網路直播績效指標之研究

網路直播蓬勃發展，其中直播電商更是目前電子商務中最新興、也相當受歡迎的一種形式。然而網路直播發展至今，「在線人數」已不能作為評量直播主的唯一績效指標。因此，本研究提出以文字探勘技術、針對直播中聊天室的觀眾及其留言進行分析，作為衡量網路直播績效的指標內涵。本研究根據前述技術與觀點，設計出四個系統模組，其中熱絡度分析模組、情感度分析模組、相關度分析模組，分別可以計算出網路直播的熱絡度、情感度、相關度；時間分析模組則可以分析出在直播的時間歷程下，熱絡度、情感度、相關度的消長變化。此外，為了驗證本研究所提出之網路直播績效指標及系統的可行性，本研究針對兩家Facebook直播電商進行系統實作與分析，經網路爬蟲程式收集其線上資料，並進行四個模組的分析，再以儀表板模式進行指標展現。本研究提出之績效指標確實有助於直播主及直播平台更有效地掌握、瞭解直播時之熱絡度、情感度、相關度，可作為精進、調整直播策略及內容之重要參考。

關鍵字

網路直播；直播電商；文字探勘；情感分析；績效指標

並列摘要

The webcast is booming, and the live e-commerce is a new and popular form of e-commerce. However, since the development of webcasting, "online number" has not been used as the only performance indicator for evaluating live broadcasters. Therefore, this study proposes the use of Text Mining techniques, analysis of viewers and their messages in chat rooms in live broadcasts, as an indicator of the performance of online live broadcast. Based on the above techniques and viewpoints. This study designed four system modules, in which the thermal degree analysis module, the sentiment analysis module and the correlation analysis module can calculate the thermal degree and emotionality of the live webcast. Correlation degree; time analysis module can analyze the change of thermal degree, affectivity and correlation under the time course of live broadcast. In addition, in order to verify the performance of the webcast performance indicators and the feasibility of the system, this study conducted systematic implementation and analysis of two Facebook live e-commerce providers, collected online information through web crawlers, and conducted four analysis of the module, and then the indicators in the dashboard mode. The performance indicators proposed in this study really help the live broadcast and live broadcast platforms to more effectively grasp and understand the hotness, affection and relevance of the live broadcast, which can be used as an important reference for improving the live broadcast strategy and content.