顧客滿意是服務之第一目標，如何有效地界定與預測服務需求、分析服務評價，以供企業服務決策之參考為企業首要課題。網路時代下，顧客需求係以非結構化的網路口碑來呈現顧客之消費經驗，卻因為人類語意的複雜性與多樣性造成傳遞過程中需求落差之問題。因此，本研究提出「語意為基礎之顧客體驗分析機制」，自動擷取文章，分析並識別口碑評價，以弭平顧客實際需求與企業認知需求間的落差。實驗結果亦證明，此機制能有效萃取出八成以上之口碑評價與七成以上之評價正確率，足見「語意為基礎之顧客體驗分析機制」之效益與在服務需求探索上之適用性。

Customer satisfaction is the major goal in service industry. How to identify service demands and analyze service appraisal are important issues for decision makers in enterprise. Customer demands are represented in an unstructured wordof-mouth, but demand gaps exist in communication process due to the complexity of human,s semantics. Therefore, this study proposes an analysis mechanism of real customer experiences based on semantic. The mechanism can capture customers, posts automatically, identify their appraisals, and shrink the gaps between customer demands and business cognitions. The results of these experiments proves that this mechanism can effectively extract more than 80% of customers, appraisals, and reach more than 70% of accuracy. The results prove the effectiveness and feasibility of this proposed mechanism in mining service demands.