HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor
Report
Share
•3 likes•474 views
![User Comments
70
“[Associations] helped me think of jokes in
a wider conceptual space than I
previously had.” (p9)
Before this, I never really knew where to
start with joke writing I just kind of sat and
thought until I came up with something. (p5)](https://image.slidesharecdn.com/chiltonmugstalk-2016-07-13v2-160720165236/75/humortools-an-adaptive-microtask-workflow-for-crowdsourcing-humor-70-2048.jpg?cb=1668449693)
HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor
•3 likes•474 views
Report
Share
Download to read offline
Background on Lydia Chilton's crowd algorithms work, Maneesh Agrawala's algorithms for generating good designs from design principles, and past and future work on crowdsourcing humor
More Related Content
What's hot
What's hot(20)
Similar to HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor
Similar to HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor(20)
Recently uploaded
Recently uploaded(20)
HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor
- 1. HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor Lydia Chilton 13 July 2016 MUGS Lunch Talk, Stanford
- 2. Outline • My Background: Crowd Algorithms – TurKit – Cascade & Frenzy • Related Work: Automated Design – LineDrive • Current Project: HumorTools: – The Onion’s American Voices – Descriptive Models of Humor – Breaking Down Humor into Microtasks – Reducing the Humor Search Space – Crowdsourcing Humor
- 3. Summary
- 4. Simple Crowd Task: Label this image Complex Crowd Algorithm: Decipher this message Greg Little, Lydia B. Chilton, Max Goldman, Robert C. Miller. TurKit: Human Computation Algorithms on Mechanical Turk. UIST 2010. Crowdsourcing: Decompose Problems so People Can Collaboratively Solve them
- 5. Decomposing and Automate Design: Model Design as a Search Problem To automate well-designed visual communication: 1. Collect design principles from experts 2. Define a search space 3. Define Constraints 4. Define an objective function to maximize Maneesh Agrawala, et al. Design Principles for Visual Communication. CACM 2011.
- 6. Humor is a Difficult AI Problem Model Humor Generation as Constraint-based Search Descriptive Theories of Humor: Constraint: “Violate an expectation”
- 7. Outline • My Background: Crowd Algorithms – TurKit – Cascade & Frenzy • Related Work: Automated Design – LineDrive • Current Project: HumorTools: – The Onion’s American Voices – Descriptive Models of Humor – Breaking Down Humor into Microtasks – Reducing the Humor Search Space – Crowdsourcing Humor
- 11. Adaptive Mechanism: Improve-and-Vote 11 initial artifact output improve vote improved original
- 12. Problems too hard for one person 12 You (misspelled) (several) (words). Please spellcheck your work next time. I also notice a few grammatical mistakes. Overall your writing style is a bit too phoney. You do make some good (points), but they get lost amidst the (writing). (signature)
- 13. 13 initial artifact output Improve-and-Vote Builds on Contextual Clues from Other Workers You (?) (?) (?) (work). You (misspelled) (several) (words). improve vote improved original You (?) (?) (?) (work). (?) (?) (?) (?) (?).
- 14. 14 MTurk TurKit Seaweed Soylent VizWiz Adrenaline Legion PlateMate Crowdy Cascade Frenzy … … … Other Crowd Algorithms TurKontrol Deluge
- 15. Cascade Distributes big problems with complex interdependencies such as creating taxonomies 15 Lydia B. Chilton, et al. Cascade: Crowdsourcing Taxonomy Creation. CHI 2013.
- 16. Frenzy Combines the knowledge of many experts to meet a global constraint such as creating conference sessions Lydia B. Chilton, et al. Frenzy: collaborative data organization for creating conference sessions. CHI 2014. 16 Used at CSCW 2013 and CHI 2014 (136 papers, 431 papers)
- 17. Outline • My Background: Crowd Algorithms – TurKit – Cascade & Frenzy • Related Work: Automated Design – LineDrive • Current Project: HumorTools: – The Onion’s American Voices – Descriptive Models of Humor – Breaking Down Humor into Microtasks – Reducing the Humor Search Space – Crowdsourcing Humor
- 18. LineDrive: Using Design Principles to Automate Design Problems 18
- 19. LineDrive: Using Design Principles to Automate Design Problems Methodology • Expert patterns • Observations from examples of expert work • Grounded in psychological principles. 19
- 20. LineDrive: Using Design Principles to Automate Design Problems Methodology • Expert patterns • Observations from examples of expert work • Grounded in psychological principles. 20
- 21. Applying Design Principles • Generative Rules: – Define design spaces • Descriptive Rules: – Identify quantitative evaluation functions 21
- 22. Outline • My Background: Crowd Algorithms – TurKit – Cascade & Frenzy • Related Work: Automated Design – LineDrive • Current Project: HumorTools – The Onion’s American Voices – Descriptive Models of Humor – Breaking Down Humor into Microtasks for Individuals – Reducing the Humor Search Space – Crowdsourcing Humor
- 23. Creativity Intelligence 23 Humor is a Desirable Property of Communication
- 24. Humor evokes Mirth 24 • Surprise • Delight
- 25. Persuasion 25
- 27. Education 27
- 28. Humor is Hard for Computers 28
- 29. Descriptive Theories of Humor
- 32. Descriptive Theories of Humor are Not Generative “Write something that violates an expectation.”
- 33. Approach To Creating a Generative Model of Humor 33 Examples + Theories Model + System + Study
- 34. 34 “I can’t believe people would just walk into an Apple store and start breaking things like it’s a Best Buy.” American Voices Examples People Bending iPhones At Apple Stores Friday September 30, 2014 Input Output
- 35. 35 1. Associations 2. Expectation Violation Books on humor by philosophers, linguists, comedians Theory
- 36. Satisfy a Constraint by searching a space of associations 36 Headline Joke Expectation Violation Apple Broken DisplaysBest Buy Best Buy Alternative Insult Associations Model
- 37. Sure, I guess UConn’s course is fine if you couldn’t get into Yale’s football clinic. UConn Holding Football 101 Clinic for Female Fans Association: Alternative Thing 37
- 38. Alright, how many ‘Summer Savannah’ Backyard Garden Lion Pedestals do I have to order to turn this thing around? Report: ‘SkyMall’ Magazine May End Print Edition Association: Details 38
- 39. Just tell me what to do and where to go to guarantee happiness forever. Poll: Elite Colleges Don’t Produce Happier Graduates Association: Personality Flaw 39
- 40. Satisfying a Constraint by searching a space of associations 40 Headline Joke Expectation Violation Sarcasm Unexpected Angle Bait-and- Switch Apple Broken DisplaysBest Buy Best Buy Alternative Insult Associations Model
- 41. Study: First Born Children More Ambitious As a second-born girl, I’d just like to say wooohoo! Spring break! Bait-and-Switch Bait: The study is false. Switch: The study is true. 41 Expectation Violation #1
- 42. Unexpected Angle “It’s an exciting time to be a shark, that’s for sure!” Great White Shark Populations Surging Off East Coast Expected Focus: This is bad for people Unexpected Focus: This is good for sharks 42 Expectation Violation #2
- 43. Sarcasm “There is absolutely no other explanation.” Mick Jagger Blamed for Brazil’s World Cup Defeat 43 Expectation Violation #3 Literal Subtext: This is true Sarcastic Flip Subtext: This is false
- 44. How Not to Make a Joke Static Workflow Why not? Does not adapt to the input The same recipe won’t apply to every headline.
- 45. 45 Headline Joke Expectation Violation Sarcasm Unexpected Angle Bait-and- Switch Apple Broken DisplaysBest Buy Best Buy Alternative Insult Associations Instead, use an Adaptive Workflow That searches the space of associations until a constraint can be satisfied
- 46. Outline • My Background: Crowd Algorithms – TurKit – Cascade & Frenzy • Related Work: Automated Design – LineDrive • Current Project: HumorTools – The Onion’s American Voices – Descriptive Models of Humor – Breaking Down Humor into Microtasks for Individuals – Reducing the Humor Search Space – Crowdsourcing Humor
- 47. HumorTools Adaptive Workflow • Apply microtasks to the headline • Until you can find a way of violating an expectation • Follow your train of thought, until you get stuck, then backtrack • Not yet crowdsourced • But it is decomposed into microtasks 47 Aspects Violation Mechanism Headline Prototype & Test Joke Expected Reactions Associations
- 48. HumorTools Interface 48 Justin Bieber was baptized this weekend in an attempt to wash away his sins following a scandal in which the singer appears in videos uses racial slurs
- 49. Justin Bieber Baptized In NYC BathtubHeadline Joke 49
- 50. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects 50
- 51. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bad. His music sucks. Expected Reactions 51
- 52. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bad. His music sucks. Expected Reactions Violation Mechanism 52
- 53. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bad. His music sucks. 53
- 54. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR peopleAssociations Bad. His music sucks. 54
- 55. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR peopleAssociations Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Violation Mechanism 55
- 56. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR peopleAssociations Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Violation Mechanism Prototype Test Joke … …… … Never let it be said that Bieber’s PR people aren’t bringing new ideas to the table. 56
- 57. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR peopleAssociations Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Violation Mechanism Prototype Test Joke … …… … Never let it be said that Bieber’s PR people aren’t bringing new ideas to the table. 57
- 58. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR people Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Joke … …… … 58
- 59. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR people Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Joke Expected Reactions Bad. Dirty, Gross … …… … 59
- 60. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR people Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Joke Expected Reactions Bad. Dirty, Gross Violation Mechanism Unexpected Angle: A Bieber fan who’d worship his dirty bathtub … …… … 60
- 61. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR people Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Joke Expected Reactions Bad. Dirty, Gross Violation Mechanism Unexpected Angle: A Bieber fan who’d worship his dirty bathtub Prototype Test …… …… …… … 61
- 62. Justin Bieber Baptized In NYC BathtubHeadline Joke Justin Bieber Baptized Bathtub Headline Aspects Bieber’s PR people Bad. His music sucks. Unexpected Angle: Expected Reaction: Bad Angle: PR people are clever Angle Reaction: Good. Joke Expected Reactions Bad. Dirty, Gross Violation Mechanism Unexpected Angle: A Bieber fan who’d worship his dirty bathtub Prototype Test …… …… …… … Oh my God! Can I lick the tub? 62
- 63. HumorTools Study • 20 users: Stanford undergraduates • Humor novices • 60-80 minutes: – Tutorial on 20 microtasks – Humor Writing • Write jokes for 3 headlines • Rate jokes against The Onion (21 people) 63
- 64. HumorTools Study Headlines 64 PETA Seeks Copyright for Primate A lawsuit filed by PETA claims that “monkey selfies” snapped by a macaque who stole a photographer’s camera should be considered the legal property of the macaque himself. Liquid Water Found on Mars NASA revealed Monday that they have found evidence of liquid water on the Mars, pointing to the possibility of life on the red planet. Can users synthesize humor?
- 65. HumorTools Evaluation • 75% of participants were able to synthesize humor • 25% of the HumorTools jokes were rated funny 65
- 66. 66 PETA Seeks Copyright for Primate This is great news for animal rights. Now my neighbor can take my dog to court instead of me the next time he deposits some of his “legal property” in their front lawn. This is why you always get the macaque to sign a release.
- 67. 67 PETA Seeks Copyright for Primate This is great news for animal rights. Now my neighbor can take my dog to court instead of me the next time he deposits some of his “legal property” in their front lawn. This is why you always get the macaque to sign a release. HumorTools 47% Funny 62% Funny
- 68. 68 Liquid Water Found on Mars I think it’s more convenient for me to get it from the tap. In other news, Pluto has been called even less of a planet now. Sounds refreshing!
