Customizing and calibrating the data relationship via the Wheel of Emotions

[[abstract]]現今，由於網路發達資訊流通迅速，造成資訊爆炸性的成長，每天、甚至每秒都會產生許多不同的資料產生，使用者卻時常來不及閱讀詳細，亦出現了許多讓使用者能夠更快速的了解資料的重點的方法。然而，很多時候分析後的結果，並不能切合使用者真正的需求。因此，我們希望能透過人機介面的設計，以簡單且直觀的方式對這樣的問題提供線索。 在本研究中，我們首先利用傳統的web mining技術以及簡單的關聯規則分析找出關鍵字資料之間的關係，並以Robert Plutchik在2001年所提出的情緒輪(Wheel of Emotions)模型為基礎，讓使用者得以利用自己的感知對畫面中的關鍵字群組做情緒與顏色對應，進而獲得個人化的關聯性並且我們會記錄對應結果以便改進往後的關係呈現。此外，為了能夠更加準確的獲得使用者的情緒顏色感知，此系統允許使用者對原始情緒輪中的基本情緒與顏色做對應校正。 為了讓畫面呈現上能夠更好，我們的介面結合完形心理學中的趨近性(Proximity)以及相似性(Similarity)等視覺組織法則，依照情緒輪上的情緒相對關係，影響字群圓之距離，關係越強則越近，反之，則越遠。且根據實驗結果，我們的系統，不僅提升了資料的呈現外，也使得使用者更加直覺切互動式地感知資料。[[abstract]]Today, because of the Internet become highly developed, the amount of information has rapidly grown and is explosively growing on the web. Every day, even every second, new datum is created and accumulated. It is unpractical and near impossible for a user to spend so much time to read all data specifically. There are amount of methodologies developed for helping users find and understand the key point of datum faster and more efficiently. However, usually, most of the output data items are not really essential to users. Therefore, in this paper, we design a human-machine interface to provide simple and intuitive visual clues for helping solve the above problems. In this research, we use the conventional web mining technologies and the simple association rule method to extract and construct the relations between data, in the first. And then, based on Robert Plutchik’s Wheel of Emotions concept, users are enabled to set up a mapping between a keyword group and a color/emotion. Moreover, these color/emotion mappings are stored in a user profile database, and are utilized next time for customizing and calibrating the relationships among data for the same user. For better presentation of data, we design our interface according to the Law of Proximity and the Law of Similarity from Gestalt psychology. For example, for users to recognize the datum’s relationships more intuitively, keyword groups with stronger association are deployed closely; and, contrarily, keyword groups with weaker association are set farther. The experiment results show that our proposed approaches do not only improve the presentation of data, but also enhance user’s perception to data in an intuitive and interactive way.[[note]]碩