Research On Dimension Obtaining And Measure Calculating Technology Of Text Olap

Information data is more and more important in modern production and life. The data warehouse is widely used for storing and processing data by large enterprises and the government. OLAP (online analytical processing) is a powerful tool of data warehouse, data processing. OLAP technology can observe and analyze the data in different dimensions, providing analysis of historical data and predicting the future trend for enterprises and the government.Information data can be roughly divided into structured data and unstructured data. For structured data, the traditional data processing methods such as relational databases have been able to quite effectively analyze it. While no effective methods can analyze the explosively growing unstructured text data, which is as important as structured data by the measure of value.If we can apply OLAP technology to the analysis of unstructured text data, this treatment of text OLAP on unstructured text data will become much effective. Some of the current research results such as Text Cube, Topic Cube, have made contribution to this direction. They have different principles and characteristics and they are classified in general into three groups of information retrieval, text mining and information extraction.On the basis of the existing research on text OLAP, this paper converts unstructured data into an intermediate form of semantic network. To calculate the text of the OLAP metrics, this paper proposes an approximation algorithm for comparing semantic networks. Based on statistics of unstructured data semantics, we assist manual work of building subject trees by semantic network fragments, and help get text dimensions half-automatically using neural network.This can improve the accuracy of current analysis on text OLAP, and reduce labor cost on building dimensions of text OLAP.