A comparative study on method comment and inline comment

Code comments are one of the important documents to help developers review and comprehend source code. In recent studies, researchers have proposed many deep learning models generate method header (i.e., comment), which achieved encouraging results. The in method, is called inline comment, also for program comprehension. Unfortunately, they not received enough attention automatic generation when comparing with comments. this paper, we compare analyze similarities differences between By applying existing generating comment generation, find that these perform worse on task generation. We then further explore possible reasons obtain a number new observations. For example, there lot templates same or similar structures) dataset, makes better. Some terms were thought be (e.g., API calls) by previous study does significantly affect quality generated comments, seems counter-intuitive. Our findings may give some implications building approaches future.