Content-based Filtering Approach Research Articles

A Reliable Prediction Algorithm Based on Genre2Vec for Item-Side Cold-Start Problems in Recommender Systems with Smart Contracts

Personalized recommender systems are used not only in e-commerce companies but also in various web applications. These systems conventionally use collaborative filtering (CF) and content-based filtering approaches. CF operates using memory-based or model-based methods; both methods use a user-item matrix that considers user preferences as items. This matrix denotes information on user preferences, which refers to the user ratings for items. The model-based method exploits the fact that the input matrix is factorized. CF approaches can effectively provide personalized recommendation results to users; however, cold-start problems arise because both these methods depend on the users’ ratings for items to predict users’ preferences. We proposed an approach to alleviate the cold-start problem along with a methodology for utilizing blockchain that can enhance the reliability of the processes of the recommendations. We attempted to predict an average rating for a new item to alleviate item-side cold-start problems. First, we applied the concept of word2vec, treating each user’s item-selection history as a sentence. Then, we derived genre2Vec based on the skip-gram technique and predicted an average rating for a new item by utilizing the vectors and category ratings. We experimentally demonstrated that our approach could generate more accurate results than conventional CF approaches could. We also designed the processes of the recommendation based on the concept of blockchain addressing the smart contract. Based on our approach, we proposed a system that can secure reliability as well as alleviate the cold-start problems in recommender systems.

MOVIE RECOMMENDER SYSTEM USING MACHINE LEARNING

The recommendation system of today has transformed the way we search for items of our interest through the use of an information filtering approach that predicts user preferences. A movie recommendation system named RECOM has been proposed in this paper. Based on the content-based filtering approach, it utilizes the information in the dataset, undergoes analysis, and recommends the most suitable movies for the user. The recommended movie list is ordered according to the IMDB ratings calculated by a standard formula. The system also enables users to search for movies based on their favourite actors/actresses. Developed in Python and Machine Learning, the recommender system generates recommendations through the utilization of various forms of knowledge and data about movies, such as vote count, vote average, mean, quantile, etc. Overall, the effectiveness and efficacy of movie searches have significantly increased thanks to the introduction of content-based filtering and machine learning techniques in movie recommendation systems, making it simpler for users to locate films they are likely to enjoy.

A Manhattan distance based hybrid recommendation system

Many online service providers use a recommendation system to assist their customers' decision-making by generating recommendations. Accordingly, this paper proposes a new recommendation system for tourism customers to make online reservations for hotels with the features they need, saving customers time and increasing the impact of personalized hotel recommendations. This new system combined collaborative and content-based filtering approaches and created a new hybrid recommendation system. Two datasets containing customer information and hotel features were analyzed by Recency, Frequency, Monetary (RFM) method in order to identify customers according to their purchasing nature. The main idea of the recommendation system is to establish correlations between users and products and make the decision to choose the most suitable product or information for a particular user. As a result of the exponential growth of online data, this vast amount of information for use in the tourism industry can be leveraged by decision-makers to make purchasing decisions[20]. Filtering, prioritizing, and beneficially presenting relevant information reduces this overload. There are following three main ways that recommendation systems can generate a recommendation list for a user; content-based, collaborative-based, and hybrid approaches[1]. This paper describes each category and its techniques in detail. RFM Analysis is used to identify customer segments by measuring customers' purchasing habits. It is the process of labeling customers by determining the Recency, Frequency, and Monetary values of their purchases and ranking them on a scoring model. Scoring is based on how recently they bought (Recency), how often they bought (Frequency), and purchase size (Monetary). Experimental results show that the accuracy of behavior analysis using Manhattan distance-based hybrid filtering is greatly improved compared to collaborative and content-based algorithms.

Anime Based on Content-Based Filtering

Anime comes from the word "Animation", nowadays it is very popular among people, mostly kids and teenagers. They are being streamed on large platforms such as Netflix and Amazon Prime. Millions of people watch anime nowadays regularly. It has a huge fan base. Due to this increment in the popularity of anime, there is a requirement for a recommendation system that can suggest user watch the best-suited anime for the user. The recommendation system is used to predict the preference of the user based on user interest. In this paper, we have proposed an anime recommendation system. It is based on a content-based filtering approach that makes use of the information provided by users, analyzes them and then recommends the different anime that is best suited to the user. This method helps users to find anime that they are interested in and can watch without concern about the content of the show. They don't have to ask other people for suggestions about the anime and they can have a lot of options at the same place. The recommended anime list is sorted according to the ratings given to these animations by the user that is stored in the dataset and it uses various algorithms such as content-based filtering and sectorization for this purpose. It doesn't allow users to waste time because of its efficient and effective technique. This system has been developed in Jupyter Notebook with the help of python and machine learning algorithms as the backend and for the front end python idle is used with various libraries.

