (ROUGE) parameters for comparison with measures ROUGE-1, 2, and L. The results include recall scores of 0.3942, 0.0952, and 0.3436 for ROUGE-1, 2, and L, respectively (when using the clustered approach). Through a juxtaposition with existing models such as BERTEXT (with 3-gram, 4-gram) and MATCHSUM, our scheme has demonstrated notable improvements, substantiating its applicability and effectiveness in real-world scenarios.

A Recommendation System Based on Implicit Data for Internet Protocol Television (IPTV)

IPTV delivers television content over Internet Protocol (IP) networks. Videos On Demand (VOD) is the most popular IPTV, allowing users to freely select from a vast pool of program genres. Therefore, it is necessary to introduce innovative features to attract new users and retain existing ones. For this purpose, IPTV systems typically use VOD recommendation engines. The primary purpose of recommendation systems is to suggest user-relevant items from various items by producing a list of recommendations for each user. In this paper, we introduce an approach to recommendation systems in IPTV. We developed this approach on implicit feedback derived from users’ interaction with movies/series sets, such as how many times they watched a movie and how long they have spent watching specific movies/series. For the previous factors, we tested a variety of recommendation algorithms, content-based, collaborative-based, and hybrid. Then applied the previously mentioned algorithms on real-life big data sets after introducing some modifications to the algorithms, then benchmarked the results on multiple performance metrics. We noticed that the applied changes achieved promising results.

초등학생을 위한 LightFM 기반의 개인별 맞춤 도서 추천 엔진 개발 및 성능평가

Objectives In this study, a personalized book system is designed for elementary school students based on machine learning technology, and a hybrid recommendation engine is developed. In addition, the purpose is to verify the performance through performance analysis of the developed recommended engine. Methods To achieve the above purpose, a toaping reading service with 9,000 members was used. 5,979,479 reading and reading activities were targeted, and a book recommendation system was developed using LightFM. The ranking for the activities was applied by applying the weight of each activity to the processing of Korean and various reading activities were applied. Results Performance analysis of the developed system was conducted. Performance analysis was conducted on AUC Score, Accuracy, Precison, Recall, and F1-Score, and as a result, 99.9%, 95.1%, 92.5%, 86.3%, and 89.3%, respectively. It was confirmed that the recommended book reflects the student's interest. Conclusions Through the results of the performance analysis, it was confirmed that the books selected by individual students and the books recommended by the developed recommendation engine were similar. It was confirmed that for elementary school students, book recommendations according to each student's area of interest should be made in addition to book recommendations by experts or related organizations. In addition, in order to improve the completeness of the system developed through this study, additional analysis and reflection of big data on book loans by public institutions are needed, and various additional studies are needed to recommend customized books.

TwinPort: 5G drone-assisted data collection with digital twin for smart seaports

Numerous ports worldwide are adopting automation to boost productivity and modernize their operations. At this point, smart ports become a more important paradigm for handling increasing cargo volumes and increasing operational efficiency. In fact, as ports become more congested and cargo volumes increase, the need for accurate navigation through seaports is more pronounced to avoid collisions and the resulting consequences. To this end, digital twin (DT) technology in the fifth-generation (5G) networks and drone-assisted data collection can be combined to provide precise ship maneuvering. In this paper, we propose a DT model using drone-assisted data collection architecture, called TwinPort, to offer a comprehensive port management system for smart seaports. We also present a recommendation engine to ensure accurate ship navigation within a smart port during the docking process. The experimental results reveal that our solution improves the trajectory performance by approaching the desired shortest path. Moreover, our solution supports significantly reducing financial costs and protecting the environment by reducing fuel consumption.

Improving recommendation diversity and serendipity with an ontology-based algorithm for cold start environments

AbstractEvery real-life environments where users interact with items (products, films, research expert profiles) have several development phases. In the Cold-start phase, there are almost no interactions among users and items content-based recommendation systems (RS) can only recommend based on matching the attributes of the items. In the transition state, items start to collect user interactions but still a significant number of items have too small number of interactions, RS does not allow users to discover cold items. In a regular state, where most of the items in the system have enough interactions, the recommendations often suffer from low diversity of the items within a single recommendation. This article proposes a general recommendation algorithm based on Ontological-similarity, which is designed to address all the above problems. Our experiments show that recommendations generated by our approach are consistently better in all environment development phases and increase the success rate of recommendations by almost 50% measured using ontology-aware recall, which is also introduced in this article.

A Modern Communication Path for Traditional Chinese Cultural Design Concepts Based on AdaBoost Model

Abstract Clarifying the integration and communication path of Chinese traditional cultural design concepts and modern art design is beneficial to inherit Chinese traditional culture and innovate modern art design. This paper proposes an SD-AdaBoost classification model based on the AdaBoost algorithm, optimizes the algorithm’s design using the sample density method, and conducts a comparative performance analysis for the model. The SD-AdaBoost classification model is used to mine two research items, application path, and dissemination mode, for the fusion of Chinese traditional cultural concepts and modern art design communication, and relevant data analysis is presented. As far as the application carriers are concerned, they embody different values, but all reflect traditional Chinese cultural ideas. Regarding dissemination methods, social media, recommendation engines, and advertising are better ways to disseminate traditional Chinese cultural ideas and modern art design, accounting for 46.96%, 47.92%, and 45.47%, respectively. Based on the AdaBoost model, we can analyze the current dissemination path of Chinese traditional culture design ideas and grasp how we can better integrate and develop Chinese traditional culture and modern art and design ideas.

Business-Driven Data Recommender System: Design and Implementation

ABSTRACT Self-Service Business Intelligence (SSBI) increases decision-making reactivity of companies by facilitating the data use by non-IT experts. An important SSBI dimension is data querying where businesspeople create their own queries by reducing the technical complexity of formal languages like SQL. However, existing solutions ignore two other key challenges of data querying identified in the literature: the databases technical jargon and the data overload. In this paper, we propose, following the Design Science Research methodology, a framework (i.e. DatAssistant) to complement existing querying solutions with two new theoretical artifacts. The first bridges the semantic gap between technical databases and businesspeople via a business-aware ontology of the Data Warehouse mapped to the business Data Catalog. The second artifact filters data overload by mobilizing a hybrid recommender engine combining semantic systems and business rules. This paper then demonstrates the validity and applicability of the framework through its technical implementation in a real-world environment.

Individualized Movie Recommendation Algorithm Based on Machine Learning

With the continuous acceleration of economic growth and the remarkable improvement of people's living standards, people's individualized needs have increased. Moreover, due to the growth of computer technology and network technology, many forms of entertainment and services have been integrated into the network. The recommendation engine actively pushes the information that meets the user's taste and preference to the user by analyzing the user's behavior data on the website when the user's needs are not clear. This article will study how to apply the deep learning algorithm to the field of movie recommendation system to solve the problems existing in the traditional recommendation algorithm based on machine learning (ML), and combine the existing mainstream big data (BD) to improve the execution efficiency of the model. Individualized movie recommendation system saves users the time of searching, viewing and screening, and discovers users' potential movie preferences at the same time. Most users think that the algorithm can help them find the type they want from a large quantity of film and television resources, so as to meet the needs of users and tap their potential interests.

Precursor recommendation for inorganic synthesis by machine learning materials similarity from scientific literature.

Synthesis prediction is a key accelerator for the rapid design of advanced materials. However, determining synthesis variables such as the choice of precursor materials is challenging for inorganic materials because the sequence of reactions during heating is not well understood. In this work, we use a knowledge base of 29,900 solid-state synthesis recipes, text-mined from the scientific literature, to automatically learn which precursors to recommend for the synthesis of a novel target material. The data-driven approach learns chemical similarity of materials and refers the synthesis of a new target to precedent synthesis procedures of similar materials, mimicking human synthesis design. When proposing five precursor sets for each of 2654 unseen test target materials, the recommendation strategy achieves a success rate of at least 82%. Our approach captures decades of heuristic synthesis data in a mathematical form, making it accessible for use in recommendation engines and autonomous laboratories.

A visual-semantic approach for building content-based recommender systems

The topic of recommending items based on multimodal content has been addressed to a limited extent, and yet this could be a potential solution to the data bottleneck problem. Content-based semantics-driven recommender systems are often applied in the small-scale news recommendation domain, founded on the TF-IDF measure but also taking into account domain semantics through semantic lexicons or ontologies. In this work, we explore the application of content-based semantics-driven recommender systems to large-scale recommendations and focus on using both textual information and visual information to recommend items that have multimodal content. We propose methods to extract semantic features from various item descriptions, including digital images. In particular, we use computer vision to extract visual-semantic features from images and use these for movie recommendation together with various features extracted from textual information. The visual-semantic approach is scaled up with pre-computation of the cosine similarities and gradient learning of the model. The results of the study on a large-scale MovieLens dataset of user ratings demonstrate that semantics-driven recommenders can be extended to visual-semantic recommenders suitable for more complex domains than news recommendation, and which outperform TF-IDF-based recommenders on ROC, PR, F1, and Kappa metrics.

Content-Based Recommender Systems Taxonomy

Abstract In the era of internet access, recommender systems try to alleviate the difficulty consumers face while trying to find items (e.g. services, products, or information) that better match their needs. To do so, a recommender system selects and proposes (possibly unknown) items that may be of interest to some candidate consumer, by predicting her/his preference for this item. Given the diversity of needs between consumers and the enormous variety of items to be recommended, a large set of approaches have been proposed by the research community. This paper provides a review of the approaches proposed in the entire research area of content-based recommender systems, and not only in one part of it. To facilitate understanding, we provide a categorization of each approach based on the tools and techniques employed, which results to the main contribution of this paper, a content-based recommender systems taxonomy. This way, the reader acquires a quick and complete understanding of this research area. Finally, we provide a comparison of content-based recommender systems according to their ability to efficiently handle well-known drawbacks.

