Research Methods for Data Science (2018/2019)

Research Methods for Data Science (2018/2019)

Literature Survey Exercise

The purposes of this exercise are to:

You will read several papers for this exercise, at varying levels of detail. To prepare yourself, first find and read:

Reading this paper first will save you a great deal of time in this exercise. A few comments in the paper are specific to networking (the author's own area), but most are highly relevant to reading a paper in any field.

Now, choose a paper that interests you and which has over 100 citations. This will be the main paper that you will work with during this course. You might want to scan (see Keshav) several papers before making your choice. We recommend choosing a paper that you expect to be relevant to your own Master's thesis. If you have difficulty finding a suitable paper, you can consult the lists of suggestions below.

Read your chosen paper thoroughly, and make sure you understand it. Expect this to take 4-5 hours (see Keshav).

Select at least one (and preferably more) older paper from the list of references in your chosen paper, and read it (1 hour, see Keshav). Try to choose a paper that is itself highly cited, and which your chosen paper builds on directly.

Select at least one (and preferably more) later paper which cites your chosen paper, and read it (1 hour, see Keshav). Try to choose a highly cited paper that builds directly on your chosen one.

Your goal is to understand your chosen paper thoroughly, and also to understand its significance in a broader setting - how it relates to what came before, and what came after.

This exercise should be done individually.

There is nothing to be submitted for this exercise, but don't delay: the next exercises in the course, with tight deadlines, build directly on this one. Don't underestimate how long it will take you to find and select the papers you want to read in detail; this exercise involves several days' work, and can't be done the day before the deadline. Aim to complete it before the next lecture.

Lists of suggested papers:


Last Modified: 19 September 2018 by Graham Kemp