因為建立或維護詞庫型意見探勘系統,需耗費大量人力或時間,本研究試圖將這些耗費大量人力的工作簡化,原來建立或維護詞庫時需要看很多文章,並挑出針對領域有用的意見詞之後新增至詞庫,現在只要執行演算法將文章內新的意見詞挑出,之後判斷演算法所挑出的新意見詞及比照新意見詞和句子是否為對領域有用的意見詞並新增至詞庫。由於檢查演算法挑出的新意見詞比直接看文章來得省時省力,進而達到降低詞庫建置或維護的成本。 Due to huge manpower and time consuming for establishing and maintaining a word library type Opinion Mining system, the study attempts to simplify the operation of huge manpower and time consuming, thus the original reading many articles for establishing and maintaining the word library, where Opinion Word useful to the domain have to be pickup and added to the word library, with the process of the study, it only needs to execute algorithms to pickup new Opinion Word in articles, then determine whether the pickup Opinion Word is useful to the domain by comparing the pickup Opinion Word and sentence, and then add to word library. Owing to checking new Opinion Word picked out by algorithms is much time and manpower saving comparing to reading articles directly, moreover word library establishing and maintaining cost can be reduced.