Context-Aware Practice Problem Recommendation Using Learners’ Skill Level Navigation Patterns

The use of programming online judges (POJs) has risen dramatically in recent years, owing to the fact that the auto-evaluation of codes during practice motivates students to learn programming. Since POJs have greater number of programming problems in their repository, learners experience information overload. Recommender systems are a common solution to information overload. Current recommender systems used in e-learning platforms are inadequate for POJ since recommendations should consider learners’ current context, like learning goals and current skill level (topic knowledge and difficulty level). To overcome the issue, we propose a context-aware practice problem recommender system based on learners’ skill level navigation patterns. Our system initially performs skill level navigation pattern mining to discover frequent skill level navigations in the POJ and to find learners’ learning goals. Collaborative filtering (CF) and content-based filtering approaches are employed to recommend problems in the current and next skill levels based on frequent skill level navigation patterns. The sequence similarity measure is used to find the top k neighbors based on the sequence of problems solved by the learners. The experiment results based on the real-world POJ dataset show that our approach considering the learners’ current skill level and learning goals outperforms the other approaches in practice problem recommender systems.

Hotel Recommendation System with Content-Based Filtering Approach (Case Study: Hotel in Yogyakarta on Nusatrip Website)

The increasing of Covid-19 pandemic has hampered people's activities, often causing stress if they are only stay at home continuously. This has led to an increasing trend of staycations or holiday activities in the city itself by renting a hotel. Hotel rental technology has begun to be transferred with the existence of OTA (Online Travel Agent). The existence of various kinds of hotels with various kinds of facilities that makes people feel confused in choosing which hotel to occupy. To help overcome this, the researchers tried to create a recommendation system to help prospective hotel residents choose the hotel according to their choice. In addition, it can also assist companies in increasing hotel room reservations through its website. In this study, researchers will build a hotel recommendation system in Yogyakarta at one of the OTAs in Indonesia using Content-Based Filtering Methods, weighting text data using Term Frequency-Inverse Document Frequency (TF-IDF) Methods and measuring document similarity using Cosine Similarity Methods. Based on the results of the Good Karma Yogyakarta hotel recommendations as a test example, 10 similar hotels were obtained, namely Happy Buddha Yogyakarta – Hostel, Nextdoor Homestay, Hotel Puspita, OYO 426 Hotel Gading Resto, Omah Jegog Homestay, Prawirotaman Homestay, RedDoorz near Prawirotaman, Ayodhya Garden Hostel Yogyakarta by HOM, Bringin House Yogyakarta, and House 24 Yogyakarta with cosine similarity values 0.956666513, 0.946570717, 0.917459394, 0.912534877, 0.886439718, 0.88221982, 0.881775275, 0.875845794, 0.872030219, and 0.871514859.

Movie Recommendation System Using Content-Based Filtering with Heroku Deployment

Abstract: Now days peoples watching movies to make their busy life calm , watching movie is not that easy as it takes a lot of time to select a movie due to large data set of movies exist in the world and it is very difficult and time consuming process to people to select a movie due to large data set but we can achieve this with the help of movie recommendation system using content based filtering ,where this recommendation system recommend a movie to the user based on the content he/she is watching. In this paper I am building a recommendation system using content- based filtering with Heroku deployment. Generally, recommendation systems are build using content -based filtering approach or collaborative filtering approach here in this paper I developed recommendation system using content-based filtering approach with Heroku Deployment. Heroku Deployment enables easy deployment of application.

