抄几段重要的：
How to choose a research topic?
我想这是很多 PhD candidates 的烦恼吧。至少我是这样，这个问题从读 master 的时候就困扰着我。直到现在，boss 给选了个不错的题目，但仍然受到方方面面的质疑，一边怀疑一边求证，也许这就是 research 的痛苦所在吧。
"Investigate a classical, fundamental, deep, and narrow topic, with some (perhaps shallow) applications to a couple of hot applied topics."
选一个有热门应用的题目，最终能把这个问题归结成一个基础的经典问题，然后把这个问题做深做透。
我想现在我做的 skyline 就是这样的一个问题，本身是一个很老很经典的数学问题，被数据库的 researcher 们用在了数据库领域的实际问题中。
The relationship between supervisor and student is?
"Like a marriage that lasts for 3-4 years?!" --b
Researchers have several roles to play:
"1. Create and adjust models of problems;
2. Solve model problems;
3. Evaluate solutions to model problems;
4. Present solutions to other researchers and customers."
How to evaluate the quality of a solution?
"1. Effectiveness: logically correct;
2. Elegance: beautiful, simple;
3. Efficiency: use computational resources efficiently." (time / space complexity)
在CS领域，好像大家比较偏重第三条，而忽视了前两条。我想第一条或者需要太多数学知识，大多数人比较头疼，所以略过不证了；或者有些道理比较显然，所以大家也略过不证了。第二条太抽象，尽量让解决问题的方法简单易懂，也就是遵循奥卡姆剃刀原理了。
Publish your paper in:
"NLCs: nice local confereces;
IK - CCs: international killer-competitive conferences;
Avoid scams!"
How to write a good conference paper?
------------------------------
| Motivation and Everyone understands.
| background
3------------------------------
| Main
| Results Experts understand.
|
|
8------------------------------
| Conclusion Everyone understands.
8.5----------------------------
| Refernces
10-----------------------------
经典啊经典！
The frequency of presenting your research:
1. At IK-CCs: at least twice in 3 years; 比较难！
2. At NLCs
3. To research visitors to your lab: twice per year; 由访问学者访问的频率决定 --b
4. As a poster / web page: often; 小气如我。。。。
5. At postGrad sessions: continuously; "postGrad session?" 是个啥？！
6. To your supervisor: often; 当然了
7. To your associate supervisor: very often; 没有
8. To your sister...: often. 没有
Something about the presentation (ppt...):
-------------------------------
| Motivation
5------------------------------
| Overview of
| the research Everyone understands.
|
15-----------------------------
| Something
| difficult Some understand.
20-----------------------------
| Overview and
| Charts of experiments
23-----------------------------
| Conclusion
25-----------------------------
每次做 presentation，如果将不明白，都会非常 frustrated。甚至跟朋友开玩笑说，做 presentationde 的时侯，会感觉自己像个厚脸皮的推销员，拼命地把自己的 idea 推销给听众，尽管连自己都怀疑这个 idea 是不是实用。 --b
How to write your 3+ years of PhD research?
Based on your thesis, thesis, thesis!!!
Content of a thesis = 3 good journal papers. 对于 CS 专业是不是有点儿高了？日本是1篇了。
Not based on:
Computer systems that you have written;
Undergraduate tutorials that you have given;
Ideas that you have had but not written down. 最遗憾的是这条。
The books recommended:
1. J. Zobel, "Writing for computer science", Springer 1997.
2. E. M. Phillips and D. S. Pugh, "How to get a PhD", open University Press, 2000.
3. D. Lindsay, "A guide to scientific writing", Longman Cheshire, 1984.
原文的 pdf 版：
