Eddi: Interactive Topic-Based Browsing of Social Status Streams
Download as PPTX, PDF1 like449 views
The document discusses 'eddi', an interactive topic-based browser designed to enhance topic exploration within Twitter feeds, addressing the challenges of topic detection with short texts like tweets. It highlights the limitations of existing information retrieval algorithms and introduces 'tweetopic', a method for real-time topic detection based on tweet analysis. The evaluation results indicate that 'eddi' significantly improves browsing efficiency and user satisfaction compared to traditional chronological feeds.
![Information Retrieval TechniquesAssume decent length to textRepetition as a measure of importance: e.g., Term Frequency – Inverse Document Frequency (tf-idf)Co-occurrence matrices: e.g., Latent Dirichlet Allocation (lda)[Blei et al., Ramage et al.]But with 140 characters, it is difficult to distinguish signal from noise, topic from commentary.katrina_Ron Rivestcracks me up. It keeps me awake when algorithm design brings the lulz.](https://image.slidesharecdn.com/eddi-uist2010-101008185149-phpapp01/85/Eddi-Interactive-Topic-Based-Browsing-of-Social-Status-Streams-11-320.jpg)
![Information Retrieval TechniquesAssume decent length to textRepetition as a measure of importance: e.g., Term Frequency – Inverse Document Frequency (tf-idf)Co-occurrence matrices: e.g., Latent Dirichlet Allocation (lda)[Blei et al., Ramage et al.]But with 140 characters, it is difficult to distinguish signal from noise, topic from commentary.katrina_Ron Rivestcracks me up. It keepsme awakewhen algorithm design brings the lulz.](https://image.slidesharecdn.com/eddi-uist2010-101008185149-phpapp01/85/Eddi-Interactive-Topic-Based-Browsing-of-Social-Status-Streams-12-320.jpg)
![Related WorkDesignTopic browsing interfaces[Kammerer et al., CHI 2009][Leskovec et al., KDD 2009][Käki et al., chi 2005]](https://image.slidesharecdn.com/eddi-uist2010-101008185149-phpapp01/85/Eddi-Interactive-Topic-Based-Browsing-of-Social-Status-Streams-24-320.jpg)
![Related WorkNoun phrases as key concepts in short segments of text[Bendersky and Croft, SIGIR 2008]Search engine callouts to find query similarity[Sahami and Heilman, WWW 2006]LDA on Twitter[Ramage et al., ICWSM 2010]Algorithms](https://image.slidesharecdn.com/eddi-uist2010-101008185149-phpapp01/85/Eddi-Interactive-Topic-Based-Browsing-of-Social-Status-Streams-25-320.jpg)
![Results: More Efficient and EnjoyableLikert Response (Agreement)941Is Quick to Scan“Eddi helps me find things that I’m interested in, faster.” EddiChrono.Is Enjoyable“I get bored faster with the traditional feed. There’s way more stuff that I’m not interested in.” EddiChronologicalI’m Confident I Saw Everything“[The chronological feed] is less enjoyable but more comprehensive.”EddiChrono.](https://image.slidesharecdn.com/eddi-uist2010-101008185149-phpapp01/85/Eddi-Interactive-Topic-Based-Browsing-of-Social-Status-Streams-37-320.jpg)
Eddi: Interactive Topic-Based Browsing of Social Status Streams
- 1. eddiInteractive Topic-Based Browsing of Social Status StreamsMichael BernsteinmitcsailBongwonSuh, Lichan Hong, Sanjay Kairam, Ed H. Chiparc augmented social cognitionJilin Chenuniversity of minnesotamit human-computer interaction
- 4. User Goal: Topic Explorationon trending topics in the feed or topics of interest
- 5. Topic Detection is DifficultExisting algorithms expect reasonably long documentsWikipedia articles: average 400 wordsTweets: average 15 wordsmsbernstmacbook died, but the Genius guys gave me a new one!Existing algorithm might find:macbookdiedguysExisting algorithm might miss:applecustomer support
- 6. eddiinteractive topic browser for twitter feedsTweeTopicTweetNoun PhrasesWeb SearchTopic Keywordsrealtime topic detection algorithm for tweets
- 10. TweeTopicfromtweetmsbernstAwesome article on some SIGGRAPH user interface work: http://bit.ly/30MJyanimationcharacter3dcomputer graphicsuser interfacetotopics
- 11. Information Retrieval TechniquesAssume decent length to textRepetition as a measure of importance: e.g., Term Frequency – Inverse Document Frequency (tf-idf)Co-occurrence matrices: e.g., Latent Dirichlet Allocation (lda)[Blei et al., Ramage et al.]But with 140 characters, it is difficult to distinguish signal from noise, topic from commentary.katrina_Ron Rivestcracks me up. It keeps me awake when algorithm design brings the lulz.
