How to skim a research paper

Objectives:

• Normalize struggle inherent in developing paper-reading skills: we’ve all been reading for most of our lives – why is reading research papers so difficult?
• Promote a growth-mindset: this is a skill that is learned, not inborn.
• Introduce idea of self-regulated learning (SRL): in the Ph.D. students become their own teachers, and must therefore thinking of how to develop the best “curriculum” for themselves – one that is the least discouraging and most pedagogical. Use three stages of SRL to explore where we might fail in constructing such a curriculum.
• Teach students how to break-down paper-reading into the smaller, simpler task of answering “guiding” questions – see guide on how to read a research paper.
• Debunk common misconceptions about paper-reading (e.g. that papers should be read linearly, like a novel) – see guide on how to read a research paper

Pre-class work

1. Summarize The PageRank Citation Ranking: Bringing Order to the Web in one paragraph, underlining/highlighting every sentence you read in the process (not every sentence you find important). Answer questions:
   • How many sentences did you highlight/underline? In which sections did you highlight the most sentences?
   • What was your strategy when reading the paper? For example, in what order did you read the sections? Which sections did you focus on most? What did you do when something didn’t make sense?
2. Fill out a poll to schedule office hours

Insight: This exercise helps students actively think about their paper-reading strategy, which they likely have not previously thought about.

In class: [slides]

This unit generally follows the structure of our guide on how to read a research paper.

1. [10min] Students socialize in groups and introduce themselves using the slides they made
2. [5min] As a group, recap takeaways from last class.
3. [10min] Go over “general tips” for how to skim a research paper
4. [Discussion + regroup] Small-group activity 1:
   • Paper: Rx: Treating Bugs As Allergies – A Safe Method to Survive Software Failures. Answer guiding questions (alternating discussion and re-grouping). For each question, ask “which section likely contains this info?” and highlight every sentence read.
   • [10 + 5min] Part A: What type of paper is it? How do we know? What’s the motivation and potential impact?
   • [10 + 5min] Part B: Guiding questions about the related work and contributions, as well as the broader impact
5. [25min] Small-group activity 2:
   • Paper: The Dimpled Manifold Model of Adversarial Examples in Machine Learning.
   • Same as above, but without the structure – just give students a block of time to work through and answer the questions.
6. [10min] Re-group and go over the paper.
7. Logistics:
   • [5min] Explain the “once a semester” assignments
   • [5min] Provide overview of the Union benefits
8. [5min] Ask students to answer the in-class survey:
   • In what ways (if any) did your paper reading strategy change?
   • Is there anything else you took away from class today?