Dialogue-Earth:-Mining-Social-Media

  • 1,502 views
Uploaded on

Social media mining presentation by Tom Masterman of the Institute on the Environment, given at the University of Minnesota's Communicators Forum Conference, 05.13.10.

Social media mining presentation by Tom Masterman of the Institute on the Environment, given at the University of Minnesota's Communicators Forum Conference, 05.13.10.

More in: Education
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads

Views

Total Views
1,502
On Slideshare
0
From Embeds
0
Number of Embeds
0

Actions

Shares
Downloads
39
Comments
0
Likes
0

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide

Transcript

  • 1. Dialogue Earth Mining Social Media Tom Masterman Institute on the Environment 2010 Communicators Forum Conference University of Minnesota 05.13.10
  • 2. Agenda • My background • Importance of social media data • Types of social media mining • Dialogue Earth’s work • What you can do today 2
  • 3. My Background • Strategy, content creation & analytics • 10+ years in media • Writer/producer at Boston TV station • Ran analytics at Internet Broadcasting • Associate Director, Dialogue Earth at IonE • Communication degree, M.B.A. 3
  • 4. Importance of Social Media Data
  • 5. Importance of Social Media Data • My broad definition of “social media”: Anywhere on the Internet where users participate in content creation and sharing, including... • Blogs • Microblogs (Twitter) • Social networks (Facebook) • Forums • Video sharing (YouTube) • Image sharing (Flickr) • Social bookmarking (Digg) • Mainstream media (comments on NYTimes) 5
  • 6. Importance of Social Media Data • Most important application: Data helps you determine if social media is important to your business, relative to other... • Marketing channels • PR efforts • Revenue streams • Expenses • Ways your employees spend time • Direct competitors • Indirect competitors 6
  • 7. Dramatic Social Media Growth Montage Mashable.com, 02.04.10 7 http://greatlaker.files.wordpress.com/2008/05/surprised_baby_2.jpg
  • 8. Insecure Social Media ROI Montage Econsultancy.com, 02.23.10 Mashable.com, 10.13.09 8 http://sabrechaser.files.wordpress.com/2009/04/nervous-woman.jpg
  • 9. Importance of Social Media Data • Things I’m confident about: • Social media is likely to grow in importance • Meaningful applications of social media data: • Identifying opinion leaders • Insights into what your customers know and feel • Awareness of potential PR problems 9
  • 10. Types of Social Media Mining
  • 11. Types of Social Media Mining • Quantitative. Engagement, actions, CONVERSION. • Qualitative. Attitudes, intentions, SENTIMENT. 11
  • 12. Quantitative Social Media Mining • Traffic. Visitors, visits, views, time spent, visits/visitor, page views/visit... • Participation. Followers, fans, friends, posts, comments, likes, shares, sign-ups... 12
  • 13. Quantitative Social Media Mining • Free traffic tools: Google Analytics,Yahoo! Analytics • Paid traffic tools: WebTrends, Omniture • Social media participation reporting: ow.ly / bit.ly, Facebook Insights,YouTube 13
  • 14. Quantitative Social Media Mining Google Analytics • Traffic Sources report to see which social media sites are driving visits • Top Content report to see which specific elements are converting 14
  • 15. Quantitative Social Media Mining ow.ly, bit.ly reporting • Clicks on your posted (url-shortened) links • Location of the conversion • Content clicked on 15
  • 16. Quantitative Social Media Mining Facebook Insights • Interactions • Comments • Wall posts • Likes • Demographics • Geographical http://www.insidefacebook.com/wp-content/uploads/2009/05/newinsights1.png 16
  • 17. Quantitative Social Media Mining YouTube Reporting • Views, comments, favorites, ratings • Event timeline 17
  • 18. Qualitative Social Media Mining • Site identification. Where are people talking about your brand? • Author influence. Who is talking about you and what is their social clout? • Text analysis. What is being said and is it positive or negative? 18
  • 19. Qualitative Social Media Mining • Focused on sites: uberVU, Backtype • Focused on influence: Klout, Twitter Grader, Technorati • Focused on sentiment: Tweetfeel, Twitter Sentiment • A little bit of everything: Social Mention 19
  • 20. Qualitative Social Media Mining uberVU • Learn top sites where topic & brand keywords are mentioned • Browse individual conversations 20
  • 21. Qualitative Social Media Mining Backtype • Similar to uberVU, browse individual conversations (in news sites, blogs, Twitter) where your topic and brand keywords appear 21
  • 22. Qualitative Social Media Mining Klout • Based on a keyword search, learn the most influential Twitter users • Detailed stats about the content they produce and who they influence 22
