Ph.D. research topic: Guiding from Dr. Ming Zhou

This morning, I got up at 8:30. When I came to MSRA, it was about 10:00. Accompanied by Yi Chen and Jizhou Huang, I came to Dr. Ming Zhou’s office. Dr. Ming Zhou was my mentor in my first four months in MSRA. I asked some questions about Ph.D. research topics. He gave me a lot of guidance, I concluded as following:

1) To a Ph.D. research topic, you could do some applied subjects. Ph.D. subject should be very different with masters’ and undergraduates’. They should be very useful, novel, and few or no people had researched. To usefulness, it must be for practical application, such as for industrial goal or some big systems modules. Novel is same as few or no people have researched it. Yeah! If many ones have do a lot research on it, you should not do it again. Because it is time consuming and could not achieve more high level.

2) To each person, you should push over and clear yourself. If you are doing things with more old experiences, you could not be fresh and novel again. Coreference resolution is a very old research topic. Nowadays, many researchers had done lots of works on it. IBM had achieved best performance on it. If you cost three months, you will finish it. But there is no more space for your Ph.D. research.

3) Considering a Ph.D. level research topic, in the beginning, you should put yourself into a blank position. You should consider nothing especially to your old experience. The current industrial hotspots and applications would be surveyed. Then you can find out the hottest and most frontal points. Surveying such point, you could find out the one or two points for your research. Then you can start your Ph.D. research. There are two examples: Kaifu’s Statistical Speech Processing and Paraphrase in NLP.

4) Research topic should be systematical and practical. You should change your viewpoint from deeply point to whole architectural one.

I conclude Dr. Ming Zhou’s suggestion to me in three words: practical, novel, and potential. In our talking, Dr. Ming Zhou disclosed some possible research topics: Blog and RSS in Web 2.0, Multi-lingual multi-document summarization for QA, Comparison Shopping. He gave me a hint: finding research topic about Web. It was the way out.

Sincerely thanks to Dr. Ming Zhou. I will call at him again when I return to Harbin.