電話行為視覺化

Abstract

行動電話近年來對人與人之間溝通的重要性越來越高，但是，除了方便和有效率之外，行動電話也帶來了許多新的問題，譬如：侵犯隱私權、打擾或是使人分心。在這些問題中，最讓人感到厭煩的就是騷擾電話，例如：詐騙電話、電話行銷等。因此，此研究我們設計出一套視覺化方法來分析通話行為，我們使用的資料長度有一個月，一天約有兩三百萬名用戶活動。我們的系統利用階層式分群法提供數個不同的面相，並且設計了光譜視覺化來顯示詳細的通話行為。我們的實驗證明我們的系統能夠有效的找出異常電話並且能夠分析各種不同通話行為的電話使用者。

Mobile phones have become one of the most important personal communication tools in our daily life. Despite its convenience and effectiveness, mobile phones also bring new problems such as invasion of privacy, interruption or distraction to normal activities. Among these problems, annoying calls such as telemartketing or fraud calls, is perhaps the most common one. In this paper, we present a visualization system to analyze the calling behaviors in a database consisting of more than 1 million mobile phone calls in Taiwan. Our system provides multiple views consisting of hierarchical clustering of users based on their calling behaviors, calling spectrum and time plot of individual users. Our experiment demonstrated the effectiveness of our system on discovering the users making annoying calls as well as visually analyzing the calling behaviors of different types of users.