Browse

Bayesian Personalized Ranking with Count Data
횟수 자료를 이용한 베이지안 개인화 순위

DC Field Value Language
dc.contributor.advisor박병욱-
dc.contributor.author김동우-
dc.date.accessioned2017-10-31T08:33:41Z-
dc.date.available2017-10-31T08:33:41Z-
dc.date.issued2017-08-
dc.identifier.other000000145015-
dc.identifier.urihttps://hdl.handle.net/10371/138091-
dc.description학위논문 (석사)-- 서울대학교 대학원 자연과학대학 통계학과, 2017. 8. 박병욱.-
dc.description.abstractBayesian personalized ranking (BPR) is one of the state-of-the-art models for implicit feedback. Unfortunately, BPR has an limitation that it considers only the binary form of implicit feedback. In this paper, in order to overcome the limitation, we suggest an adapted version of BPR regarding the numeric value of implicit feedback like count data. Furthermore, we implement our model and original BPR in R and compare the results. This model may be useful to reflect implicit feedback more intensively than BPR.-
dc.description.tableofcontents1.Introduction 1
2.Preliminaries 3
2.1 Implicit Feedback 3
2.2 Matrix Factorization (MF) 4
2.3 Matrix Factorization for Implicit Feedback 8
2.4 Bayesian Personalized Ranking (BPR) 9
3 Bayesian Personalized Ranking with Count Data 14
3.1 Notation 14
3.2 Model 15
3.3 Meaning of cui 16
3.4 SGD Algorithm 18
3.5 Implementation 19
4 Discussion 24
Reference 25
Abstract in Korean 27
-
dc.formatapplication/pdf-
dc.format.extent2058504 bytes-
dc.format.mediumapplication/pdf-
dc.language.isoen-
dc.publisher서울대학교 대학원-
dc.subjectRecommendation system-
dc.subjectImplicit feedback-
dc.subjectCount data-
dc.subjectMatrix factorization-
dc.subjectBayesian personalized ranking (BPR)-
dc.subject.ddc519.5-
dc.titleBayesian Personalized Ranking with Count Data-
dc.title.alternative횟수 자료를 이용한 베이지안 개인화 순위-
dc.typeThesis-
dc.contributor.AlternativeAuthorDongwoo Kim-
dc.description.degreeMaster-
dc.contributor.affiliation자연과학대학 통계학과-
dc.date.awarded2017-08-
Appears in Collections:
College of Natural Sciences (자연과학대학)Dept. of Statistics (통계학과)Theses (Master's Degree_통계학과)
Files in This Item:
  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse