An implementation of graph based text classification technique for social media

The data mining is a task of analyzing the data and their hidden patterns. Therefore the classification and clustering approaches are used for performing this task. The classification of the data needs to be performing training over the similar patterns and using the training knowledge the upcoming patterns are recognized. In the data mining the text analysis is one of the popular aspects of the data mining. In this presented work the text analysis and their classification techniques are studied. In addition of that a new text classification approach is introduced for analysing the social media text. The proposed technique helps for automatically analysis of the text patterns and obtaining their positive and negative orientation according to the learned patterns. The proposed technique utilizes the word probability, sentence formation probability and the weighted indexing of the data for recognizing the orientation of the text data. The proposed technique organized in such manner by which the different combinations of the words can be analysed using the graphical approach. Therefore the computed probability and weights are used for generating the directed weighted graphs. Using this graph the upcoming technique is classified and their orientation is obtained. The implementation of the proposed text classification data model is provided using JAVA technology and their performance is also evaluated. The performance of the system is reported in terms accuracy, error rate, memory usage and the time complexity. According to the different experimental analysis the proposed technique is found accurate and efficient for text classification. Additionally that can also be extendable for finding the classes more than two classes.