The Design And Application Of Document Model In Consensus Information System

Consensus referrs to the newest socio-political attitudes held by people along with the occurrence, development and change of social events. People act as the main body and politicians act as the object. Consensus information is a description of and reflection of public opinion.In the consensus information system, the performance of public opinion can have a variety of forms of information. There is no fixed standard. As the access to public opinion, the information system of the consensus comes from a variety of data sources. Information extraction and analysis of public opinion used in the procedures and algorithms are fixed. This requires a variety of data sources to build a unified storage model.This paper presents a tree-based document model which has close connection with the operation of the document model of learning to meet the needs of users.This article is based on a cooperation project with Beijing Municipal Center. It is mainly to complete the following work:1 Get abstract data from different sources and the establishment of tree-based document model;2 After document model is established, go on to do information extraction, information filtering, and model the various branches of the extraction and filtration; 3 After treatment, the document model can be serialized to the database is stored in, and the classification and the provision of retrieval; 4 Users operate by tapping into the information provided, so that the document model can be self-adjusting,self-optimization, to achieve feedback loop of the features that make the document more accurate model can exist in the systems.In the design and development process, the application of a number of computer technology, such as database connection pooling, object-oriented java programming language, GUI part of the library and JfreeChart used in swing package, the software unit testing and a number of development ideas of Extreme Programming and so on. In the information extraction, each data source including PDF, Word and Web Page has a corresponding data adapter. The classification of the document model, using Bayesian classifier, can dynamically adjust the parameters of feedback devices.