arrangement from link
read a paper
Abstract -> Introduction -> Important Diagrams -> (understand main topic) -> Conclusions
-> [Technical Details]
Abstract -> Conclusion -> (how the problem is solved in the paper) -> Diagrams -> (impression)
-> (known what to find) -> focus on such infomation
Abstract -> Diagrams -> statement on topics
-> Full Content -> Previous Work from the group
-> interesting reference
-> think about
-> if data is consistent with my own. if not, why?
-> what if I adopt his way
-> why he did the experiment like this
Abstract -> (it is interesting) -> Introduction -> (understand problems) -> Figures and Tables
-> discuss and find
-> his new ideas
-> related work to mine
-> same as mine before (works and assumptions)
take it as writing -> find sentences about topics -> write it in a word document
related works -> find out method and result (important)
-> experiment design
-> data analysis
read -> create it on your own
detailed operation on start topic survey
summarized from 台湾清华彭明辉教授的研究生手册.pdf
Abstract and Introduction
1. find survey / review -> 30 ~ 40 papers
2. Abstract + Introduction, sometimes Illustrative examples & Conclusions (read those with sensible Intro)
- a. what are the common methods in this domain ?
- b. how could the methods be divided into different branches ?
- c. what are the characteristics, merits and defects for methods of each branch ?
3. if still have trouble for the above questions:
- 1. find all papers based on (2a), put them into categories, ordered by date
- 2. read one branch a time, Abstract + Intro: find the basic petitions and ideas, note down the merits
- 3. read Intro: how they criticize other branches, note down the drawbacks
4. research and retain mostly 20 papers related to your work, even remove some branches reasonably that you don’t want to use
5. answer the additionnal questions based on (2c), get the main application and matters needing attention for each application, in this domain
- 1. what are the key problems in this domain?
- 2. what characteristics(merits and drawbacks) people care usually?
- 3. under what circumstances people will care more about the merits and drawbacks?
- 4. when do people care less about them?
6. organize papers into categories, ordered by date. read main bodies for them according to their relation and importance to your work
Main body, experiment and simulation
before reading, take care of the three questions:
- 1. what are the main assumption of it, how hard to meet the assumption in reality
- 2. based on those assumptions, what are the good points for this paper
3. what’s the representation of these good points, (which formula item is simplified)
- only care the key formulae (in final application)
- focus on the assumptions rather than the induction, if you really need to read formulae
read along all the papers in this domain easily, find those easier to read and the most cited, answer the three questions above.
After all, based on (5) and these 3 questions, answer the 4th: main drawbacks of this branch.
Use the 4 answers to make conclusion: when the branch is suitable and when it’s not.
when confronted with unknown things
- search engine for passages
- read after I have some background knowledge
- search keywords and so on so forth, end with a relationship between them
- ask the author if it make trouble for understanding
- ask colleagues or the author