A design of recommendation based on flexible mixture model considering purchasing interest and post-purchase satisfaction

Takeshi Suzuki, Gendo Kumoi, Kenta Mikawa, Masayuki Goto

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

The recommender system is an effective Web marketing tool that havve been used especially on electric commerce sites in recent years. The recommender system provides each user with a list of new recommended items that are predicted to be preferred by the user. Collaborative filtering is one of the most representative and powerful methods to predict user preference in the recommender system. Collaborative filtering measures the similarity of preference between users and uses it to decide items to be recommended. Based on previous researche on this method, user preference is considered to have two aspects: Purchasing interest for items and post-purchase satisfaction with items. However, the conventional methods do not consider the two different preferences at the same time. This paper suggests taking these two preferences into account and proposes a new method that allows users to choose the balance between them. The proposed method is evaluated through simulation experiments with MovieLens data. It demonstrates the effectiveness of our proposal in precision and average rating compared with a previous method.

Original languageEnglish
Pages (from-to)570-578
Number of pages9
JournalJournal of Japan Industrial Management Association
Volume64
Issue number4
Publication statusPublished - 2014

Fingerprint

Purchasing
Recommender systems
Mixture Model
Recommendations
Collaborative filtering
Recommender Systems
Collaborative Filtering
User Preferences
Marketing
Simulation Experiment
Design
Mixture model
Purchase
Choose
Experiments
Predict
Demonstrate
User preferences

Keywords

  • Collaborative filtering
  • Flexible mixture model
  • Latent class model
  • Probabilistic models
  • Recommender systems

ASJC Scopus subject areas

  • Industrial and Manufacturing Engineering
  • Applied Mathematics
  • Management Science and Operations Research
  • Strategy and Management

Cite this

@article{fb565ec12c1242138f6762b90fd2b22a,
title = "A design of recommendation based on flexible mixture model considering purchasing interest and post-purchase satisfaction",
abstract = "The recommender system is an effective Web marketing tool that havve been used especially on electric commerce sites in recent years. The recommender system provides each user with a list of new recommended items that are predicted to be preferred by the user. Collaborative filtering is one of the most representative and powerful methods to predict user preference in the recommender system. Collaborative filtering measures the similarity of preference between users and uses it to decide items to be recommended. Based on previous researche on this method, user preference is considered to have two aspects: Purchasing interest for items and post-purchase satisfaction with items. However, the conventional methods do not consider the two different preferences at the same time. This paper suggests taking these two preferences into account and proposes a new method that allows users to choose the balance between them. The proposed method is evaluated through simulation experiments with MovieLens data. It demonstrates the effectiveness of our proposal in precision and average rating compared with a previous method.",
keywords = "Collaborative filtering, Flexible mixture model, Latent class model, Probabilistic models, Recommender systems",
author = "Takeshi Suzuki and Gendo Kumoi and Kenta Mikawa and Masayuki Goto",
year = "2014",
language = "English",
volume = "64",
pages = "570--578",
journal = "Journal of Japan Industrial Management Association",
issn = "0386-4812",
publisher = "Nihon Keikei Kogakkai",
number = "4",

}

TY - JOUR

T1 - A design of recommendation based on flexible mixture model considering purchasing interest and post-purchase satisfaction

AU - Suzuki, Takeshi

AU - Kumoi, Gendo

AU - Mikawa, Kenta

AU - Goto, Masayuki

PY - 2014

Y1 - 2014

N2 - The recommender system is an effective Web marketing tool that havve been used especially on electric commerce sites in recent years. The recommender system provides each user with a list of new recommended items that are predicted to be preferred by the user. Collaborative filtering is one of the most representative and powerful methods to predict user preference in the recommender system. Collaborative filtering measures the similarity of preference between users and uses it to decide items to be recommended. Based on previous researche on this method, user preference is considered to have two aspects: Purchasing interest for items and post-purchase satisfaction with items. However, the conventional methods do not consider the two different preferences at the same time. This paper suggests taking these two preferences into account and proposes a new method that allows users to choose the balance between them. The proposed method is evaluated through simulation experiments with MovieLens data. It demonstrates the effectiveness of our proposal in precision and average rating compared with a previous method.

AB - The recommender system is an effective Web marketing tool that havve been used especially on electric commerce sites in recent years. The recommender system provides each user with a list of new recommended items that are predicted to be preferred by the user. Collaborative filtering is one of the most representative and powerful methods to predict user preference in the recommender system. Collaborative filtering measures the similarity of preference between users and uses it to decide items to be recommended. Based on previous researche on this method, user preference is considered to have two aspects: Purchasing interest for items and post-purchase satisfaction with items. However, the conventional methods do not consider the two different preferences at the same time. This paper suggests taking these two preferences into account and proposes a new method that allows users to choose the balance between them. The proposed method is evaluated through simulation experiments with MovieLens data. It demonstrates the effectiveness of our proposal in precision and average rating compared with a previous method.

KW - Collaborative filtering

KW - Flexible mixture model

KW - Latent class model

KW - Probabilistic models

KW - Recommender systems

UR - http://www.scopus.com/inward/record.url?scp=84923256957&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84923256957&partnerID=8YFLogxK

M3 - Article

VL - 64

SP - 570

EP - 578

JO - Journal of Japan Industrial Management Association

JF - Journal of Japan Industrial Management Association

SN - 0386-4812

IS - 4

ER -