隨著媒體多元化、資訊技術的進步與網路應用普及化,消費者與客服中心接觸之管道亦已呈現多元化的趨勢。越來越多的使用者利用網路向企業提出問題。但使用者所輸入的問題,多半是自然語言問句,而且都是屬於非結構化的文字資料,而系統對於這一類的訊息處理能力卻有所限制。本研究設計一個語意感知機制,應用於處理客服問題詢答的問題,此機制根據文件的語意,研判相關的問題概念屬性,開發以語意規則(Semantic Rule)為基礎的語意解釋系統,運用知識本體(Ontology)的明確性表達為問題領域知識(Domain Knowledge)的概念塑模,分析組織概念、職務概念與答案概念之間的屬性關係,建立問題本體知識庫,再利用語意規則定義出職務、答案與問題之間的關係,最後結合推論引擎完成問題分派及篩選出可能答案的工作。經實驗結果顯示:其問題語意判斷正確率為82.47%,經由系統的判斷可以縮短人工研判的時間,讓相關處理部門可以在第一時間收到新的問題,快速的解決客戶所面臨的問題;並篩選可能的答案,提供客服人員參考,以縮短回應問題的時間。
As media diversify, information technologies develop and internet usage becoming more popular, consumer and customer service contact methods also show a trend towards diversification.More consumers are using the internet to mention customer service concerns. However, the problem with consumer input through internet is that most questions are presented in an unstructured manner, which can not be processed by some systems. Thus, this research invented a semantic mechanism, which can assist machines in processing questions proposed by customers.The semantic mechanism defines a set of concepts, taxonomy, and interrelated rules that dictate these concepts in a way that can be understood by machines. Using the clear expression of ontology as the modeling concept for domain knowledge, this research developed a semantic interpretation system based on semantic rules.This research then analyzed and defined the relationships among concepts of organizations, functions, and answers, then established a large domain of ontology. Next, utilizing the defined relationship among functions, answers and problems, this research finally completed the job of assigning and screening possible answers after forming a rule-based inference engine. Experimental results indicate that the semantic mechanism was accurate 82.47% of the time, and the system shortened the time for manual analysis. Thus, customer service received new problems more efficiently, and solved customer concerns more effectively. The system also screened and then selected possible answers as references for customer service staff to shorten problem response time.