Ali Naserasadi - Computer Group, Zarand Higher Education Complex, Zarand, Kerman, Iran,
Majid Estilayee - Technical and Engineering, Payam-e Nour, Tehran, Iran,

The automatic identification of text subject is the basis for a variety of information analysis mechanisms, such as document classification, retrieval, and frontier identification of fields. Therefore, the research on automatic identification methods of text subject is significant. In This paper the key technologies of current text subject recognition are systematically investigated, including the method of obtaining topic words, the calculation of the correlation strength of knowledge units, and the subject analysis methods and practices for multi-relationship fusion. On the basis of summing up the shortcomings of the current text subject recognition methods, the paper proposes a comprehensive method for acquiring topic words, which is used in combination with the extraction range and the grammatical and semantic levels.

Topic Recognition; Text Analysis; Subject Mining; Semantic Analysis