Hybrid content and collaborative filtering based recommendation system for e-learning platforms

Recommendation systems, although a well-studied topic, experience several shortcomings when applied on e-learning platforms. While collaborative filtering methods have enjoyed great success in making recommendations on large scale e-commerce and social networking and observation, users of e-learning platforms have continually evolving preferences, which render collaborative filtering methods weak. On the other end of the spectrum are content-based filtering approaches. Although such methods are more suited for e-learning platforms, the primary concern is that these methods find it hard to generalize across content sources and content types. In this work, we present a hybrid recommendation system that combines the desirable characteristics of collaborative filtering, as well as from content-based filtering, for the task of recommending course content/curriculum to users of an e-learning system. Our recommendation easily incorporates changing user profiles (as learners step through course content) and also generalize across content sources (courses taught by various departments) and types. We apply our system on a real dataset comprising 111 students organized into interdisciplinary groups. Our results showcase the clear benefits that our hybrid recommendation system enjoys, showing more than 30 percentage points of improvement over conventional filtering techniques.

An Improved Recommender System Solution to Mitigate the Over-Specialization Problem Using Genetic Algorithms

Nowadays, recommendation systems offer a method of facilitating the user’s desire. It is useful for recommending items from a variety of areas such as in the e-commerce, medical, education, tourism, and industry domains. The e-commerce area represents the most active research we found, which assists users in locating the things they want. A recommender system can also provide users with helpful knowledge about things that could be of interest. Sometimes, the user gets bored with recommendations which are similar to their profiles, which leads to the over-specialization problem. Over-specialization is caused by limited content data, under which content-based recommendation algorithms suggest goods directly related to the customer profile rather than new things. In this study, we are particularly interested in recommending surprising, new, and unexpected items that may likely be enjoyed by users and will mitigate this limited content. In order to recommend novel and serendipitous items along with familiar items, we need to introduce additional hacks and note of randomness, which can be achieved using genetic algorithms that brings diversity to recommendations being made. This paper describes a Revolutionary Recommender System using a Genetic Algorithm called RRSGA which improves the fitness functions for recommending optimal results. The proposed approach employs a genetic algorithm to address the over-specialization issue of content-based filtering. The proposed method aims to incorporate genetic algorithms that bring variety to recommendations and efficiently adjust and suggest unpredictable and innovative things to the user. Experiments objectively demonstrate that our technology can recommend additional products that every consumer is likely to appreciate. The results of RRSGA have been compared against recommendation results from the content-based filtering approach. The results indicate the effectiveness of RRSGA and its capacity to make more accurate predictions than alternative approaches.

WEB PAGE RECOMMENDATION SYSTEM BY INTEGRATING ONTOLOGY AND STEMMING ALGORITHM

In this research, we offer a customized-recommendation system that uses item representations and user profiles based on the ontologies that provide personalized services to semantic applications. To develop and implement the personalized-recommendation system, a system that uses the representations of the items and the user profiles based on the ontologies to provide the semantic applications with personalized services. Recommendation systems can use semantic reasoning capabilities to overcome present system limits and increase the quality of recommendations. The recommender makes use of domain ontologies to improve personalization: on the one hand, a domain-based inference method is used to model user interests more effectively and accurately; on the other hand, a semantic similarity method is used to improve the stemmer algorithm, which is used by our content-based filtering approach, which provides a measure of the affinity between an item and the user. In recommender systems and web personalization, Web Usage Mining is crucial. This study presents an effective recommender system based on ontology and web usage mining. The approach's first step is to extract features from online documents and build on related ideas. Then, they create an ontology for the website using the concepts and relevant terms retrieved from the records. The semantic similarity of web documents is used to group them into multiple semantic themes, each with its own set of preferences. The suggested solution incorporates ontology and semantic knowledge into Web Usage Mining and personalization procedures, as well as a stemming algorithm, and gets an overall accuracy of 90%.

What books will be your bestseller? A machine learning approach with Amazon Kindle

PurposeWith the rapid increase in internet use, most people tend to purchase books through online stores. Several such stores also provide book recommendations for buyer convenience, and both collaborative and content-based filtering approaches have been widely used for building these recommendation systems. However, both approaches have significant limitations, including cold start and data sparsity. To overcome these limitations, this study aims to investigate whether user satisfaction can be predicted based on easily accessible book descriptions.Design/methodology/approachThe authors collected a large-scale Kindle Books data set containing book descriptions and ratings, and calculated whether a specific book will receive a high rating. For this purpose, several feature representation methods (bag-of-words, term frequency–inverse document frequency [TF-IDF] and Word2vec) and machine learning classifiers (logistic regression, random forest, naive Bayes and support vector machine) were used.FindingsThe used classifiers show substantial accuracy in predicting reader satisfaction. Among them, the random forest classifier combined with the TF-IDF feature representation method exhibited the highest accuracy at 96.09%.Originality/valueThis study revealed that user satisfaction can be predicted based on book descriptions and shed light on the limitations of existing recommendation systems. Further, both practical and theoretical implications have been discussed.

