Natural language processing based interpretation of skewed graphs

Different graphical tools such as pie charts, line charts, bar charts, scatter diagram, histogram etc. are used for data representation in statistics. Textual information along with these graphical tools plays vital role in the analysis of quantitative data. Multimodal documents use skewed graphics as a significant tool for the representation of visual information. It is frequently experiential that the communicative objectives of the skewed graphics are not confined by credentials accompanying text. To distinguish the symbolized information using skewed graphics is assertive job for greenhorn readers. An approach to mechanize the progression of graph classification and information withdrawal is offered in this paper. This study spotlights on the skewed graphics that are vital type of area charts used for probability distribution and testing of hypothesis process. To begin with, we have classified the area charts into diverse classes and then designed structural design for graph image classification and information extraction from every class of area chart. The extorted information is represented in the structure of natural language abstract using pattern based approach.