- 12. Information Retrieval TechniquesAssume decent length to textRepetition as a measure of importance: e.g., Term Frequency – Inverse Document Frequency (tf-idf)Co-occurrence matrices: e.g., Latent Dirichlet Allocation (lda)[Blei et al., Ramage et al.]But with 140 characters, it is difficult to distinguish signal from noise, topic from commentary.katrina_Ron Rivestcracks me up. It keepsme awakewhen algorithm design brings the lulz.
- 13. Information Retrieval Techniqueskatrina_Ron Rivestcracks me up. It keepsme awakewhen algorithm design brings the lulz.
- 14. TweeTopic: IntuitionTweets look like search queries, and search results can be mined for topics.
- 15. TweeTopic: IntuitionTweets look like search queries, and search results can be mined for topics.TweetmsbernstAwesome article on some SIGGRAPH user interface work: http://bit.ly/30MJyTweetNoun PhrasesNoun Phrasesarticle SIGGRAPH user interface workSearchWeb SearchTopic KeywordsWeb SearchTopic KeywordsSIGGRAPH 2004 Trip ReportThis year’s themes at SIGGRAPH … good navigation interface …www.stoneschool.com/Work/Siggraph/2004/index.htmlWIMP (computing) – WikipediaPossibility ... (like the noun GUI, for graphical user interface) ...en.wikipedia.org/wiki/WIMP_(computing)SIGGRAPH: Specialty 3D ApplicationsStandalone programs give alternatives to the toolset of a 3D ... maxon.digitalmedianet.com/articles/viewarticle.jsp?id=55098
- 16. 1Noun phrase detectionNoun PhrasesWeb Search Topic Keywordsmsbernst Awesome article on some SIGGRAPH user interface work: http://bit.ly/30MJy
- 17. 1Noun phrase detectionNoun PhrasesWeb Search Topic Keywordsmsbernst Awesome article on some SIGGRAPHuser interfacework: http://bit.ly/30MJy
- 18. 1Noun phrase detectionNoun PhrasesWeb Search Topic Keywordsmsbernst Awesome articleon some SIGGRAPHuser interfacework: http://bit.ly/30MJy
- 19. 2Query a search engineNoun PhrasesWeb Search Topic Keywordsarticle SIGGRAPH user interface workSearch
- 20. 2Query a search engineNoun PhrasesWeb Search Topic KeywordsSIGGRAPH 2004 Trip ReportThis year’s themes at SIGGRAPH … Automatic Distinctive Icons for Desktop Interfaces … such that they actually do provide a good navigation interface …www.stoneschool.com/Work/Siggraph/2004/index.html<html>WIMP (computing) – WikipediaAnother possibility is to have the P in WIMP stand for Program, allowing it to be used as a noun (like the noun GUI, for graphical user interface) rather ...en.wikipedia.org/wiki/WIMP_(computing)SIGGRAPH: Specialty 3D ApplicationsAug 4, 2006 ... SIGGRAPH: Specialty 3D Applications Standalone programs give alternatives to the toolset of a 3D animation application By Frank Moldstad ...maxon.digitalmedianet.com/articles/viewarticle.jsp?id=55098Graphical specification of flexible user interface displaysGraphical specification of flexible user interface displays. Full text, Pdf (983 KB). Source, Symposium on User Interface Software and Technology archive ...portal.acm.org/citation.cfm?id=73673UIST 2010UIST (ACM Symposium on User Interface Software and Technology) is the premier forum for innovations in the software and technology of human-computer …www.acm.org/uist/
- 21. 3Mine topics from resultsNoun PhrasesWeb Search Topic KeywordsSIGGRAPH 2004 Trip ReportThis year’s themes at SIGGRAPH … Automatic Distinctive Icons for Desktop Interfaces … such that they actually do provide a good navigation interface …www.stoneschool.com/Work/Siggraph/2004/index.htmlTF-IDF on a web corpus:sketchmodelpaperGollumcardsanimationmaptextureSIGGRAPHfluidsskincharactershadercolladareal-timeclothsubsurface scatteringBalrogspecial session
- 22. 3Mine topics from resultsNoun PhrasesWeb Search Topic KeywordsKeep terms inat least 50% of search resultsUse less common termsas suggestions
- 23. AppleW00t! Snow Leopard gave me 10 gigs back!RT @username: gmail is down, but the imap connection on my iphone still works (fingers crossed!)My iPhone 3GS cracked-on-a-rock, @username’s swam in a toilet, both repaired/replaced in 20 min @ Boylston Apple Store. Total cost: $0.ObamaI think the most striking thing about Obama’s speech + GOP response for casual listeners would be how much agreement there was.Watching Obama attempt to #reversethecursehealthcareRT @username: The fastest way to prove you are an idiot is to call the President a liar on live TVResearch@username Congratulations on the CSCW best paper nomination!Stanford scientists turn liposuction leftovers into embryonic-like stem cells: http://bit.ly/3GHsw9CORRECTION: the deadline for submissions to the Graduate Student Consortiumfor TEI ’09 is October 2 http://bit.ly/15D8Mv
