2002 Volume 16 Issue 2
Published: 15 April 2002
-
Select all|
-
2002, 16(2): 1-6.Identifying company names in running texts plays a significant role in financial information extraction. Based on the thoroughly investigations of financial articles ,the relevant structural features and contextual constraints were obtained. In this paper ,a company name identification system is proposed ,which is built on the six knowledge-bases and a twice-scan method. The experiment achieved 97.3% precision and 89.3% recall respectively by close test ,and 62.8% precision and 62.1% recall respectively by open test .
-
2002, 16(2): 7-13.In the paper ,a method to improve the generalization performance of the Chinese text classifier is put forward. Generally speaking ,a text classifier is obtained by training text set with a machine learning method. A kind of common sense about text semantic invariance is introduced. A method to improve the text classifier is put forward by fusing the common sense into it . With the combination with a Support Vector Machine ,we design and implement the improved text classifier. The experiment shows that the generalization performance of the text classifier is improved with the method.
-
2002, 16(2): 14-19,26.In the paper ,we discuss chinese text structure from the point of view of formalization. Formal descriptions for the heading ,subheading and paragraph as well as their structural relations in a text are present and a systematic tagging method for chinese text structure is proposed. And then , we introduce the conceptions of the normal text and quasi-normal text . On the basis of these ,indexing algorithm for formal chinese text structure is discussed. The methods and results presented in the paper are of direct and practical significance for fulltext indexing and structural analysis and processes of chinese text .
-
2002, 16(2): 20-26.Prediction of Chinese phrase boundary location is the base of shallow parsing or chunk parsing. It is also very important for processing real texts. With the support of our Chinese treebank including 64426 words , this paper designs and implements a method for automatic prediction of Chinese phrase boundary location based on neural network. The preliminary results show that the precision is 93.24%(close testing) and 92.56%(open testing) respectively.
-
2002, 16(2): 27-33.This paper introduces emotions and their computation to machine understanding of Chinese.Based on the existing research of the machine understanding of Chinese ,the problem of the emotional meaning tagging of Chinese was studied by using the multiple relaxation alternate algorithm. We designed an experimental system and obtained accurate Emotion tagging matching by using the context information. It is the first step to the machine understanding of the emotional meaning of Chinese sentences ,thus the foundation for the future research is laid.
-
2002, 16(2): 34-39,59.This paper presents an English generation system based on the case grammar representation of Japanese which is used in a Japanese - Chinese machine translation system developed by the paper′s authors. The case grammar based Japanese analysis and the representation of analysis results are given first . Then ,the generation policy ,especially ,the generation rule system is presented. At last ,some difficult issues in English generation are discussed.
-
2002, 16(2): 40-46.“是”sentence is a special and often used sentence in mandarin Chinese. This paper examines“是”sentences in the corpus ,and presents some rules for automatic bracketing the subject and object of them. The result is agreeable :the precision is over 99% in the sentences containing no comma and 89% in the sentences containing comma.
-
2002, 16(2): 47-53.The automatic evaluation of output quality for machine translation systems is a difficult problem. Most research in this section is for writing language ,which is quit different from oral language. Therefore ,this paper provides a new automatic evaluation algorithm for speech translation system. It defines the block according to the word position ,tags the information weight and designs several matching methods. All these make the automatic evaluation results more closely to the score of human. And the automatic evaluating can also process more different sentences. Several examinations are given at the end of the paper. They show the algorithm has a good sensitivity for different output quality and its evaluation results is as same as the human’s feeling.
-
2002, 16(2): 54-59.With the improvement of character recognition capability ,processing of the more and more complex layout is a very important part in the document processing system. In order to adapt the characteristics of Chinese layout , based on the component-based layout analysis algorithm , this paper proposes an adaptive morphology based layout analysis algorithm. This algorithm adds the adaptive expedition into the kernel of component-based layout analysis-character combination ,and gives it more adaptive capability to process a large range of multi font size and multi interval samples.
-
2002, 16(2): 60-64.Combining with the development of financial document OCR system ,this Paper presents a new idea about knowledge-based binarization for financial document . Aiming at different kinds of recognition fields (fields of financial document that need to be recognized) ,this paper constructs a suit of binarization methods. The effectiveness of those methods is proved by practically applying.