- 69. 69 Liquid Water Found on Mars HumorTools HumorTools I think it’s more convenient for me to get it from the tap. In other news, Pluto has been called even less of a planet now. Sounds refreshing! 15% Funny 45% Funny 45% Funny
- 70. User Comments 70 “[Associations] helped me think of jokes in a wider conceptual space than I previously had.” (p9) Before this, I never really knew where to start with joke writing I just kind of sat and thought until I came up with something. (p5)
- 71. Outline • My Background: Crowd Algorithms – TurKit – Cascade & Frenzy • Related Work: Automated Design – LineDrive • Current Project: HumorTools – The Onion’s American Voices – Descriptive Models of Humor – Breaking Down Humor into Microtasks for Individuals – Reducing the Humor Search Space – Crowdsourcing Humor
- 72. Problem: The Search Space is Big
- 73. Solution: Maximize expected utility by selecting microtasks that are correlated with successfully making jokes
- 74. Headline Types (8 total) • Science – Scientists Developing Heat-Resistant Chickens To Withstand Climate Change • Celebrity / Entertainment – Mattel Announced Barbie Movie • Absurd – Company Unveils 'Drinkable Sunscreen'
- 75. Same Related Insult Alternative Detail POV AVERAGE 0.876 1.170 0.104 0.456 0.328 0.452 American Politics 1.00 1.093 0.130 0.389 0.407 0.440 Business 0.821 1.321 0.154 0.436 0.295 0.530 Entertainment 0.753 1.358 0.049 0.420 0.444 0.421 Science 0.877 1.231 0.036 0.536 0.290 0.588 Celebrity 0.636 1.273 0.242 0.394 0.636 0.389 Sports 1.133 1.133 0.333 0.933 0.200 0.333 Tech 1.111 0.944 0.139 0.417 0.500 0.429 Absurd 0.884 0.977 0.109 0.403 0.194 0.487 Frequency per joke (588 jokes total) Red indicates “well above average” Association TypesHeadline Type P(joke contains an association type | Headline Type)
- 76. Headline Type Correlation with Joke Strategy
- 77. Joke Strategies for Entertainment Headlines: BE TOO SERIOUS
- 78. Joke Strategies for Absurd Headlines ACCEPT THE ABSURD PREMISE
- 80. Future Work: Automate Some Microtasks Train ML system with user collected data 81 • Entity extraction • Sentiment analysis • Semantic parsing
- 81. Grant: Be funnier than the Onion 82
- 82. Summary
- 83. HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor
- 84. Crowdsourcing: Decompose Problems so People Can Collaboratively Solve them Simple Crowd Task: Label this image Complex Crowd Task: Decipher this message Greg Little, Lydia B. Chilton, Max Goldman, Robert C. Miller. TurKit: Human Computation Algorithms on Mechanical Turk. UIST 2010.
- 85. To Automate Design, People Model Design as a Search Problem 1. Collect design principles from experts 2. Define a search space 3. Define Constraints 4. Define an objective function to maximize Maneesh Agrawala, et al. Design Principles for Visual Communication. CACM 2011.
- 86. Humor is a Difficult AI Problem Model Humor Generation as Constraint-Based Search Descriptive Theories of Humor: Constraint: “Violate an expectation”
- 87. HumorTools: An Adaptive Microtask Workflow for Crowdsourcing Humor • My Background: Crowd Algorithms – TurKit – Cascade & Frenzy • Related Work: Automated Design – LineDrive • Current Project: HumorTools – The Onion’s American Voices – Descriptive Models of Humor – Decomposing Humor into Microtasks for Individuals – Reducing the Humor Search Space – Crowdsourcing Humor
Editor's Notes
- Thanks for inviting me. I’m a post-doc here at Stanford in the HCI group. I’ll be starting as Computer science faculty at Columbia in 2017.
- I’m going to talk about humor but, I’m also going to give you some background on my area, which is Crowdsourcing, and HCI.
- Let me quickly summarize everything I’m going to say in the next 90 slides.
- The objective of Crowdsourcing is to decompose problems so people can collaboratively solve them. Crowdsourcing can be very simple tasks like image labeling or more complex tasks like Deciphering this message where no one person knows the answer. My work is on creative crowd algorithms that decompose harder and harder problems.
- Other people in HCI have decomposed hard problems like design. In fact, didn’t have to use the crowd, there were able to automate some design problems. Maneesh Agrawala in the HCI group here has several papers that automatically create well-design Visual Communication: Generating readable map directions “how things work diagrams” Assembly instructions. All of these problems were decomposed in a similar way:
- Humor is a difficult AI problem and very hard to decompose My work is inspired by Maneesh’s work, and treats humor generation as a constraint-based search.