Spam Detection and Spammer Behaviour Analysis in Twitter Using Content Based Filtering Approach

Twitter is one of the most popular social media networks and therefore it is prone to misuse. One of the ways in which people misuse Twitter is by spamming. Spam becomes an issue once a communication medium especially one, which enables global communication and handle huge amount of online data. Since Twitter is popular among so many people, it makes it easy for spammers to thrive. Spammers are people who send unwanted messages to people to either advertise a product or lure the victims into clicking malicious links, which may affect their user systems. The main objective of these spammers is usually to make money from their victims. In the last years, several systems has-been made with the aim of determining whether a user is a spammer or not. However, these systems cannot filter each spam message and a different account can be created by a spammer and used to send other messages. This paper proposes a content-based approach, which can be used to filter spam tweets. The approach involves using tweets in machine learning and compression algorithms in order to filter the undesired tweets.

Hybrid Recommender System for Music Information Retrieval

Music plays an integral role in our lives as the most popular type of recreation. With the advent of new technologies such as Internet and portable media players, large amount of music data is available online which can be distributed and easily made available to people. Enormous amount of music data is released every year by several artists with songs varying in features, genre and so on. Because of this, a need for reliable and easy access of songs based on user preferences is necessary. The recommender system focuses on generating playlists based on the physical, perceptual and acoustical properties of the song (content based filtering approach), or on commonalities between users on a particular basis like ratings or user data history (collaborative filtering). The system thus developed is a hybrid music recommender tool which creates a user centric suggestion system accompanied by feature extraction which in turn enhances the accuracy of music recommendations.

CBRec: a book recommendation system for children using the matrix factorisation and content-based filtering approaches

Promoting good reading habits among children is essential, given the enormous influence of reading on students' development as learners and members of the society. Unfortunately, very few (children) websites or online applications recommend books to children, even though they can play a significant role in encouraging children to read. Given that a few popular book websites suggest books to children based on the popularity of books or rankings on books, they are not customised/personalised for each individual user and likely recommend books that users do not want or like. We have integrated the matrix factorisation approach and the content-based approach, in addition to predicting the grade levels of books, to recommend books for children. Recent research works have demonstrated that a hybrid approach, which combines different filtering approaches, is more effective in making recommendations. Conducted empirical study has verified the effectiveness of our proposed children book recommendation system.

HCF-CRS: A Hybrid Content based Fuzzy Conformal Recommender System for providing recommendations with confidence.

A Recommender System (RS) is an intelligent system that assists users in finding the items of their interest (e.g. books, movies, music) by preventing them to go through huge piles of data available online. In an effort to overcome the data sparsity issue in recommender systems, this research incorporates a content based filtering technique with fuzzy inference system and a conformal prediction approach introducing a new framework called Hybrid Content based Fuzzy Conformal Recommender System (HCF-CRS). The proposed framework is implemented to be used in the domain of movies and it provides quality recommendations to users with a confidence level and an improved accuracy. In our proposed framework, first, a Content Based Filtering (CBF) technique is applied to create a user profile by considering the history of each user. CBF is useful in the situations like: lack of demographic information and the data sparsity problems. Second, a Fuzzy based technique is incorporated to find the similarities and differences between the user profile and the movies in the dataset using a set of fuzzy rules to get a predicted rating for each movie. Third, a Conformal prediction algorithm is implemented to calculate the non-conformity measure between the predicted ratings produced by fuzzy system and the actual ratings from the dataset. A p-value (confidence measure) is computed to give a level of confidence to each recommended item and a bound is set on the confidence level called a significance level ε, according to which the movies only above the specified significance level are recommended to user. By building a confidence centric hybrid conformal recommender system using the content based filtering approach with fuzzy logic and conformal prediction algorithm, the reliability and the accuracy of the system is considerably enhanced. The experiments are evaluated on MovieLens and Movie Tweetings datasets for recommending movies to the users and they are compared with other state-of-the-art recommender systems. Finally, the results confirm that the proposed algorithms perform better than the traditional ones.

CBRec: a book recommendation system for children using the matrix factorisation and content-based filtering approaches

Promoting good reading habits among children is essential, given the enormous influence of reading on students' development as learners and members of the society. Unfortunately, very few (children) websites or online applications recommend books to children, even though they can play a significant role in encouraging children to read. Given that a few popular book websites suggest books to children based on the popularity of books or rankings on books, they are not customised/personalised for each individual user and likely recommend books that users do not want or like. We have integrated the matrix factorisation approach and the content-based approach, in addition to predicting the grade levels of books, to recommend books for children. Recent research works have demonstrated that a hybrid approach, which combines different filtering approaches, is more effective in making recommendations. Conducted empirical study has verified the effectiveness of our proposed children book recommendation system.

A Prototype for an e-Recruitment Platform using Semantic Web Technologies

1 IntroductionSince the late 1990s when the labour market faced important economy changes and had a high demand for qualified candidates, online web recruitment domain has grown steadily, surpassing the traditional methods. An e-Recruitment solution offers a significant gain in terms of efficiency by speeding up the recruitment process and saving time to all those involved. A job recommender system is supposed to help the recruiters, by recommending the most eligible candidates for a particular job and, on the other hand, to propose jobs to the aspiring candidates, matching their profiles with the existing job offers.It is difficult for a candidate to analyse all the open positions from one or more companies and then select from them those that really fit his profile. In many cases it can occur that, although the candidate possesses all the required abilities for a job, he may not be aware of the job existence. Automatic matching between the skills and abilities that a company requires for a specific job and those an applicant has, helps decrease the errors that can occur when the process is done manually (having in mind the huge amount of CVs that need to pass by a screening process before they are deep checked). What is important to note is that these e-Recruitment platforms are assisting and not replacing the recruiters in their decision-making process.Moreover, using an online recruitment system to connect people with job opportunities has also other benefits. For example, it provides all the needed information to do a comprehensive analysis about the labour market, including what are the skills most of the employers are seeking, what is the level of training and expertise of the candidates, what are the salary expectations of the applicants and so on.The rest of the paper is organized as follows. The next section presents a literature review about the different approaches used to implement a job recommender system. In section 3 we propose an ontology for the IT recruitment domain that helps to automatically match job offers with the candidates' profiles and in reverse, designed to simplify and streamline the recruitment process. Section 4 presents an architecture that uses the defined ontology to implement an e-Recruitment platform, based on Java technologies. In section 5 we illustrate the main features of the implemented prototype of the job recommender system. Finally, section 6 presents the conclusions to our work and the future work remarks, which can improve and extend the presented solution.2Related WorkIn the existing literature [1] [2] there are four main categories used to classify recommender systems:* Collaborative Filtering (CF) - uses the user-to-user correlation method to predict the unknown preferences of a new user based on the preferences of similar users;* Content-Based Filtering (CBF) - recommends items having similar content with the one the user has previously viewed or selected;* Knowledge-Based Approach - makes suggestions based on inferences about the needs and preferences of the user;* Hybrid Approach - combines one or more of the already mentioned methods to obtain better performance.Developing a strong job recommender system is definitely not an easy task. There have always been different problems that needed to be solved. The first encountered problem that matching systems had was the amount of semi-structured data which contained many free text fields remaining unfilled. [3] proposes using Structured Relevance Models (SRM) as a solution for the fields that are missing when a candidate fills an online CV. The authors also emphasize that a blank field value can be inferred based on the context the other fields in the record provide. The SRM is a first example of a Collaborative Filtering approach.Further studies in the e-Recruitment domain suggested developing an automated recommendation system based on supervised machine learning [4], technique that uses the Content-Based Filtering approach. …

