2008The KIPS Transactions PartDOpen access

A Contents Recommendation Scheme Based on Collaborative Filtering Using Consumer's Affection and Consumption Type

In-Bok Choi, Tae-Keun Park, Jae-Dong Lee

Open full text 0 citations

Abstract

협업여과 기법은 추천 시스템에서 널리 사용되는 기술이지만, 소비자의 참조그룹을 선정하는 방법에 따라 추천의 정확도가 달라지는 특성을 가지고 있다. 이에 본 논문에서는 콘텐츠 추천의 정확도를 높이기 위하여 소비자의 감성과 소비유형을 참조그룹으로 하여 협업여과기반으로 콘텐츠를 추천하는 기법을 제안한다. 소비자의 감성을 기쁨, 슬픔, 혐오, 행복, 이완 다섯 가지로 구분하고, 소비유형을 저실용/저쾌락, 저실용/고쾌락, 고실용/저쾌락, 고실용/고쾌락 네 가지로 구분하여 콘텐츠 추천 기법의 성능을 분석한 결과, 본 논문에서 제안하는 기법으로 콘텐츠를 추천한 경우가 소비자 감성과 소비유형을 고려하지 않은 전체 참조그룹으로 추천한 경우보다 정확도가 향상됨을 확인하였다. Collaborative filtering is a popular technique used for the recommendation system, but its performance, especially the accuracy of recommendation, depends on how to define the reference group. This paper proposes a new contents recommendation scheme based on collaborative filtering technique whose reference groups are created by consumer's affection and consumption type in order to improve the accuracy of recommendation. In this paper, joy, sadness, anger, happiness, and relax are considered as the consumer's affection. And, low-utility / low-pleasure, low-utility / high-pleasure, high-utility / low-pleasure, and high-utility / high-pleasure are considered as the consumer's shopping types. Experimental results show that the proposed scheme improves the accuracy of recommendation compared to the recommendation scheme considering neither consumer's affection nor consumption type.

Open-access reader

About this research paper

What this paper is about

협업여과 기법은 추천 시스템에서 널리 사용되는 기술이지만, 소비자의 참조그룹을 선정하는 방법에 따라 추천의 정확도가 달라지는 특성을 가지고 있다. 이에 본 논문에서는 콘텐츠 추천의 정확도를 높이기 위하여 소비자의 감성과 소비유형을 참조그룹으로 하여 협업여과기반으로 콘텐츠를 추천하는 기법을 제안한다. 소비자의 감성을 기쁨, 슬픔, 혐오, 행복, 이완 다섯 가지로 구분하고, 소비유형을 저실용/저쾌락, 저실용/고쾌락, 고실용/저쾌락, 고실용/고쾌락 네 가지로 구분하여 콘텐츠 추천 기법의 성능을 분석한 결과, 본 논문에서 제안하는 기법으로 콘텐츠를 추천한 경우가 소비자 감성과 소비유형을 고려하지 않은 전체 참조그룹으로 추천한 경우보다 정확도가 향상됨을 확인하였다. Collaborative filtering is a popular technique used for the recommendation system, but its performance, especially the accuracy of recommendation, depends on how to define the reference group. This paper proposes a new contents recommendation scheme based on collaborative filtering technique whose reference groups are created by consumer's affection and consumption type in order to improve the accuracy of recommendation. In this paper, joy, sadness, anger, happiness, and relax are considered as the consumer's affection. And, low-utility / low-pleasure, low-utility / high-pleasure, high-utility / low-pleasure, and high-utility / high-pleasure are considered as the consumer's shopping types. Experimental results show that the proposed scheme improves the accuracy of recommendation compared to the recommendation scheme considering neither consumer's affection nor consumption type.

Why it matters

A significance statement is not available in the OpenAlex record.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

협업여과 기법은 추천 시스템에서 널리 사용되는 기술이지만, 소비자의 참조그룹을 선정하는 방법에 따라 추천의 정확도가 달라지는 특성을 가지고 있다. 이에 본 논문에서는 콘텐츠 추천의 정확도를 높이기 위하여 소비자의 감성과 소비유형을 참조그룹으로 하여 협업여과기반으로 콘텐츠를 추천하는 기법을 제안한다. 소비자의 감성을 기쁨, 슬픔, 혐오, 행복, 이완 다섯 가지로 구분하고, 소비유형을 저실용/저쾌락, 저실용/고쾌락, 고실용/저쾌락, 고실용/고쾌락 네 가지로 구분하여 콘텐츠 추천 기법의 성능을 분석한 결과, 본 논문에서 제안하는 기법으로 콘텐츠를 추천한 경우가 소비자 감성과 소비유형을 고려하지 않은 전체 참조그룹으로 추천한 경우보다 정확도가 향상됨을 확인하였다. Collaborative filtering is a popular technique used for the recommendation system, but its performance, especially the accuracy of recommendation, depends on how to define the reference group. This paper proposes a new contents recommendation scheme based on collaborative filtering technique whose reference groups are created by consumer's affection and consumption type in order to improve the accuracy of recommendation. In this paper, joy, sadness, anger, happiness, and relax are considered as the consumer's affection. And, low-utility / low-pleasure, low-utility / high-pleasure, high-utility / low-pleasure, and high-utility / high-pleasure are considered as the consumer's shopping types. Experimental results show that the proposed scheme improves the accuracy of recommendation compared to the recommendation scheme considering neither consumer's affection nor consumption type.

Key concepts: Affection, Pleasure, Sadness, Happiness, Collaborative filtering, Consumption (sociology), Computer science, Recommender system

Related papers

Back to paper searchBrowse research topicsOriginal source
A Contents Recommendation Scheme Based on Collaborative Filtering Using Consumer's Affection and Consumption Type — Research Paper | ScholarLens