近幾年來，新型的傳播媒體形式層出不窮，電腦互相通訊的形式已經從文本和圖像擴展到音頻和影像，因此移動直播平台應運而生。移動直播平台創造出一群素人網紅，也讓移動直播平台變現方式更為多元化。網紅們可以透過觀眾流量等與平台分潤，接受觀眾贈送的虛擬禮物後與平台分帳，甚至平台會直接收取觀眾或者投資方的贊助等方式來取得收益。在眾多社交平台中都有好友推薦功能，它的實現提出了一種基於三度影響力理論的好友推薦演算法。它顯著提升了社交平台好友推薦的效果，從而進一步促進了用戶活躍。本研究欲在移動直播平台中實作基於三度影響力的三度關係推薦系統，按照推薦重合度並擇優展示，可以幫助用戶發現更多他喜歡的主播，提高網紅主播曝光量，並促進用戶活躍，並能維持平台的用戶人數，獲得更多收益。透過本研究完成的系統回饋問卷得知，本研究開發的推薦系統使平台可以達到顯著的用戶觸及量超過9人，而推薦的平均準確率為79.42%，因此三度關係推薦系統可以使移動直播平台在一定程度上促進其用戶使用平台的參與度及活躍度。

In recent years, new forms of communication media have emerged in an endless stream. The forms of computer communication have expanded from text and images to audio and video, and online live broadcast platforms have emerged as the times require. The live broadcast platform has created a group of amateur Internet celebrities, which also makes the live broadcast platform realize more diversified ways. Internet celebrities can share profits with the platform through audience traffic, share accounts with the platform after accepting virtual gifts from the audience, and even the platform will directly collect sponsorship from audiences or investors to obtain income. There is a friend recommendation function in many social platforms, and its implementation proposes a friend recommendation algorithm based on the three-degree influence theory. It significantly improves the effect of friend recommendation on social platforms, thereby further promoting user activity. This study intends to implement a three-degree relationship recommendation system based on three-degree influence in the live broadcast platform. According to the recommendation coincidence degree and displaying the best, it can help users find more anchors he likes, increase the exposure of Internet celebrity anchors, and promote users. Be active, increase the number of users on the platform and maintain the number of users on the platform, and get more benefits. According to the system feedback questionnaire completed in this study, the recommendation system increases the average reach of platform users to 9 people, and the average recommendation accuracy rate is 79.42%. Therefore, the three-degree relationship recommendation system can make the live broadcast platform promote its The degree of participation and activity of users using the platform.