Γραφοθεωρητικές μέθοδοι κοινωνικής σύστασης
Social recommendation: a graph – based approach
KeywordsΕκτίμηση άγνωστης βαθμολογίας ; Αλγόριθμοι ; Βάσεις δεδομένων ; Επεξεργασία δεδομένων ; Αξιολόγηση
The scope of this thesis is the development of an algorithm which can find the possible rating that a user of a rating website would give to a product, depending on various factors such as the real given ratings of the user’s friends and the correlation value between the users whether they are friends or not. With this method, we can look into the preferences of a user, to group the user into bigger users’ groups with a higher correlation value as also to do more correct product or activities suggestions to users. Specifically, the development of a tool which can provide us with the data that we would need from the website that we have selected and then submit the data into our database, was necessary. The website that we have selected and fulfills the requirements of our research is https://www.goodreads.com/ . This website hosts a global books catalogue. The registered users of this website can select the books that they have read and rate them, they can select the books that they would like to read and they can become friends with other users. So, by covering the three main requirements needed for the realization of this research, list of products, ratings and friendship, this website was selected. The present thesis, initially, presents the way – tool used in order to collect the data needed for the research and the development and application of two algorithms to the collected data.