近年來隨著Facebook的盛行，Facebook已成許多人紀錄生活或關注朋友近況不可或缺的管道。許多人文社會學者意圖透過其用戶在Facebook上活動行為的紀錄(如打卡、轉貼分享連結等)，進行學術研究分析；而Facebook公司相對也提供應用程式介面並以特定資料格式提供，但考量用戶之個人隱私及平台的可用度，制訂了相關的使用規則；因此本研究以協助學者達成研究分析Facebook用戶動態消息資料為目的，設計與實作一個「臉書使用者行為資料蒐集系統(Facebook User Activities Data Collector) 」。可相容Facebook應用程式介面使用規則及穩定收取其動態消息資料，讓學者們可針對其研究項目分別設定相關「蒐集專案」，並可使特定用戶(即蒐集專案之參與者)於本系統進行同意及授權提供特定資料，系統則每日定時依蒐集專案設定進行蒐集作業，待作業完成，再將資料以圖形化介面呈現。 因考量本系統可同時運行多個蒐集專案，為優化其效能，本研究在整合「蒐集專案」、「參與者」後提出一套有效的歸戶機制(Account Aggregation Mechanism)，透過重整參與者的機制，可顯著降低重複蒐集的次數，以提升資料蒐集的效率；並為因應Facebook所制訂使用規則，設計一套臉書蒐集相容機制(Facebook Collector Compatible Mechanisms)，在蒐集過程中對於可能產生錯誤情況加以防範，以及錯誤產生時能以對應之處理程序進行自動化處理，以確保資料蒐集過程的穩定及結果的完整。最後，在系統開發完成後，設計一套驗證計畫，以驗證本研究之蒐集成果以及相關機制成效。 / Recently Facebook has become an indispensable platform for many people to record their lives and to concern their friends. Many Humanities and Social Sciences scholars are interested in conducting research on the activity records of the Facebook users, such as check-ins and shared story. Therefore, there is a large demand in collecting these user activity records while respecting user privacy. Based on the open API (Application Programming Interface) by Facebook, we design and implement a “Facebook User Activities Data Collector” to help scholars collect the activity records with the permission from Facebook users. Using the collector, researchers can create projects for Facebook data retrieval and solicit volunteers to participate those projects by giving their permissions to enable the collector to retrieve their activity records following the Facebook privacy rules. Our tool offers a friendly user interface for researchers and participating users. Besides, as a user may participate in multiple data retrieval projects in the same period of time, our collector is equipped with an “account aggregation” mechanism that will retrieve all the records of a user for the projects he participated at once without repetitive collection work that will degrade the performance of the tool. Another feature of the collector is a robust exception handling mechanism that will handle frequent occurring exceptions in data retrieval and reports unknown exceptions to system administrators to improve the system’s reliability. To verify our collector, we have also recruited several volunteers and conducted some experiments to retrieve their activity records. The preliminary results show that our collector performs well and collects the records correctly.

近年來隨著智慧型手機的興起，人們開始利用智慧型手機處理日常生活事務及社群娛樂。無形中，已逐漸的改變日常行為模式。因此，許多學者也開始進行人們使用智慧型手機狀況的研究。 若透過傳統問卷等方式蒐集分析使用記錄，受訪者可能因訪問員的外在因素而影響問卷有效性。為解決此問題，很多研究開始在智慧型手機中安裝應用程式，蒐集使用記錄，藉以分析使用行為。然現今此類應用程式，都是為特定的研究進行開發，缺乏了彈性的考量。因此，本論文設計一個是用於安卓智慧型手機的可客製化行為感測記錄器，提供研究學者依各自需需求產生實驗項目規格檔，由受測者下載至手機，透過我們的行為感測記錄器，記錄平時手機操作行為之相關記錄，並於指定之時機下將記錄上傳供研究學者統計分析，以符合個別實驗之要求。 本研究重點在於如何定義彈性且多樣化的實驗設定檔，並考量載記錄感測過程中，電力大量消耗和受測者隱私資保護的議題，提出因應的解決方法。我們設計與執行幾個實驗來驗證此行為感測記錄器，初步看來，此記錄器可以滿足我們的主要需求，但結果也導引出一些需改進的設計議題。 / In recent years, with the rise of smart phones, the public began to take advantage of smart phones in daily life and social engagements. Virtually, smart phones have gradually changed people's behavior patterns so that researchers are eager to understand the how people use smartphones in their daily life. In this thesis, we present a configurable activity logger for Android smartphones for recording a smartphone’s sensor data and the user’s operations. Instead of one Android App for each researcher’s experiment, our approach is to devise a general logger App that can be customized to collect different data items according to the needs of various researchers. Specifically, our logger App accepts experiment configuration files in which researchers specify what data items to collect and at what frequency, and then conduct the logging task accordingly for each experiment. In other words, our logger works like an interpreter on such experiment configure files. Besides customizability of logging items, our logger concerns the saving of electricity and the preservation of user privacy. Firstly, we have applied various power saving tips published in the literature to the design of our logger. Secondly, we have refrained from collecting any contents that may reveal the user’s identify or that may invade user’s privacy. To evaluate our logger, we have designed and implemented a few data collection experiments. The preliminary results show that our logger can fulfill most of the requirements we set up, though thy also reveal some improvements we need to work on in the future.