System of monitoring and analysis of reviews in social networks for purchase recommendations
Students Name: Tverdokhlib Yurii Petrovych
Qualification Level: magister
Speciality: Data Science
Institute: Institute of Computer Science and Information Technologies
Mode of Study: full
Academic Year: 2022-2023 н.р.
Language of Defence: ukrainian
Abstract: The rapid growth of data volumes and the number of Internet visitors has created the problem of information redundancy, which prevents timely access to the necessary resources. Information and search systems partially solved this problem, but prioritization and personalization of information were missing. This has contributed to a significant demand for recommender systems. They are needed to help users when the system needs to make a decision, based on the fact that in everyday life people often make decisions based on the recommendations of others [1]. The product purchase recommendation system will be used by customers to find a suitable product. The choice of which is extremely important, because by choosing a bad product you can lose money and waste a lot of time, so the recommendation system will help to avoid unwanted incidents, namely, it will analyze product reviews and build a list of recommended products based on them. Goal of research is to create a system for monitoring and analyzing feedback in social networks to form recommendations for the purchase of goods aimed at facilitating the selection of the desired product. The main tasks that are solved in the work are: • description, analysis of perspectives and features of the studied subject area; • conducting an analysis of known means of solving the problem; • system analysis and modeling of the subject area; • models construction, selection and justification of problem solving methods; • design, development and testing of a platform for a system of monitoring and analysis of reviews in social networks for purchase recommendations; • economic analysis of the software product. Study object – the process of forming recommendations for the purchase of goods. The subject of the study is the process of monitoring and means of reviews analysis for the formation of recommendations for the purchase of goods. System of monitoring and analysis will be used by customers to speed up and facilitate the search for the necessary products on e-commerce resources. The successful selection of a quality product is extremely important, as it saves time and money in the search. Analyzing comments on the network, the information system recommends the product if there is a preponderance of positive feedback on it. In the work, a system analysis was carried out for the system of monitoring and analysis of reviews in social networks, a conceptual model was developed. Modeling of the project requirements was carried out, a tree of goals was built, the criteria put forward when defining goals were determined, and goals were set. Using the method of analysis of hierarchies, it was determined that the type of product under development is a decision support system. Systems used in the field of e-commerce were used as a prototype of the system for generating recommendations based on analysis of reviews. For example, the recommendation system of the online store Rozetka, which is able to intelligently analyze and predict the preferences of customers to offer them a list of recommended products. However, Rozetka’s algorithm selects recommended products for each user based on their previous purchases, interactions and ratings of other featured products and matches them with similar products viewed by users with similar preferences and interests. The basis of the developed decision-making system is the sentiment analysis algorithm [2] using logistic regression. Logistic regression is a classification model that is very easy to implement and works very well on linearly separable classes. It is one of the most common classification algorithms, which makes it attractive to work with. Logistic regression is a good model because it learns very quickly despite large data sets and guarantees very reliable results. The main advantage of logistic regression is that it is much easier to set up and train than other machine learning and artificial intelligence programs. Another advantage is that it is one of the most efficient algorithms when the different results or differences represented by the data are linearly separated. This means that you can draw a straight line separating the results of the logistic regression calculation. The created system will allow every person who is looking to buy a product to receive a quality recommendation in a matter of seconds, which will save the time spent on searching and will allow finding the product according to the desired criteria. The following effects should be expected from such a recommendation system: • economical - the system will help customers save money by recommending cheaper and equally high-quality products; • functional – automation of product search according to specified criteria based on analysis of product reviews; • financial - it is planned that the use of the system will be paid, which will bring some income to the developers; • time-saving for a person looking for a product, because without this system, it can take up to several hours. Keywords - social networks, monitoring system, analysis of feedback, formation of recommendations, sentiment analysis algorithm. References. 1. Recommender Systems: Algorithms and Applications / P. Pavan Kumar, S. Vairachilai, P. Sirisha, S. Nandan Mohanty. – Boca Raton, London, New York: CRC Press, 2021. – 248 с. 2. Aakanksha S. New Opportunities for Sentiment Analysis and Information Processing / S. Aakanksha, G. R. Sinha, S. Bhatia, 2021. – 311 с.