Unit:
Τομέας Θεωρητικής ΠληροφορικήςLibrary of the School of Science
Author:
Δρόσου Αικατερίνη
Supervisors info:
Μαρία Γρηγοριάδου Ομότιμος Καθηγήτρια , Παναγιώτης Γεωργιάδης Ομότιμος Καθηγητής , Διονύσιος Μάργαρης Διδάκτωρ
Original Title:
Συγκριτική μελέτη μετρικών απόστασης για σύστημα προτάσεων
Translated title:
Distance metrics comparison for recommender systems
Summary:
This thesis's target is to make the optimal recommendation of a product or
service, according to the users' preferences. Its source code uses stored
information concerning the ratings of selected products and its test
scenario concerns a movie recommendation system.
As an input it uses only two plain text files. The first one contains
movies rated by users, while the second one contains the category that
each movie belongs to. The initial target is to preprocess these files and
to evaluate which metric is the best option, as far as recommendation is
concerned, according to each user’s rating background.
This thesis consists of six chapters. The first one presents the
theoretical background underlying our study and the tools used, while the
second chapter introduces the scenario concerning our study. The third
chapter analyzes the procedure of data preprocessing, the next chapter
presents the metrics to be used, while the fifth chapter analyzes each
metric’s results. Finally, the last chapter evaluates the results and
presents the conclusions of our study.
Keywords:
Personalization, Recommender systems, Proximity metrics, Database, Aging factor
Number of index pages:
6-10
File:
File access is restricted.
document.pdf
845 KB
File access is restricted.