TY  - BOOK
AU  - Samantha Gwyn M. Aranzamendez, Joshua Caleb D. Bolito, Aron Christopher R. Rafe.
AU  - 
ED  - 
ED  - 
ED  - 
ED  - 
SN  - 2
PY  - 4541///346
CY  - 
PB  - 
KW  - 
KW  - 2
KW  - 0
KW  - 6
KW  - 20
N1  - Undergraduate Thesis : (Bachelor of Science in Computer Science) - Pamantasan ng Lungsod ng Maynila, 2024; 5
N2  - ABSTRACT:  Content-based Filtering is a recommender system that provides recommendations based on the description of an item a user interacted with. It is used in various consumer domains to personalize recommedantions to users.  Despite its relevant functionality, Content-based Filtering is riddled with limitations as well. Among these limitations of the recommendations provided by Content-based Filtering are overspecialization - where the algorithm recommends those items that are directly related to the user which rules out those newer sets of items, cold-start - where the algorithm cannot produce appropriate recommendations for new users of the system that have no and/or limited rated features in their profile, and synonymy issues - where the algorithm lacks the ability  to distinguish semantic relationships between words. In this study, an enhanced Content-based Filtering was developed which addresses the issue of overspecialization, cold-start, and synonymy issues by using Maximal Marginal Relevance, Temporary Preference Profile, and FastText.  Results show that the proposed enhancement with n=0.5 showed the most prominence among the enhanced variants of Content-based Filtering, having a good balance between relevance and diversity of recommendation which on average, improves upon the original algorithm in terms of Precision by 45.83%, in terms of Recall by 24.45%, in terms of F-Score by 38.51%, and in terms of Diversity by 276.79%
ER  -