3js : cspc online dataset repository with an integrated recommender system using content based filtering algorithm /
Estallo, Joshua P.
3js : cspc online dataset repository with an integrated recommender system using content based filtering algorithm / Joshua P. Estallo, Janben Aldrich H. Quiapo, and Jan Dale M. Parañal - 101 leaves.
Undergraduate Thesis (BSCS) - Computer Science.
Includes bibliographic references.
The significance of data collection in research is crucial, as it provides essential in-formation for hypothesis testing and drawing meaningful conclusions. Novice researchers often face challenges in accessing relevant datasets, impacting their work significantly due to issues like accessibility, completeness, and biases within datasets, leading to errors. This study aims to enhance user experience by developing a prototype dataset repository using content-based filtering algorithm on a centralized web platform. Utilizing a Python pro-gram with TF-IDF and Cosine Similarity, the system achieved a high user rating of 0.9481, effectively handling file uploads and resolving dataset query indexing. With an average pre-cision and recall score of 0.60, the system accurately predicts and delivers relevant datasets, emphasizing the precision in analyzing metadata to meet user needs. Researchers advocate for future improvements, including item ratings integration and advanced natural language processing techniques, to refine the system's accuracy in recommendations based on user interactions.
3js : cspc online dataset repository with an integrated recommender system using content based filtering algorithm / Joshua P. Estallo, Janben Aldrich H. Quiapo, and Jan Dale M. Parañal - 101 leaves.
Undergraduate Thesis (BSCS) - Computer Science.
Includes bibliographic references.
The significance of data collection in research is crucial, as it provides essential in-formation for hypothesis testing and drawing meaningful conclusions. Novice researchers often face challenges in accessing relevant datasets, impacting their work significantly due to issues like accessibility, completeness, and biases within datasets, leading to errors. This study aims to enhance user experience by developing a prototype dataset repository using content-based filtering algorithm on a centralized web platform. Utilizing a Python pro-gram with TF-IDF and Cosine Similarity, the system achieved a high user rating of 0.9481, effectively handling file uploads and resolving dataset query indexing. With an average pre-cision and recall score of 0.60, the system accurately predicts and delivers relevant datasets, emphasizing the precision in analyzing metadata to meet user needs. Researchers advocate for future improvements, including item ratings integration and advanced natural language processing techniques, to refine the system's accuracy in recommendations based on user interactions.