The analysis of recommender systems: Attacks, sentiment analysis, evaluation and hybrid methods

Abstract

Attacks on recommender systems varies depending on the type of the recommender system being targeted. This paper will focus on the methods of attacks against Content-based and Collaborative Filtering recommender system. Sentiment Analysis is an important branch in the discipline of natural language processing, which is widely used in analyzing public opinion and recommendation systems. It is a computational study of the emotions, opinions and attitudes expressed by people about products, services, organizations, individuals, events, topics, etc. Sentiment analysis is a common application of natural language processing (NLP). Sentiment Analysis, also known as tendency analysis or opinion mining, is an important information analysis and processing technology. Its research purpose is to automatically mine the position, viewpoint, opinion, emotion, likes and dislikes in the text. Evaluating recommender systems refers to the process of assessing the quality and effectiveness of recommendation algorithms. This involves determining how well the systems are able to provide accurate and relevant recommendations to users, as well as measuring other important metrics such as coverage, diversity, and novelty. Hybrid recommender systems aim to address these limitations by combining different methods to make inferences more powerful. These systems can leverage multiple data sources, such as user behavior, item attributes, and contextual data, to offer personalized recommendations that meet the diverse needs of users. By using a combination of methods, hybrid recommender systems can overcome the limitations of individual algorithms and provide more effective recommendations.