The Design And Implementation Of Assisting System Used For Manually Annotating Text Information

For the past few years, with the impulse of the information search, information extraction and machine translation's technology and demands, natural language processing technology developed rapidly and became an independent subject, which has drawn greater attention. Language resources construction is the basic research field of natural language processing research; discourse annotation is the important aspect of text information processing and language resources construction. As a powerful tool for discourse annotation, the development of the system is an important portion, which has a direct impact on the efficiency and the quality of discourse annotation, is also an important research direction and tries in the text information processing.Based on the corresponding theory of annotation, this paper designed and achieved the discourse annotation system for context computing, and assisted built a text information annotation corpus for the content computing.Extracting information of the sudden events timely and effectively is an important aspect of the sudden events' response. According to the relatively strict format of sudden events, we choose the sudden events' news as annotating targets.This thesis is based on the background of relevant team's research, the theories on discourse annotating, and the text-based annotating tools. This research regards the meaning units in the sudden events' news as the annotating targets, and accomplishes a hierarchical and categorized annotating.The study consists of the following tasks:1. Identify sudden events news text classified and hierarchical annotating set, find out the generated XML document's elements and hierarchical structure: According to the identified text annotating for the annotate theory, it can select the appropriate keywords for annotating content and related attributes, also reflect the hierarchical structure, the keywords and their values in the XML file.2. Perform the transform of meaning units from the linear modules to the structure text: the entity, entity-relationship, event, event relationship, time and any other meaning units, should be extracted from the original text, and we add the keywords, which in the corresponding annotation congregation, finally it becomes the structure text of the XML format.3. Complete the text information supporting system design and implementation: According to mark demands, it designed and implemented a text annotating system, and completed a certain number of text annotating.In this paper, we design and achieved a interface-friendly discourse annotation system, thus enhance the efficiency and quality of the discourse annotation, and automatic annotation has been carried on for as many fixed patterns as possible. It's the concrete practice of discourse annotation theory, which provides a reference annotation solutions and modules for content computing.