Academic Recommendation

The proposed work is intended to develop a semi-automatic technique for classifying the sentiments based text.
Basically in this presented work for text classification the decision trees are implemented which are the
supervised learning algorithms. But the additional effort for tagging of data is necessary therefore that is known
as the semi supervised model. The technique is applied on the online student’s learning and discussion to find
the students experience and improve the experience in further learning processes. Therefore the entire system
development is performed on two major modules first for generating the student’s communication data and then
uses it with the supervised model for text classification according to the user emotions. In this technique the web
data is first pre-processed for improving the quality of learning data. After that the data is used with the NLP
parser for finding the communicated text features. The computed text features are than used with two different
supervised learning models namely C4.5 and the ID3. These models are basically a kind of decision trees,
during training of these algorithms the algorithm generates the tree. These generated decision are termed here
as the trained model. The trained classifier is further used for real time classification of communicated text for
the binary classification. The implementation of the entire text mining concept is performed using the JAVA
technology and the classifiers performance is compared using the accuracy, error rate, memory consumption
and time consumption. According to the computed classifiers performance the proposed technique namely c4.5
based classifier perform more accurate classification as compared to the traditional ID3