Research-paper recommender systems: a literature survey

In the last 16 years, more than 200 research articles were published about research-paper recommender systems. We reviewed these articles and present some descriptive statistics in this paper, as well as a discussion about the major advancements and shortcomings and an overview of the most common recommendation concepts and approaches. We found that more than half of the recommendation approaches applied content-based filtering (55 %). Collaborative filtering was applied by only 18 % of the reviewed approaches, and graph-based recommendations by 16 %. Other recommendation concepts included stereotyping, item-centric recommendations, and hybrid recommendations. The content-based filtering approaches mainly utilized papers that the users had authored, tagged, browsed, or downloaded. TF-IDF was the most frequently applied weighting scheme. In addition to simple terms, n-grams, topics, and citations were utilized to model users' information needs. Our review revealed some shortcomings of the current research. First, it remains unclear which recommendation concepts and approaches are the most promising. For instance, researchers reported different results on the performance of content-based and collaborative filtering. Sometimes content-based filtering performed better than collaborative filtering and sometimes it performed worse. We identified three potential reasons for the ambiguity of the results. (A) Several evaluations had limitations. They were based on strongly pruned datasets, few participants in user studies, or did not use appropriate baselines. (B) Some authors provided little information about their algorithms, which makes it difficult to re-implement the approaches. Consequently, researchers use different implementations of the same recommendations approaches, which might lead to variations in the results. (C) We speculated that minor variations in datasets, algorithms, or user populations inevitably lead to strong variations in the performance of the approaches. Hence, finding the most promising approaches is a challenge. As a second limitation, we noted that many authors neglected to take into account factors other than accuracy, for example overall user satisfaction. In addition, most approaches (81 %) neglected the user-modeling process and did not infer information automatically but let users provide keywords, text snippets, or a single paper as input. Information on runtime was provided for 10 % of the approaches. Finally, few research papers had an impact on research-paper recommender systems in practice. We also identified a lack of authority and long-term research interest in the field: 73 % of the authors published no more than one paper on research-paper recommender systems, and there was little cooperation among different co-author groups. We concluded that several actions could improve the research landscape: developing a common evaluation framework, agreement on the information to include in research papers, a stronger focus on non-accuracy aspects and user modeling, a platform for researchers to exchange information, and an open-source framework that bundles the available recommendation approaches.

Infinite Relational Model 기반 Co-Clustering을 이용한 영화 추천

사람의 영화에 대한 선호도에는 개인의 특성과 영화의 속성을 기반으로 하는 다양한 요인이 연관되어 있다. 영화 추천을 위한 사용자-영화-선호도 연관 관계의 분석 기법으로서, 다중 개념 탐색 기법의 특성을 지닌 infinite relational model (IRM)의 활용 가능성을 확인하고, 이를 기초로 영화 선호 유형에 따른 사용자-영화 군집을 탐색한다. 별점으로 표현되는 명시적인 선호도 데이터에 영화 컨텐츠 관련 메타데이터를 추가하여 학습 데이터를 구성하고, 이에 IRM을 적용하여 공군집화(co-clustering)를 수행한 결과, 해석 가능한 다양한 명시적 연관 관계를 발견하였다. 공군집화 결과를 기초로 개인화 추천에서의 다양한 활용 방안을 논의한다. Preferences of users on movies are observables of various factors that are related with user attributes and movie features. For movie recommendation, analysis methods for relation among users, movies, and preference patterns are mandatory. As a relational analysis tool, we focus on the Infinite Relational Model (IRM) which was introduced as a tool for multiple concept search. We show that IRM-based co-clustering on preference patterns and movie descriptors can be used as the first tool for movie recommender methods, especially content-based filtering approaches. By introducing a set of well-defined tag sets for movies and doing three-way co-clustering on a movie-rating matrix and a movie-tag matrix, we discovered various explainable relations among users and movies. We suggest various usages of IRM-based co-clustering, espcially, for incremental and dynamic recommender systems.

An Empirical Evaluation Of Learner Performance In E-Learning Recommender Systems And An Adaptive Hypermedia System

This paper introduces a novel architecture for an e-learning recommender system which is based on good learners’ average ratings strategy and content-based filtering approach. The feasibility of the proposed system is conducted by comparing its performance against other recommender systems and an adaptive hypermedia system in order to measure the effectiveness of the proposed strategy in improving students’ learning performance. Experimental result has shown that the recommender strategy can improve students’ performance by at least 12.16%, as compared to other recommendation techniques. A performance evaluation with an adaptive hypermedia system that uses knowledge level as its adaptation feature also showed a positive increase of 14.99% in terms of students’ performance.