Recommendation Services based on Business Processes Analysis for Online Stores

Recommender systems for E-commerce are software service that helps customers to find the most relevant product according to customers’ needs. Recommender systems have significantly changed the way users of online stores search and choose products. Recommendations are made on the basis of user’s profile: sex, age, interests, search history, purchasing habits, etc. The following paper examines the existing problems in the field of E-commerce concerning recommender systems. The purpose of research is to examine the work of recommender services for E-commerce and test one of the third-party software solutions, implementing product recommendations for online stores, on efficiency. To achieve the goal, several tasks were committed. Firstly, studying and classifying recommender systems, as well as identifying strengths and weaknesses of such systems. Secondly, it was decided to estimate one of the recommender system’s effectiveness. Chapter I concerns the main problems in this area, as well as an overview of the recommending tools used in online retailers, such as Amazon, which is one of the major leaders in the field of E-commerce. Chapter II scrutinizes some of the existing mathematical algorithms implemented in recommender systems, as well as methods of collection, processing and interpretation of data. One of the paragraphs includes overview of some well-known recommender service providers. An overview of metrics to evaluate the effectiveness of the recommendations in the E-commerce is described on eBay example in the final part of the chapter. In the final chapter of the thesis an experiment was carried out. One of the recommender systems was installed into an online-shop. Changes of the key indicators of its work were analyzed by A / B testing method. In conclusion of the experiment, the practical usefulness of embedded recommender system was assessed. Processing of the data demonstrate the effect of the recommender system in this shop.