- Crowdsourcing was invented to use people for tasks that computer can’t do like labeling images.
- Mechanical Turk is a platform where you can pay people a few cents to do a simple task. Turk quickly became very popular, But for a long time, Turk was only used to do embarrassingly parallel tasks like image labeling. So I asked: Can we do more complex things?
- But my coauthor and I built a system called TurKit. which introduced the concept of Crowd Algorithms. You can write JavaScript programs that makes function calls to the crowd.
- The most basic adaptive mechanism is improve-and-vote. Which allows workers to build on the work of others.
- For example, Here is some messy handwriting. No one person could read it. I can’t even read it! Using improve and vote, the crowd could decipher it. Here’s how it works…
- It starts with the image, At first we have no idea what it say, so we start with question marks Then we ask a worker to improve it, and they get parts of the answer, but other parts are wrong. Then we vote on which version is better. Usually the improved version is better, but not always. Sometimes you get spam, or just really weird interpretations. You take the version workers voted for, and you improve it again, until you get it all deciphered. This little portion says “You misspelled several words…” and goes on.
- TurKit was widely adopted in research. Systems I built are in red, but other people used TurKit and built crowd-powered systems, too. Notably, Michael Bernstein at Stanford and Jeff Bigham at CMU built their first crowd systems in TurKit.
- Cascade distributes big problems with complex interdependencies like creating taxonomies
- Frenzy combines the knowledge of experts to meet a global constraint like Grouping papers into conference sessions.
- Those patterns are all completed by humans, but if we understand the space well enough, it is possible to solve problems automatically with design principles. LineDrive is a system for creating useful driving directions based on visual principles.
- But what people who give good directions do is more like this: How can a computer do that automatically? It’s so stylized? So different from a map. Methodology for automating the creation of these is to look at expert patterns.
- Expert patterns, Observations from examples of expert work Grounded in psychological principles.
- Design Space: angle you show the turn at. Can exaggerate it to emphasize the orientation of the turn. All Roads must be visible. Quantitative Evaluation: minimum road length (visibility threshhold), You want as many of the important roads as possible to meet this visibility threshhold.
- Humor is a highly valued human skill. It is a sign of creativity and intelligence (Images – Monty Python. Obama laughin?)
- Humor is used in many domains: - Persuasion in advertising and politics
- Humor is used in many domains: - Persuasion in advertising and politics
- It is used to capture attention
- And it helps us learn. Because when we are having fun and at ease we learn better than when we are stressed and pressued.
- Humor is hard. Humor is an AI Challenge Robots can’t do it. Why? Level of understanding experience and cretivity they don’t have. Data.
- Incongruity theories the notion that humor violates our expectations. It is the detection of the incongruity that we find funny. Aristotle mentions this, modern philosophers such as Kant and Schopenhauer latched on It is now the dominant theory of humor.
- Incongruity theories the notion that humor violates our expectations. It is the detection of the incongruity that we find funny. Aristotle mentions this, modern philosophers such as Kant and Schopenhauer latched on It is now the dominant theory of humor.
- Humanity has been wondering about humor for over 2,000 years, so I’m obviously not go to start from scratch. My approach is to mine lots of examples and theories, and build a system that embodies them. It basically takes descriptive theories and turns them into generative theories.
- For examples, I decided to narrow in on one very special form of humor. It’s made by The Onion, called American Voices. It takes real news headlines like this one: “People Bending iPhones At Apple Stores” And The Onion write fake man-on-the-street style responses: What’s special about this is that the headline and joke represent input and output pairs that we can study.
- But what are we going to going to analyze these jokes for? Well, I distilled a lot of books on humor by philosophers, linguists, and comedians, and two ideas kept coming up: Associations Expectation Violation.
- I prototyped a number of different generative models of humor, I settled on this model: Satisfying a constraint by search a space of associations. Expectation Violations is a goal. But it’s too hard to hit right off the bat. So I use associations to search for possible ways to violate expectations. By looking at examples, I found concrete mechanisms that could be applied. For example, Apple Store to Best Buy are Alternatives. That’s one type of association Best Best to Broken Displays is an Insult. That’s a different type of association. And there are multiple types of expectation violation. Let me explain each of those.
