LIP: finding good/relevant papers

finding-papers.md

LIP: Finding good/relevant papers

Search/read with intentionality

        old <-------------------> new
theoretical <-------------------> system

"Most similar papers" will probably be towards the (new, concrete) end

• "Good"
  • Types of contributions: http://faculty.washington.edu/wobbrock/pubs/interactions-16.pdf
    • Research is novel work; the contribution is what specifically is novel about it
  • Good writing: good papers have a way of articulating your nebulous ideas into something really catchy and concrete
• "Relevant"
  • Just starting a new project and trying to form a RQ: understand theoretical underpinnings/consensus ⇒ older, more theoretical work
  • Once you have an idea of an intervention: get state of the art ⇒ new, concrete systems papers
  • Inspo/refinement while building
    • Tech approach: newish, concrete. Look at technical approaches (may be before or after description of system)
    • Conceptual: broadest category, anything helps here (introduction and results/discussion)
    • Study design: from pretty much any time (unless there are specific surveys that were developed recently), papers that are measuring the same thing as you (evaluation)
  • Writing first draft of manuscript
    • Backsolve the citations for theoretical claims you're making (Example: intro from http://brrian.org/papers/uist2015-explain-visual-changes.pdf)
    • Make sure your related work section is super complete: look at MSPs, take union of their RW sections, etc.
      • Look at other people's reviews
      • If your work differs in a significant way from existing SOA systems (as it should), you may need to bridge RW sections from more distant fields! (Example: http://users.eecs.northwestern.edu/~scl025/files/ply.pdf)
    • Look for nice ways to state concepts: MSPs (introduction, discussion)

Iterative algorithm

1. Get roots: disciplinary terminology, authors, paper titles/system names
   • Ask for examples of good papers and authors related to your area
   • Google ideas that seem relevant to you, or phrases you've heard (then identify the most promising ones)
2. Follow footnotes
   1. Decide whether you're going to do this DFS or BFS
   2. Look mostly at related work
      • When you're actually writing, it's not unreasonable to include a superset of the closest paper's entire related work section
   3. Identify the most promising ones

Evaluating how useful a paper is

• Authors in a field who tend to be widely-cited (look at the first, last author)
• Seminal theoretical papers (tend to be older, may be from other areas like cogsci, LS, AI, etc.)
  • Look at impact factor or # of citations on ACM DL or Google Scholar
• Conferences/publication venues: CHI, UIST, HCOMP, Ubicomp, VL/HCC, TOCHI
  • Overlapping HCI conferences: ICER, SIGCSE, ...
  • Google "SIG" + domain
• Keep track of terms of art: domain-specific phrases or names for concepts
  • exploratory programming, end-user programming, cognitive load

Keeping track of papers

• Use NU VPN for access to ACM DL
  • Most papers in CS and mathematics are open-access, authors will have preprints on their websites, CHI proceedings are open-access for 1 year after each conference
  • Much less true for other disciplines: psych, cogsci, LS
• Citation managers and browser extensions
  • Papers - Your personal library of research
  • Mendeley - Reference Management Software & Researcher Network
  • Zotero | Home
• Tag or make notes of papers: key takeaways, reasons you might want to reference it later, highlight important parts
  • Example: KA Hivemind
  • Example: Ply related work collection
  • Easy export to BibTeX