Abstract: Taking sentence in reports as analysis unit, this paper proposed a new research framework that can analyze the content structure of government work report based on word correlation in order to visually and quantitatively reveal the development of government works in our country. This paper chose Chinese government work reports as essential data, and extracted subject words from the text. Correlation networks at various levels were constructed according to co-occurrence relationship between pairs of words, and then indicators of the whole and individual network were calculated. In addition, the network structure, the evolution venation and trends were visualized to reveal main directions of government works, key fields, as well as their interactive structures and future trends. Results indicate that Chinese government work covered a wide range of fields during the past years. It is promoted steadily in a holistic approach, taking programmatic and responsive characteristics into consideration. Chinese government work remained consistent and continuous in changes of plans and the central government rotation, as well as greatly responsive to external environment and changes of tasks. Chinese government work mainly concentrated on 10 directions and 9 key fields, and there were two large-scale evolution venations. In conclusion, this paper effectively reveals connotations, structural characteristics, and development trends of government work through deep text mining of Chinese government work reports.

Key words: Government work, Government work report, Subject correlation network, Co-word analysis, Visualization