The Construction of Cultural Translation Research System under Big Data Technology

Abstract In the context of big data development, one of the reasons for the ineffectiveness of Chinese cultural translation research is the lack of attention to the “system construction”, this paper focuses on the construction of the cultural translation research system under the big data technology, based on the MOOC recommendation system using the recommendation engine group for the hybrid recommendation, the recommendation engine group integrated the use of the mainstream collaborative filtering algorithm based on foreign translation literature in the recommendation algorithm. Through the experimental analysis, the model performance reaches the maximum when the number of cultural literature is 7, and the F_1 of the traditional method is improved by about 2.6%. Compared with the original recommendation algorithm, the improved recommendation algorithm of the MOOC implicit scoring model can get more accurate recommendation results to promote the construction of a Chinese cultural foreign translation discourse system. It is expected to promote the construction of Chinese cultural foreign translation discourse system, show the vitality, staying power, and influence of China’s excellent culture to the world, enhance the international discourse power and cultural soft power of Chinese culture in the new era, and contribute Chinese wisdom to global cultural exchange and governance.

Food Dish Pedictor using YOLOv4

Abstract: In recent years, the demand for automated food recognition systems has increased due to the large variety of dishes available especially in Indian cuisine and also due to the growing awareness of the importance of a healthy diet. Identifying multiple food items from an image is a challenging task, especially when it comes to Indian cuisine, which is known for its diverse range of dishes and ingredients. The goal of this paper is to create a food dish predictor for five common Indian cuisines utilizing the cutting-edge object detection algorithm - YOLOv4. The five foods selected for this study are Aloo paratha, Biryani, Poha, Khichdi, and Chapati. A dataset is created by taking the images of selected dishes from various platforms such as social media and forming classes, which was then used to train the YOLOv4 model. To accurately train the machine to identify the dishes, the dataset was manually labeled. The YOLOv4-based food dish predictor that has been proposed could be put to use in a number of applications. The predictor can assure proper order fulfillment in food delivery services by recognising food plates in real-time from customer-provided photos. The predictor's capacity to automatically extract dish information can assist menu recognition software, saving restaurant personnel time and allowing for rapid menu revisions. Meal recommendation engines can utilize the predictor to provide personalized meal choices based on user preferences and previous eating experiences. Overall, the YOLOv4-based food dish predictor helps multiple sectors of the food business to increase productivity, consumer experiences, and decision-making capacities. The findings of this investigation show how well YOLOv4 can recognise and identify food items, particularly Indian food items.

A Novel Classical Music Recommendation System Using User Facial Expression

Abstract: In today's world, feelings play an important part in many aspects of life. The spectrum of human feelings, whether or not those sensations are verbalised, serves as the foundation for emotions. A diverse range of emotional expressions can be used to convey a person's sense of individuality. It is possible to infer a person's state of mind simply by observing their facial expressions. This project aims to determine people's feelings based on images of them and then match those feelings with suitable musical selections. The majority of today's most popular music discovery tools make use of a social or content-based recommendation engine of some kind. However, a listener's taste in music is dependent on a variety of factors in addition to their previous listening habits and the subject matter of the songs they hear. The feelings of other users are also important.

Understanding the Effects of Personalized Recommender Systems on Political News Perceptions: A Comparison of Content-Based, Collaborative, and Editorial Choice-Based News Recommender System

ABSTRACT With the increasing implementation of algorithms across various news platforms, understanding news consumers’ subjective perceptions of algorithmic-based news recommender systems has become critical. A between-subjects experiment (News Recommender System type: content-based filtering vs. collaborative filtering vs. human editorial choice-based recommender system) with 161 participants revealed that participants tended to trust the collaborative filtering system and perceive news recommended by the system to be more credible and less biased compared to editorial choices-based or content-based recommender systems – due to the triggering of the homophily heuristic – even though the three systems recommended the same set of news. Implications were discussed.

Fast and Frictionless: A Novel Approach to Radiology Appointment Scheduling Using a Mobile App and Recommendation Engine.

Many outpatient radiology orders are never scheduled, which can result in adverse outcomes. Digital appointment self-scheduling provides convenience, but utilization has been low. The purpose of this study was to develop a "frictionless" scheduling tool and evaluate the impact on utilization. The existing institutional radiology scheduling app was configured to accommodate a frictionless workflow. A recommendation engine used patient residence, past and future appointment data to generate three optimal appointment suggestions. For eligible frictionless orders, recommendations were sent in a text message. Other orders received either a text message for the non-frictionless app scheduling approach or a call-to-schedule text. Scheduling rates by type of text message and scheduling workflow were analyzed. Baseline data for a 3-month period prior to the launch of frictionless scheduling showed that 17% of orders that received an order notification text were scheduled using the app. In an 11-month period after the launch of frictionless scheduling, the rate of app scheduling was greater for orders that received a text message with recommendations (frictionless approach) versus app schedulable orders that received a text without recommendations (29% vs. 14%, p < 0.01). Thirty-nine percent of the orders that received a frictionless text and scheduled using the app used a recommendation. The most common recommendation rules chosen for scheduling included location preference of prior appointments (52%). Among appointments that were scheduled using a day or time preference, 64% were based on a rule using the time of the day. This study showed that frictionless scheduling wasassociated with an increased rate of app scheduling.