Μελέτη συστήματος συστάσεων βασιζόμενο στην έκπληξη των χρηστών
View/
Keywords
Serendipity ; Diversity ; Coverage ; Python ; Recommendation systems ; Συστήματα συστάσεωνAbstract
Search engines and a host of other applications provide recommendations based on user preferences and correlations. Bias and fairness in machine learning techniques are issues that have aroused the interest of researchers.
Good prediction accuracy alone does not guarantee users an effective and satisfying experience. In this regard, another factor that seems to play an important role in whether users value the recommendation system is serendipity (unexpectedly pleasant surprise).
In this postgraduate thesis we will study different approaches to fairness as well as the implementation of a political fairness in recommendation systems. Fairness may be related to the fair treatment of users by the system in relation to the quality of the recommendations it provides and / or the fair distribution of the proposed objects (no bias in the groups of objects proposed).
The aim of the postgraduate thesis is to design and implement an approach that takes into account justice in the recommendation process as well as serendipity. In particular, the system will calculate the serendipity for new movies that the user has not seen and make the appropriate suggestions. In terms of fairness, the movies that the system will monitor and recommend are of different categories and not just those that the user is used to watching.
Postgraduate Studies Programme
Ψηφιακά Συστήματα και ΥπηρεσίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών ΣυστημάτωνNumber of pages
47Language
GreekCollections
Except where otherwise noted, this item's license is described as
Attribution-NonCommercial-NoDerivatives 4.0 Διεθνές
Attribution-NonCommercial-NoDerivatives 4.0 Διεθνές