Skip to content

Latest commit

 

History

History
215 lines (140 loc) · 7.89 KB

lecture1.md

File metadata and controls

215 lines (140 loc) · 7.89 KB

    home | syllabus | submit | chat

    © 2017 tim@menzies.us

Lecture001

Before we get down to SE theory, our first lectures will explore the tools and techniques needed for Project1 (due end of January).

So here is a pretty messy collection of notes and tips on various tools. Some of this we will work through in class, some you'll need to read for yourself.

Research

Finding Choice (TM)

Exercise1:

  • Problem: "Make it easier to track day to day expenses"
  • Think, pair, share:
    • 2 minutes. Dream of options
    • 5 minites: pair with your neighbor, share ideas, try to find better ones
    • 8 minutes: around the room "what are our best ideas?"

Exercise2:

  • Problem: "how to find a car park"
  • Go to Google scholar and see what you can find

Reading the literature (TM)

Dancing through the years with google scholar

  • Case study "problems with parking"

Places to search for papers:

Note that these search engines have "cited by" fields so if you find a good paper, you can run forward in the literature looking for subsequent studies.

Pruning rules (heuristic, take with a grain of salt):

  • Recent better than past
  • Highly cited better than not

Methodology:

  • To grab an initial set of documents, what were your search terms? How many documents did it match (hint: usually 100s to 1000s)
  • To prune the intial set , what were your strategies (extra search terms? quick scans of abstracts? number of citation counts? restrictions by year?)
  • After pruning, how many docs did you have (hint, usually dozens)?
  • How did you study the pruned docs (e.g. what particular questions did you ask of each document? how did you collate the results)?
  • What sanity checks were applied? - Did the literature review find known key articles in the field - Potentially, is it possible to operationalize fixed to the identified problems? e.g. lack of domain knowledge very hard to operationalize; not closing matching brackets (easier to operationalize). Recall the Royce results: 22% of their problems were deep semantic while the result were simple logic and syntax problems.

(R.C. Bryce, A. Cooley, A. Hansen, N. Hayrapetyan, A one year empirical study of student programming bugs in Frontiers in Education Conference (FIE), 2010. )

Github (EF)

<iframe width="560" height="315" src="https://www.youtube.com/embed/0fKg7e37bQE" frameborder="0" allowfullscreen></iframe>

Create an Organziation in Github

One organization has many repositories.

Github Isses and Labels

Github issues: tutorial and A simple styleguide for tagging Github issues.

Using master and branches

Newbie GH:

  • everyone commits to master. nothing fancy
  • FYI- I commit several times a day
  • probably suffices for your small class projects

But if you want to see how the gurus do things:

Data Collection Tools

Googleforms (TM)

See https://support.google.com/docs/answer/87809?hl=en

<iframe width="560" height="315" src="https://www.youtube.com/embed/cm3KyqbaMJA" frameborder="0" allowfullscreen></iframe>

Mechanical Turk (Jack)

Introduction to crowd sourcing

Talking to people (TM)

Good news:

  • To evaluate your software, you have been given access to a group of 65× 10 / 20 = 32 people you can use a 1 hour eval session. - Which you can use for your January and March work when you ahve to evaluate software - So, in theory, groups of 16 people

Bad news:

  • That group is.... you. To earn 10 marks for this subject, you will have to volunteer for 5 × 1 hour eval sessions in Jan and March
  • Scheduling nightmare. It'll be... messy. Realistically, getting groups of 5 to 10 to comment on your software in Jan/Mar will be more achiveable.

In any case, once you've got those people, what can you do with them?

A wide range of options. You will most probably use or or more of:

  1. Semi-structured or structured interviews
  2. Questionaire
  3. Focus groups
  4. Think aloud
  5. a analysis methods

Exercise 3: Think-pair-share

  • Rank each of the above 5 points, plus GoogleGorms and Mechanical Turk, in terms of cost and effectiveness in gaining insight on software
    • 2 minutes. Dream of options
    • 5 minites: pair with your neighbor, share ideas, try to find better ones
    • 8 minutes: around the room "what are our best ideas?"

Polcies with People Data

No matter what is done, all sessions must start with

  • Identifying who you are (some contact details passed to subject)
  • A gift. Muffins. Coffee. Pizza. Something to make them feel wanted.
  • Explanation of what we are doing
  • Stress the user's rights - The right not to participate (they can leave, now, if they want to) - The right to privacy (user identifiers will NEVER be stored with data) - The right to be forgotten (on request, you WILL delete this user's data)

After that, you could...

  • Run sample exercise 1 (done by experimenters)
  • Run structured exercise 2 (what the user does, perhaps with a thinking
  • Conduct structured debrief (e.g. ask them to fill in a questionniare)
  • Do something unstructured (e.g. ask them to think aloud, reflect; here you are looking for

Reporting (TM)

Pecha Kucha

20 slides, 20-30 secs each slide. For example: http://tiny.cc/timm7.

Go to slides.google.com

  • Write some slides (20, max). - Be kind. Add page number and short URLs to all slides - Always have a “for more info” slide
  • File > Publish to web
  • Edit link. - Set start=true (auto start's slides) - Set delayms=20000 (20 secs per slide)
https://docs.google.com/presentation/d/
1GgwBtcyni0VXvm9PLzx3oQl8C4_lcdKNK8T2SWTohQI/
pub?start=true&loop=false&delayms=20000&slide=id.p

Pecha Kucha: Tips, Resources & Examples

Latex (EF)

Sharelatex

  • http://sharelatex.com
  • Remember: push the big blue button.
  • Link the sharelatex project to a repo in your organization
  • at least one member of each group has to get the "collaborator" plan (normally $15/month, student plans for $8/month) - Can link to Github - can create a share project and invite others.
  • the rest can live in free "Personal" land

Bibtex

Adding figures (btw, NOT excel graphics)

You probably won't need this, but for completenesss.