本研究旨在探討如何以目標導向方法建構一個智慧型人機互動界面，並將 其應用於行動式裝置上。本研究之重點有二: 建立目標導向模型及其分析方 法以分析及表示使用者之明確及不明確意圖；以案例式推理(Case-Based Reasoning, CBR)及資料導向剖析(Data-Oriented Parsing, DOP)方法為基礎，開 發具學習能力之智慧型人機互動界面。我們將使用者之輸入分為明確輸入 及不明確輸入。明確輸入可以直接對應到相關之系統服務以滿足使用者要 求；不明確輸入則須利用建立之背景知識進行推理以轉成明確目標，以對 應到相關之服務。背景知識來自於系統分析及設計階段，利用目標導向軟 體需求工程之方法分析結果。我們將可能之使用者輸入，建立成相對應之 目標模型，以作為使用者意圖判別之基礎。利用此背景知識，系統可以推 理出在使用者輸入字面之外的隱含意圖，以利系統主動提供更方便的服 務。當具有此界面之行動裝置接收到使用者之語音輸入時，界面會將語音 輸入轉成相對之字串，再利用案例式推理方法，依據過往經驗及事先建立 之背景知識，將輸入轉成相對應之目標模型，最後產生一系列行動目標作 為滿足使用者之服務指引。我們將以數位家庭為應用環境，設計在行動裝 置上之智慧型數位家電控制界面。 The purposes of this research are to discuss how to construct an intelligent human-computer interface and use it in mobile devices. There are two research issues in this project: constructing a goal-driven model and its analysis method for explicit and implicit user intention representation and analysis, and developing an intelligent human-computer interface with learning capability based on Case-Based Reasoning (CBR) and Data-Oriented Parsing (DOP). The user’s input will be divided into explicit and implicit input. Explicit input can be mapped to related system services to satisfy user requests. Implicit input will be transfer to explicit input by reasoning with background knowledge and map to related services. The background knowledge comes from results of goal-driven software requirement engineering. The possible user’s input will be analyzed and used to construct goal models. Using these goal models, system can infer the implicit intention in user’s input and provide more convenient services actively. When a mobile device with the intelligent interface receives user’s vocal input, the interface will transfer the vocal input into a string. Based on previous parsing experience and back ground, the CBR approach will generate related goal models. Finally, based on these goal models, the system can provide a series of actions to satisfy user’s requests. Based on our research results, we will design an intelligent mobile interface for digital homes.
