Hacker News new | past | comments | ask | show | jobs | submit login

I ask myself two questions -

1. Is this paper of interest to me? 2. How will I assimilate the knowledge of this paper into my existing framework?

For (1), the title+abstract helps a lot. Followed by the conclusion, which can indicate whether there is something really to be absorbed.

For (2), I need some grasp of the "related work" - i.e. "how have others tried to solve this problem before this work?" So once I affirm interest, I might scan some of the references. If I'm familiar with them, I know enough to assimilate. Otherwise if the paper describes them adequately for the purpose, I settle for that with a note-to-self to dive into anything in detail.

A third kind of "always on" mode for me is "can I reuse any techniques or results in this work?" and I read more carefully to gain that.

For abstract and structure, my preferred structure (because I was trained like this) is .. in this specific order .. -

1. Introduction - broad sweep description of area to clue people in.

2. Related work - a survey of how people have solved some key thing.

3. Problem statement - What remains unsolved/unaddressed relative to the "related work".

4. Solution - How does this solve it. This includes methods.

5. Validation - How does the author claim that their solution actually solves the problem they set out to tackle. In some cases, this may involve user studies, in others it may be a mathematical result, etc.

This is both my communication framework as well as a knowledge assimilation framework.

edit: I find an abstract that's pretty much like one sentence or so for each of those items in that order easiest to assimilate. However, I admit this wasn't easy to develop initially and I struggled a lot with it before it became normal.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: