行動裝置上之動態廣告推薦訊息

Abstract

行動裝置在這幾年來廣為人們使用和傳播。人們透過在行動裝置內各種應用程序，可以容易地在任何地方與人聯繫或做任何事情。因此，行動環境下的智慧管理已成為一個重要的研究課題。由於應用程序在提供其主要服務的同時也可以作為廣告的載體，適當的行動廣告推薦策略有利於用戶，應用程序提供商（即，廣告運營商），和廣告提供商。所以，在這篇研究中，我們提出了一個新的框架程序DRAM以在行動裝置上的應用程序推薦廣告。我們提出的框架會萃取如使用者的基本資訊、使用的應用程序、廣告點擊的歷史紀錄等各方面的性質和特徵以進行分析。 我們提出的框架程序會根據所萃取的各性質特徵將使用者分群，並計算各群使用者適合推薦的行動廣告。以行動紀錄檔的真實數據做實驗的結果顯示，與原方法相比， DRAM更可以為使用者提供其有興趣的廣告，並提昇使用者的平均廣告點擊率。

Mobile devices have been dramatically spread in recent years. People could do anything anywhere and connect with others easily via various apps on their devices in hand. Therefore, knowledge management in mobile environments becomes an important research topic. Since apps can act as the carriers of advertisements while providing the main functions, proper mobile advertising strategy should be developed to benefit users, app providers (i.e., advertisement carriers), and advertisement providers. Thus, in this work, we propose a novel framework DRAM to recommend advertisements for apps on mobile devices. The proposed framework extracts features in different aspects, such as client profiles, carrier apps, and clicked advertisement history. With the extracted features, the framework partitions the clients into groups and decides the suitable advertisements for each group of clients by computing the target advertisements. The experimental results on real-world mobile log data demonstrate that DRAM can provide interesting advertisements to each user such that the average CTR (click-through rate) is improved comparing with the baselines.