在行動社群網路中考慮地點屬性偏好的傳播機率估計方法

Abstract

由於智慧型手機很容易取得目前位置資訊，使用者可以輕易地在行動社群網路上透過打卡和朋友分享目前的位置和活動。而在社群網路中，病毒式行銷利用使用者之間互相傳播廣告資訊，快速地散佈資訊。若要利用病毒式行銷在行動社群網路上行銷一個地點，則必須先評估使用者之間對於不同地點的打卡傳遞機率。在本篇論文中，我們專注於在行動社群網路中計算使用者之間的打卡傳遞機率。之前的研究中只考慮行銷地點和使用者過去打卡地點的遠近關係。然而，由於每個使用者有不同的地點屬性偏好，使用者可能會因喜愛的地點屬性而願意移動較遠的距離，反之亦然。因此，我們預估傳播機率時，不僅考慮在不同類別的打卡行為，也考慮個人對地點屬性偏好。基於兩個真實資料集的實驗結果，顯示我們所提出的方法能夠有效反映行動社群網路中打卡資訊的傳遞。

With the explosion of smartphones, users are easy to share their current location and activities with their friends by the check-in function in location-based social networks. Moreover, the success of viral marketing in social networks since users are more likely to accept the information from their friends. To promote a target location attracting as more as possible users to visit in location-based social networks via viral marketing, we have to estimate the propagation probability from users’ check-in records. In this thesis, we focus on estimate propagation probability of each social connection. Prior works only consider the distance between users’ visited locations and the target location. However, different users have different category preferences. For instance, most users would like to move far away if the category of target location is preferred. Therefore, we consider not only the individual check-in behavior in different categories but also the individual category preferences. To derive the propagation probability in LBSNs, experiments are conducted on two real datasets, and the results show that our proposed approach can truly reflect the information propagation in LBSNs.