Chest X-ray is a prime imaging test in the healthcare sector to detect and treat abnormalities and diseases in the lungs, heart, and bones. Proper interpretation and annotation of chest radiographs by the radiologist or Artificial Intelligence Clinical Assistant Diagnosis system should be ensured for high-accuracy computer-aided diagnosis. By treating chest X-ray projections Posterior-Anterior (PA) and Anterior-Posterior (AP) as separate datasets to train the deep learning models, the accuracy of disease prediction can be improved. Automatic segregation of AP and PA view CXR images reduces the burden and time of manual process. Also in a Healthcare Digital Archiving System, the chest radiographs can be maintained separately as AP and PA groups under frontal view. The proposed work exploits the major techniques in Natural Language Processing (NLP) i.e., Optical Character Recognition (OCR) and pattern matching to detect the projection type of the labeled chest X-ray images and then classify them into two views, AP and PA. We created the dataset with chest X-ray images collected from various dataset sources including NIH etc. This work also compares the proposed NLP-based CXR view classification system with a transfer learning-based system to classify chest X-ray images into PA and AP projections.