- 24. Related WorkDesignTopic browsing interfaces[Kammerer et al., CHI 2009][Leskovec et al., KDD 2009][Käki et al., chi 2005]
- 25. Related WorkNoun phrases as key concepts in short segments of text[Bendersky and Croft, SIGIR 2008]Search engine callouts to find query similarity[Sahami and Heilman, WWW 2006]LDA on Twitter[Ramage et al., ICWSM 2010]Algorithms
- 26. EvaluationHow does TweeTopic compareto other topic detectionalgorithms?How does Eddi compareto a typical chronologicalTwitter interface?TweetNoun PhrasesWeb SearchTopic Keywords
- 27. TweeTopic EvaluationComparison topic detection algorithmsRandom UnigrammsbernstAwesome article on some SIGGRAPH user interface work: http://bit.ly/30MJy
- 28. TweeTopic EvaluationComparison topic detection algorithmsRandom Unigram
- 29. Inverse Document Frequency (idf)msbernstAwesome article on some SIGGRAPHuser interface work: http://bit.ly/30MJy
- 30. msbernstAwesome article on some SIGGRAPH user interface work: http://bit.ly/30MJymsbernstAwesome article on some SIGGRAPH user interface work: http://bit.ly/30MJymsbernstAwesome article on some SIGGRAPH user interface work: http://bit.ly/30MJyTweeTopic EvaluationComparison topic detection algorithmsRandom Unigram
- 31. Inverse Document Frequency (idf)
- 32. Latent Dirichlet Allocation (lda)msbernstAwesome article on some SIGGRAPH user interface work: http://bit.ly/30MJygraphics
- 33. TweeTopic Evaluation100 random tweets from Twitter’s streamThree human coders rated the top five recommendations from each algorithm (Fleiss’sκ=.70)Logisticregression analysis for binary outcomesvideo gamesmedal of honorreviewshonorYup, Medal of Honor will have a demo http://bit.ly/bx6PSG
- 34. Results: TweeTopic Doubles BaselineTopic Labeling AccuracyOdds Ratio (baseline = 1 at Random Unigram)
- 35. LDA vs. TweeTopicI’m off to take a nap now. See y’all in a few hours!LDAbedhalfhoursleepTweeTopicnaptimepower napsleeptake a nap
- 36. Eddi EvaluationRecruited active Twitter users, preferring those who followedmore than 100 peopleGave users 3 minutes to browse 24 hours of their feed using Eddi or a chronological interface, over 6 total trials
- 37. Results: More Efficient and EnjoyableLikert Response (Agreement)941Is Quick to Scan“Eddi helps me find things that I’m interested in, faster.” EddiChrono.Is Enjoyable“I get bored faster with the traditional feed. There’s way more stuff that I’m not interested in.” EddiChronologicalI’m Confident I Saw Everything“[The chronological feed] is less enjoyable but more comprehensive.”EddiChrono.
- 38. Results: Twice As EffectiveTrack tweets remaining onscreen for > 2 secondsGet relevance judgments from users:“I’m glad that I saw this tweet in my feed.”Users consume a purer feed:
- 39. Discussion and Future WorkEddi is most useful for overwhelming feeds @msbernst follows 1000 @msbernst follows 100 @msbernst follows 10peoplepeoplepeopleUse case: filter accounts with selective interests “Show me @GuyKawasaki when he tweets about social computing; ignore the rest.”
- 40. eddiInteractive Topic-Based Browsing of Social Status StreamsExplore an overwhelming feed by topics of interestUncover the central topic of a tweet,given very little text
- 44. TweeTopic EvaluationTweeTopic VariantsTransformed vs. Raw:Do we massage the tweet to look like a query?
- 45. Iterated vs. None:Do we keep removing words if the search engine fails?4Iterate to remove words if neededarticle SIGGRAPH user interface work
- 46. Results: Noun Phrase Analysis UnnecessaryTopic Labeling AccuracyOdds Ratio (baseline = 1 at Random Unigram)
- 47. Related WorkTwitter and DesignCommon uses of Twitter: information sharing, opinions, status[Naaman et al., CSCW 2009]50%40%30%% of all tweets20%10%0%InformationSharingOpinionsRandomThoughtsPersonalStatus
- 48. edchli
Editor's Notes
- #3: Twitter is exploding in popularity as a communication medium and as an information source. But it’s also exploding in diversity and scale. Here’s a small slice of my feed – it covers a wide variety of topics. Some of these I care about and some of these I don’t.
- #10: TODO: photoshop the image to call out the items via color, rather than using these shapes
- #14: You don't want those as topic headings. "Cracks, keep, and lulz."
- #25: Todo: make full-slide related work
- #30: Shout out to ramage
- #31: Controlled for tweet, rater, and topic rank
- #32: Wald tests with bonferonni correction confirmed that it outperforms all the others
- #37: He tweets ~once every four minutes
- #44: Wald tests with bonferonni correction confirmed that it outperforms all the others
- #45: Todo: make full-slide related work