CMPT 880/890 Research Process: The Five Questions.

Post on 27-Dec-2015

217 views 2 download

Transcript of CMPT 880/890 Research Process: The Five Questions.

CMPT 880/890Research Process: The Five Questions

OutlineWays of doing researchProblem-based researchThe five questions

What’s the problem?Why should I pay you to solve that problem?What’s the solution?What are the steps from here to the solution?How do you know your solution is a good solution for

the problem?

Ways of doing researchInvent something and tell people about it

“I have created a carrot that tastes like ginger ale”Observe the world and write up your observations

“I have discovered a new species of elephant”Work on a topic out of a book

“I have solved Fermat’s Last Theorem”Compare things

“The Subaru has more leg room than the Honda”

Problem-based researchIn many areas of computer science, our research is

about solving problemsProblems of the real world

E.g., “it is too time-consuming to learn a foreign language”Problems defined by the research community

E.g., “there is no known constant-time algorithm for XYZ”(implicit: “…which makes it too time-consuming to …”)

Problem-based research:The Five Questions

1. Problem: What’s the problem?2. Motivation: Who cares?3. Solution: What’s the solution?4. Steps to the solution: How do we get to the solution?5. Evaluation: How do we know the solution is good?

The ProblemWhat counts as a problem?

“People can’t do X”<Subject> <problem word> <activity or state>

Subjects: People, subgroup of people, machines, algorithms, processes

Problem words: can’t <verb>, has difficulty <verb>ing, can’t <verb> <quality>

Activity / state words: depends on the domain

Problems (?)“I am terrible at Mario Kart”“Nobody knows whether the Loch Ness monster exists”“Android is not as smooth as iPhone”“It’s too hard to learn Icelandic”“We don’t know whether Fermat’s Last Theorem is true”“Frequency caches are better than recency caches”“Mobile technology distracts drivers”“It takes three hours to play League of Legends”

The MotivationWho cares?Why should I pay you to solve this problem?What benefit will there be in solving this problem?

What things do people care about?Why do people to want to solve problems?

The MotivationWho cares?Why should I pay you to solve this problem?What benefit will there be in solving this problem?

What things do people care about?Why do people to want to solve problems?

MoneyHappinessSafetyJustice

What’s the motivation?“I’m terrible at Mario Kart”“iPhones aren’t very customizable”“Activists have their online activity tracked”“People stop paying attention to the world around them

when they are talking on a cellphone”“People have difficulty learning Icelandic”“People get RSI from using a mouse for long periods”

The SolutionIf you have stated the problem clearly, the solution

should be easy to stateProblem: I am terrible at Mario KartSolution: Make it so that !(I am terrible at Mario Kart)

Real solution is in figuring out why the problem existsThe hook

Goal is to discover a general principleSome knowledge we didn’t know before

The SolutionProblem: I am terrible at Mario Kart because I always

oversteer with the game controllerSolution: Make it so that !(oversteer with controller)

Problem: some people oversteer with controllers because the controller joystick is too sensitive for their movements

Solution: Make it so that !(too sensitive)

Principle: input sensitivity should match capability

What’s the solution?“People have difficulty learning Icelandic”

What’s the solution?“People bend their iPhones by sitting on them”

Steps in the SolutionHow do we get to the solution from here?Tells you whether your solution is realistic / feasibleProblem: People can’t jump over buildingsSolution: More practice

(determine human limits on jumping)Solution: Shorten all buildings

(determine architectural constraints)Solution: Rocket boots

(determine cost/benefit tradeoff)

What are the steps?“People with low vision lose the mouse cursor”

The EvaluationIs your solution a good solution to the problem?(First: does your solution even match the problem)What thing would you measure to show that your

solution was good?Should be clear from your explication of the problem

“People have difficulty learning Icelandic”What does “difficult” mean here?People have difficulty pronouncing Icelandic vowel

sounds that are not in their language

What evaluation?Problem: People have difficulty pronouncing Icelandic

vowel sounds that are not in their languageSolution: VowelMeIceland appEvaluation:

What evaluation?Problem: People with low vision lose the mouse cursorSolution: Big cursorEvaluation:

What evaluation?Problem: Pharmacists put the wrong ingredient in a drugSolution: IngredientMinder appEvaluation: