用于引书识别的引书引用模式研究

Study on the Citing Patterns for the Cited Book Recognition

现代信息技术飞速发展,为古籍引书计算机识别提供了可能性。本研究的目的是研究古籍引书的引用模式,进而探讨模式识别方法在引书识别中的应用。具体方法是以明、清及民国时期广东地方志中有关物产门目资料(《广东方志物产》)为语料,抽取所有引书的引用模式(引书名称及其表达方式),分别对引书名称模式和引用表达方式进行研究。

It is possible to recognize the cited books with the rapid development of modern information technology. This paper attempts to explore the citing patterns, and the pattern recognition applying in the cited books recognition. The authors take Local Chronicle of Guangdong： Produce as an example, extracting all the citing patterns from the documents. After that, all the citing patterns are surveyed, including the title citing ways and the citing expression patterns. The title citing ways are classified as the document name, the author name, the author name ＋ document name. The citing expression patterns are classified as pre - symbol, pos - symbol, the seal. In the end, the authors draw out 12 documents randomly to test the citing expression patterns, and that the recall is 84.95%, the precision is 72.88%.