文字探勘（Text Mining）是以資料探勘的方式進行文件的文字資料分析，並透過這些分析取得文字間的相關性，進行分類、比較、判別。近十年來，搜尋引擎崛起，文字探勘的技術被更有效應用，創造新的商業價值。隨著網際網路的日新月異，網路資料量的累積使得搜尋引擎的發展愈發快速，改寫了資料檢索不變的定律。 文本相似度（Text Similarity）透過將文字型態之間予以權重（或做：距離），計算文字型態間的相似程度，並加總比較以取得資訊、分類或二元判斷。透過此方法將大量的文章段落進行分析，並取得富含價值的有用資訊。 本研究將提出一個新的相似度比對方法。我們將文件中任意一個連續的文字視為一個段落（Segment），將此段落與其他句子之間比對獲得評分，並從評分分數的高低與分佈，找出在同一文件中相似的目標段落。本研究以聖經四福因書作為案例，演示演算法運作方式與預期結果，並針對不同的參數之欲其結果進行比較。;Text Mining is known as data analysis to documents based on data mining. Main purpose of text mining is to obtain the relevance between text, through these analyzes conclude classification, comparison and discrimination. Over the past decade, search engines have emerged, and text search techniques have been more effectively applied to create new business value. With the ever-changing Internet, the accumulation of information on the network makes the development of search engines more quickly, also makes a huge on change data retrieval. Text Similarity, the degree of similarity between the text types is calculated by weighting (distance). Calculate the degree of similarity between text types and obtain information, classify or binary judgments, observe the valuable information through analysis a big quantity of articles. In this research, we raised a new method of similarity calculation. We treat any part of continuous sentences in the document as a Segment. Compare this segment with other sentences to get scores, and find the similar target segment in the same document from the rank and distribution of the scores. In this research, we use the four gospels in holy bible as cases study. The cases study demonstrate the operation of the algorithm and the expected results.
