本論文在於提出並擬建構一套中文化數字輔助記憶系統,透過系統的運作,將一串無意義的數字輔換成有意義的中文詞句,以輔助使用者記憶。主要利用中央研究院詞庫小組的八萬詞目中文詞庫,分析注音符號聲母與韻母,找出114組與0~9數字發音近似的注音,分別給予不同的配分,再與詞庫中的詞彙比對,找出12680組與數字發音近似的詞彙。利用字串轉換機制(Transformation mechanism)將一連串數字予以切割,找出與詞庫中發音近似的詞彙予以組合配對。 本研究中配對以dynamic programming方式處理,時間複雜度為O(n3*(m+y)),其空間複雜度為O(n2*m),其中n為數字字串長度,m為欲挑選組句的個數,y為挑選m組時,失敗的組合次數。詞句組合上,透過評估函數F(evaluation function)分別給予不同權值,經過自然語言的分析後,從中挑選評估函數值前m個組句,做為我們挑選的組句。並在組句過程中自動產生二元剖析樹(Binary parsing tree),以方便後續研究之用。實驗結果顯示我們的系統對大多數的使用者有不錯的助憶成效。
In this thesis, we develop a system called Computer-Assisted Mnemonic System (CAMS), which can assist people to memorize some meaningless numeric strings with Chinese sentences. We analyze phonetics of the numbers from 0 to 9 and find out 114 rules of similar syllables. The rules are matched with the lexicons corpora established by CKIP (Chinese Knowledge Information Processing) Group of Academia Sinica in the Republic of China. Our system finds 12680 lexicons that can be used in composing sentences. We use a transformation mechanism to segment a series of numbers and thus compose sentences. In the research, we use dynamic programming to implement the unification process of CAMS in O(n3*(m+y)) time by using O(n2*m) space of memory, where n is the length of the input numeric stings and m is the number of compositional sentences that we want to output, y is the times that our system gets unsatisfied composition results. Among the satisfied composition results, we utilize evaluation function to judge whether each of those composition sentences is suitable for us to memorize or not. In this process, the system will generate a binary parsing tree of sentences for further research in the future. Experimental results show that our system has a high degree of mnemonic effect for most users.