対話型遺伝的アルゴリズムにおける評価方法と個体生成方法の検討 Discussion of the Evaluation Method and the Generation Method of the Individuals for interactive Genetic Algorithm

この論文にアクセスする

この論文をさがす

著者

抄録

本研究ではショッピングサイトなどにおいて,商品選択時のユーザの嗜好を学習し,ユーザの嗜好に合った商品提示を行う仕組みとして対話型遺伝的アルゴリズム (iGA) を用いる.iGA を用いることにより,商品選択時のユーザの嗜好を反映した商品提示が可能になると考えられる.しかし,iGA を用いる際はユーザ負担の軽減と嗜好の多峰性について検討する必要がある.本論文では,前者に対し,提示個体からユーザの嗜好に合った個体のみを選択する評価方法,後者に対し,クラスタリングを用いた個体生成方法を提案し,その有効性を検証する Tシャツを対象として,iGA を用い,これら2つの提案手法を組込んだ商品選択支援システムを構築した.構築したシステムにおいて,評価方法,個体生成方法,それぞれの評価実験を行い,提案する評価方法がユーザ負担の軽減に有効であること,提案する個体生成方法が嗜好の多峰性に対応した個体を生成し,提示することができていることを確認した.We utilized an interactive Genetic Algorithm (iGA) as the method to reflect the user's taste in the product recommendation system. By applying iGA to the recommendation system, products that suit the user's taste can be presented. When applying iGA to the system, the following two points should be considered; a mechanism to reduce a user's fatigue and a mechanism to treat a user's multi-modal preferences. First, we proposed an evaluation method where the user only selects individuals that suit their taste from the presented individuals, so that the fatigue of the user can be reduced. Secondly, we also considered in the proposed system a method to generate and present individuals that correspond to the user's multi-modal preferences by applying a clustering technique. In this paper we developed a T-shirts selection support system based on iGA with the proposed mechanisms. Experiments were conducted to verify the effectiveness of the proposed method. From these experiments, we found out that the proposed method was effective in reducing fatigue, and that it was capable of presenting individuals that reflected the multi-modal taste of the user.

We utilized an interactive Genetic Algorithm (iGA) as the method to reflect the user's taste in the product recommendation system. By applying iGA to the recommendation system, products that suit the user's taste can be presented. When applying iGA to the system, the following two points should be considered; a mechanism to reduce a user's fatigue and a mechanism to treat a user's multi-modal preferences. First, we proposed an evaluation method where the user only selects individuals that suit their taste from the presented individuals, so that the fatigue of the user can be reduced. Secondly, we also considered in the proposed system a method to generate and present individuals that correspond to the user's multi-modal preferences by applying a clustering technique. In this paper we developed a T-shirts selection support system based on iGA with the proposed mechanisms. Experiments were conducted to verify the effectiveness of the proposed method. From these experiments, we found out that the proposed method was effective in reducing fatigue, and that it was capable of presenting individuals that reflected the multi-modal taste of the user.

収録刊行物

  • 情報処理学会研究報告数理モデル化と問題解決(MPS)

    情報処理学会研究報告数理モデル化と問題解決(MPS) 2008(17(2008-MPS-068)), 113-116, 2008-03-05

    一般社団法人情報処理学会

参考文献:  6件中 1-6件 を表示

各種コード

  • NII論文ID(NAID)
    110006827840
  • NII書誌ID(NCID)
    AN10505667
  • 本文言語コード
    JPN
  • 資料種別
    Technical Report
  • ISSN
    09196072
  • NDL 記事登録ID
    9458072
  • NDL 雑誌分類
    ZM13(科学技術--科学技術一般--データ処理・計算機)
  • NDL 請求記号
    Z14-1121
  • データ提供元
    CJP書誌  NDL  NII-ELS  IPSJ 
ページトップへ