应用BTM模式之中文短语切分的的研究.pdfVIP

應用BTM 模式之中文短語切分研究 77 東南技術學院 東南學報 第二十八期 中華民國九十四年六月出版 第77-86頁 應用BTM模式之中文短語切分研究 A Study of Applying BTM Model on the Chinese Chunk Bracketing 蔡 佳 麟 Jia-Lin Tsai Tung Nan Institute of Technology, Department of Information Management, Assistant Professor Abstract The purpose of this paper is to automatically generate Chinese chunk bracketing with bottom-to-top mapping (BTM) model. The BTM model is designed as a supporting model with parsers. We define a word-layer matrix to generate the BTM dataset from CKIP Chinese Treebank (CCTB). Our model matches auto-learned patterns and templates against segmented and POS-tagged Chinese sentences. A sentence that can be matched with some patterns or templates is called a matching sentence. The experimental results have shown that the chunk bracketing of the proposed approach on the matching set (the set of matching sentences) for perfect and actual input is able to achieve more than 95% F-measure when training size is ≥ 5,000, POS layer number is ≥ 2 and BTM threshold value is ≥ 0.5. Keywords: phrase, parser, template matching 摘 要 本篇文章描述一個可自動產生中文短語切分的BTM模式。此BTM模式是用來支援 中文剖析器。我們定義了一個詞層矩陣，用以轉換樹圖資料庫為BTM資料庫。一個句 子若可對應到一個BTM特徵或模版，此句子被稱為可對應句。我們的實驗顯示，此 BTM模式在訓練語料大於等於五千句，詞性標記大於等於兩層，及BTM閥值大於等於 零點五的條件下，在可對應句上，可達到大於95%F-measure的短語切分能力。 關鍵詞 ：短語，剖析器，模版匹配 78 東南學報第二十八期 I. Introduction The definition of chunk, which has been represented as groups of words between square brackets, was first raised by [Abney, 1991]. A chunker is to divide sentences into non- overlapping phrases by starting with finding correlated