Literature reading is a process from "beyond reproach", "specious", "big right and wrong" to "much ado about nothing".

At the beginning of reading, due to my limited knowledge, the methods and results proposed in the previous article are brand new to me, and my knowledge reserve is not enough to judge my point of view. So when you start reading, you will accept everything in the article, and it is difficult to have a questioning opinion. Everything you read is beyond reproach. After reading ten articles and eight articles, I know more methods and viewpoints, and some may have different opinions. This is because I doubt that I will jump out and criticize the arguments I see, but I can't determine that one method is completely superior to the other with one or two articles. Each method has its own advantages and disadvantages, so as to reach the realm of "specious". Only by reading enough articles can we make a complete evaluation, and the results of the comparison between the advantages and disadvantages also have sufficient arguments. At this time, it reached a state of "big right and wrong".

Summarize the article, part of which is to synthesize other people's work and part of which is to discuss one's own views. Right and wrong is only a comprehensive step, and it should be further. By grasping the "right and wrong", we should be able to find new problems, new advantages or shortcomings, put forward improvement methods, and put forward suggestions and ideas for future work and development prospects. It is necessary to jump out of the discussion of the details of the original experiment and evaluate it from a higher level such as principle, methodology and system. Of course, this is very difficult, so most of the summary articles are not described, so that readers can understand the past after reading it, and they are still at a loss about the future. Of course, the more you talk, the more time and energy you spend. A prominent example is that it took Zhao Liping two years, adding and deleting more than 20 times, to write an article with his own unique views on the development direction in the next 5- 10, and then he realized that "description" was not easy.

If you can achieve the "right and wrong" economy after reading, you can start writing. In writing, we should first establish a framework to classify and subdivide the content to be summarized. The finer the points, the better, at least three levels, and there may be four or five titles under the three-level title. This has the following advantages:

Frame the questions to be summarized and narrow the scope. Don't let your writing deviate from the direction and lose focus. It's also easy to prioritize, rather than grabbing at the eyebrows and beards. This is not a book, and there is not much space to broaden the comprehensive discussion, so we should limit our review scope.

@ writing pressure is small. When I think of writing an article, I think of "writing at least 5000 words and reading 100 files", and the pressure suddenly increases, and I don't know where to start. After subdivision, all you want is to discuss a parameter of a method, just a short paragraph. This is much simpler and can be finished immediately without too much pressure.

@ organized. All the questions to be discussed are listed there, just like a to-do list, one after another, in an orderly way, and the progress is very intuitive.