A serendipity oriented recommendation system
View/
Keywords
Information retrieval ; Recommender systems ; SerendipityAbstract
This thesis explores the concept of serendipity and the challenges associated with incorporating into
recommendation systems. Through the study of related research in the field and the detailed analysis of
the available dataset, three features were generated to be used for predicting a defined serendipity score
based on real user feedback in movie recommendations. A Random Forest model was selected, trained,
and evaluated on the generated dataset. The results offer insights into the potential of using predictive
models for reranking recommendation lists to maximize serendipity for users. Through this work, it is
hoped to provide a contribution towards a better understanding of serendipity and offer a starting point
for future work in this area.
Postgraduate Studies Programme
Πληροφοριακά Συστήματα και ΥπηρεσίεςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Ψηφιακών ΣυστημάτωνNumber of pages
41Language
EnglishCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα