Research Of Text Visualization Based On Semantic

With the development of information technology and network technology, With the development of information technology and network technology, text visualization as a way to quickly understand text information, has been got more and more attention by researchers.It has been made a great progress in the field of text visualization. Researchers have proposed many different ways for presentation. Word Cloud is one of them. It could show the core content of the text for users by the display of key words.This paper firstly introduces the background of text visualization, and then gives a summary of the existing methods. By comparing with the existing frequency-based,semantic-based and multi-level-based visualization methods, we illustrate their advantages and disadvantages, and find that the existing works focus on showing the semantic information of text as much as possible. Since simply random placement of the words has been unable to convey the deeper semantic information of text for users, semantic-based visual text display mode is put forward. How to express the implicit message of text is becoming a problem which needs to be resolved for visualization researchers. For the problem of overlap between different words in text visualization, it also has a very important significance on how to reduce the loss of semantic information after adjustment.Therefore, a new method of text visualization based on semantic is proposed in this paper. By calculating the semantic relationships among key words in the text, we can build a word cloud to put the words by energy model, and mark different types of semantic information by a modular clustering. On the problem of overlap among different words, this paper introduces a new method to solve the problem of word overlap by Archimedes screw adjustment. This method can make a smaller damage after the adjustment of word clouds, and the original semantics are preserved as much as possible. The experiment results show that our algorithm can reduce the damage of the structure of the word cloud compared with existing methods. It satisfies the design requirements of our word clouds. It demonstrates that semantic information of the text can be showed much better by using the real text set of news.