Abstract: This study presents a systematic approach for the classification of large corpus based on concept granulation and equal partition approach. The proposed research has three main processes which are the preprocessing treatments to text documents, feature extraction and finally the classification. The proposed approach is concentrated in the feature extraction phase. Almost bird eye view like approach is the feature extraction method. So, the proposed research concept granulation and equal partition approach has been named as Immune Term (TIM) which finds the immunized terms from the information system. At first, documents are preprocessed from text to numerical form, i.e., word frequency is calculated for each document. Second, sets of features are extracted using TIM. In the third step, the TIM treated feature is introduced to Principal Component Analysis (PCA) and Latent Semantic Indexing (LSI) for global set extraction or dimension reduction. Finally, Naive Bayes (NB) and Support Vector Machine (SVM) are used to classify the documents. The proposed research seems to be fruitful when compared to the conventional word frequency approach.