- Like all good writing, details matter. Details bring the reader into the world your are writing.
- Like all good writing, details matter. Details bring the reader into the world your are writing.
- I prototyped a number of different generative models of humor, I settled on this model: Satisfying a constraint by search a space of associations. Expectation Violations is a goal. But it’s too hard to hit right off the bat. So I use associations to search for possible ways to violate expectations. By looking at examples, I found concrete mechanisms that could be applied. For example, Apple Store to Best Buy are Alternatives. That’s one type of association Best Best to Broken Displays is an Insult. That’s a different type of association. And there are multiple types of expectation violation. Let me explain each of those.
- In the Best buy joke, it starts my aggressing with your expectation. This is horrible, I can’t believe it. Then at the last minute, it switches to be a Best Buy insult. I call this “bait-and-switch”.
- For another example of Violation Mechanisms, here’s a different joke. The headline is HEALDINE The Onion writes: “So did he think Transcendence was good, or what?” You expect the focus to be the end for Mankind, and how horrible that would be. But instead, the focus is on the movie.
- Lastly is sarcasm. Which hardly needs to be explained.
- I prototyped a number of different generative models of humor, I settled on this model: Satisfying a constraint by search a space of associations. Expectation Violations is a goal. But it’s too hard to hit right off the bat. So I use associations to search for possible ways to violate expectations. By looking at examples, I found concrete mechanisms that could be applied. For example, Apple Store to Best Buy are Alternatives. That’s one type of association Best Best to Broken Displays is an Insult. That’s a different type of association. And there are multiple types of expectation violation. Let me explain each of those.
- I built a system called HumorTools that teaches people the microtasks for creating humor and helps them follow them. The general guideline is that you Apply microtasks… Find a way of violating an expectation There are a lot of options, so just follow your train of thought until you get stuck, then backtrack.
- Here is what one real user interface looks like: This asks users to decompose a headline into aspects. There are 20 of microtasks.. It’s too big to show you all of it. Instead, let me walk you through an example that illustrates the process.
- We start with the headline Justin Bieber Basptized in NYC Bathtub
- We break it into aspects
- We pick the aspect “Justin Bieber” and start writing our expected reactions. I expect this is bad. His music sucks, he’s a bad role model.
- But I can’t think of a violation mechanism for that.
- So the I forget that chain of reasoning, and backtrack
- Let’s try an association. Bieber has PR people who cover up for his stupid behavior. They’re the ones working hard to make it seem like he’s repentant.
- Ah! And there’s a violation here! Although we expect the headline to be bad, From the perspective of the PR people it’s good
- “Never let it be said that Bieber’s PR people aren’t brining new ideas to the table.”
- And to get there we followed a train of thought, and did a little backtracking.
- We can write multiple jokes this way.
- If we start with Bathtub, we expect that to be dirty and gross
- But to a fan, that they’d worship him anyway.
- That’s a different unexpected angle joke.
- I ran a study on HumorTools using 20 Stanford undergraduates.
- I’m going to read you two headlines from the study. … The question is can users synthesize humor for them?
- The evaluation showed that 75% of participatns were able 25% of the jokes were rated funny
- Let me read show you some jokes: For the PETA headline where PETA seeks copyright for primate Here are two jokes for this headline: (read) Both have the look and feel of the Onion, and are amusing.
- The first one was actually written by The Onion. The second was synthesized with HumorTools Raters found the HumorTools joke funnier than The Onion
- For the Mars headline where Liquid Water Found on Mars, indicting the strong possibility of life on the red planet. Here are three jokes for this headline: (read) All 3 have the look and feel of the Onion, and are amusing.
- The first two are humorTools And “sounds refreshing” is The Onion Both HumorTools jokes were funnier than The Onion
- HumorTools is currently done completely by people, but since each microtask is fairly small, we could start to Train Machine learning systems with user-collected data, leading to hybrid human-computer joke generation
- I have a grant to fund myself and 6 students this summer to be funnier than The Onion and release HumorTools