80 likes | 197 Vues
RecULike is an innovative open-source academic recommender system designed to enhance research discovery. It recommends research papers based on user behavior, offers related papers for any given document, and fosters connections between users and experts in the field. Built to serve as an illustrative example for an upcoming book, RecULike features a robust architecture that includes a data importer, search engine, and recommender system. This tool aims to advance academic collaboration and information accessibility, promoting a richer research ecosystem.
E N D
RecULike : Open Source Academic Recommender System Liang Xiang Researcher@Hulu
Main Contents • What is RecULike? • Why build RecULike? • How to build RecULike? • Architecture • Data Importer • Search Engine • Recommender System • Site/API
What is RecULike? • RecULike is a academic recommender system which can: • Recommend research papers to users by analyzing their historical behaviors • Provide related papers of given paper • Connect users and experts
Why build RecULike • We want to write a book • We want to use a real recommender system as our example in the book
How to Build RecULike • Architecture • Data Importer • Search Engine • Recommender System • Site/API
Architecture Metadata Page DBLP Meta Data Database Search Page Importer Sphinx CiteSeer CiteULike Cold Start Recommendation Page User Recommender System User Behavior Database Log System
Recommender System • Cold-Start • Content Filtering • Collaborative Filtering