著者

清田 雅子 山田 泰宏 服部 哲 速水 治夫

雑誌

研究報告グループウェアとネットワークサービス(GN)

巻号頁・発行日

vol.2013, no.3, pp.1-6, 2013-03-11

---

テレビドラマは四半期毎に新たな作品が多く放送されているため,視聴者は過去に見た作品すべてを覚えていることは困難である.また,どの作品が自分の嗜好に合うか判断するための数量的な基準が存在しない.そのため,過去に好んで見ていた作品を忘れてしまい再視聴する機会の損失や,誤った判断により自分の嗜好に合わない作品を視聴する可能性がある.本論文ではユーザが過去に見た作品を管理し,再視聴の機会を増やすとともに,ユーザの嗜好に合った未視聴作品を自動的に推薦することで新たな作品を見る機会を作ることを目的とする.そのため,本論文ではユーザが過去に見た作品を 5 段階評価で登録し,その情報を基に未視聴作品の評価を推測し,オススメ度を具体的な数値で示す機能を有した推薦システムを提案する.実験により,提案システムは過去に見た作品の管理に有効であり,ユーザの嗜好に合ったオススメ作品を提示できることが確認できた.For the Japanese television drama, new works are released and broadcast on a quarterly basis. Thus, it is difficult for the viewer of the work to remember most works that the viewer had seen. Moreover, there are not quantitative criterions which would indicate whether or not the work is suitable to the viewer. For those reasons, it is common that to lose the opportunity to view the work again, because the viewer forgot the works of which the viewer had been looking in favor, and there is a possibility to view the work that does not suitable to their taste occurred by the user's dis-appropriate decision. Our study is aiming at managing the work that was seen in the past by the user, making the chance to see new works by recommending automatically unviewed works that suit the every user's preference. Therefore we propose the recommender system. The user of the proposed system can register works that the user has seen, and score the work in the 5-point score scale. The proposed system could calculate the score that indicates how the work is suitable to the user. The score is specific numerical values: based on the evaluation of the score of work. From the experiments, the proposed system is effective in the management of work seen in the past, We were able to confirm that the work that is suitable to the user's preference presented by our recommendation function.