結合頻率和相似度之個人本體論

Abstract

近年個人化已快速成為一個在任何軟體系統上不可或缺的使用者體驗,尤其是需要和用戶直接互動的介面,更需要提供使用者獨一無二的體驗。因為每個人感知世界的方式各不相同,因此規格統一的軟體解決方案將逐漸式微。透過個人化,人們能有更高的機會取得有用的結果,無論結果落入或超出使用者期待都能有效提升使用者體驗。 本論文旨在塑模和建構一個完整的個人本體論系統,同時該系統也能與時俱進。本研究將該系統應用於行動裝置輸入法上,作為鍵盤的預測引擎。建構在個人本體論之鍵盤能從使用者行動裝置中的訊息提取出現頻率高、相似度高的詞彙,並向使用者推薦最合適的選項。鍵盤不僅會根據個人寫作風格來推薦下一個詞彙,也會持續學習用戶的使用習慣,提供更為實用的預測。

Nowadays, personalization is fast becoming a must have in any system that directly interacts with a user. The reason is because different people have different views towards the world that they perceive, as a result one size fit all software solutions are on the edge of becoming obsolete. By personalization, we can achieve a higher chance that people obtain the results that they anticipate either consciously or unconsciously, which in turn provide the users with an overall improved experience. The goal of this thesis is to create a system that generates users’ Personal Ontologies based on their personal writings which are extracted from their mobile smartphones. These ontologies are used for refining users’ queries by keywords that the users are not only interested in, but also important words in their day to day vocabulary. As a result the Personal Ontology has been designed to function as a prediction engine suggesting the optimal next words whilst typing in a software keyboard. Not only is this keyboard already based on a person’s writing style, but it will continue to become more intelligent, in that it provides more accurate predictions as time goes by.