本研究中利用人臉辨識技術輔助，進行收視者身份辨識，調閱其收視資料，供所提出的群體推薦演算法進行節目推薦。此外，所提出的群體推薦演算法，將群體成員間的關係納入考量，推薦更合理的節目選單給使用群體。其主要概念在於利用計算群體與各成員的相關度，以推測出群體中各成員所佔的權重，進而估算出群體尚未收視的節目評比，並選出評比分數最高的N個節目，產生節目推薦清單給該群體。值得注意的是，本系統將利用推薦演算法持續更新成員權重，以求得更精確的群體成員關係，在約10次的節目推薦後，即可達成收斂。

In this study, the face recognition technique is applied to recognize the user’s identity. Thewatching-history of the recognized user is provided to the proposed group recommendation algorithm forrecommending programs. Besides, the proposed group recommendation algorithm takes the members’interaction into account for providing more reasonable program list to the group. The concept is based oncomputation of correlation between the group and its members for predicting the members’ weights. Finally,the unwatched programs’ ratings can be predicted, and the recommended Top-N programs will be provided tothe group. The group members’ weights are continuously updated by the recommendation algorithm to obtainmore accurate relationships of the group members. The proposed method can reach convergence status afterapproximately 10 recommendations.