  • 23. Qualitative Social Media Mining Twitter Grader • Search Twitter users to determine the most influential people • Timeline of their followers and following counts 23
  • 24. Qualitative Social Media Mining Technorati • Based on a keyword search, learn who the blogging “authorities” are on a given topic or brand 24
  • 25. Qualitative Social Media Mining Tweetfeel • Get the sentiment (positive or negative) of tweets containing keywords 25
  • 26. Qualitative Social Media Mining Twitter Sentiment • Similar to Tweetfeel, get the sentiment (positive or negative) of tweets containing keywords • Look at trends over time • Reclassify tweets • Save queries to track trends 26
  • 27. Qualitative Social Media Mining Social Mention • Learn top sites where topic & brand keywords are mentioned • Browse individual conversations • Learn the top users talking about various keywords 27
  • 28. Dialogue Earth’s Work
  • 29. Dialogue Earth’s Work • What is Dialogue Earth? • Program at the U of M’s Institute on the Environment • Started in October 2009 • Team includes experts in science, Web technology and social media, and U of M research assistants 29
  • 30. Dialogue Earth’s Work • Background: Polarized debate over environmental topics, such as the existence, and importance of global warming Telegraph.co.uk, 11.20.09 Gore Beck Time.com, 12.17.09 Time.com, 02.10.10 30 Gore image: http://thebreakthrough.org/blog/2008/03/20/enviro%20gore.jpg Beck image: http://radicalrecruit.files.wordpress.com/2009/11/glenn_beck.jpg
  • 31. Dialogue Earth’s Work • Mission: Increase shared understanding on topics related to the environment • Process: • Analyze social media dialogue current focus • Create relevant, trustworthy content • Establish a meaningful Web presence 31
  • 32. Dialogue Earth’s Work • Analyzing social media dialogue • Mine sites for topics related to the environment • Model the drivers of dialogue • Develop an accurate sentiment analysis engine • Produce summaries and visualizations 32
  • 33. Dialogue Earth’s Work • Mining sites for environmental topics • Extracting data via free APIs • Focused on Twitter Search API • Previous work done with NYTimes API • Investigating APIs for competing analytics tools 33
  • 34. Dialogue Earth’s Work Twitter Search API • Filtered on “global warming” and “climate change” • User data includes: • Followers, following • Geo-location 34
  • 35. Dialogue Earth’s Work • Mining sites for environmental topics • Using Yahoo! Pipes and Google News to create a custom news aggregator 35
  • 36. Dialogue Earth’s Work • Modeling the drivers of dialogue News Media Coverage Social Media Activity Personal Experiences Social media image: http://www.webtricksbox.com/wp-content/uploads/socialmedia03.jpg 36 Blizzard image: http://media.mlive.com/grpress/news_impact/photo/blizzard-holland1jpg-5d978a3343bd6797_large.jpg Gas pump image: http://www.musclecarclub.com/petrol/free_gasoline_prices.jpg
  • 37. Dialogue Earth’s Work • Modeling the drivers of dialogue • Analysis: How correlated are global warming tweet volume and severe weather? 37
  • 38. Dialogue Earth’s Work • Developing an accurate sentiment analysis engine • Sentiment: attitude, opinion, emotion • Text analysis and social media mining disciplines are merging • Pricey and customizable: Radian 6, Clarabridge • Free, but limited: Social Mention, uberVU • Free and Twitter-focused: Twitter Sentiment, Tweetfeel 38
  • 39. Dialogue Earth’s Work • Developing an accurate sentiment analysis engine • There is tremendous potential for new, powerful marketing intelligence • Plus, it’s really cool data! Facebook Gross National Happiness 39
  • 40. Dialogue Earth’s Work • Developing an accurate sentiment analysis engine • Focusing on Twitter sentiment, there are two main issues we are addressing: • Accuracy of polarity (+/-) coding • Moving beyond +/- to answer specific questions 40
  • 41. Dialogue Earth’s Work • Some issues coding +/- for a tweet • Sarcasm: • News headlines: • Irrelevance: 41
  • 42. Dialogue Earth’s Work • Our approach to improving +/- sentiment • Combine a best-of-breed computer algorithm with a dedicated social media research team • Partnered with Twitter Sentiment creators • Dialogue Earth research team will: • Provide nuanced data for comparison • Highlight promising, new directions for the algorithm 42
  • 43. Dialogue Earth’s Work • Moving beyond +/-, to answer questions • Business leaders, marketers and researchers want to answer specific questions: • Do consumers like my new product features? • What do potential voters think of my candidate? • Will moviegoers recommend my movie to friends? 43
  • 44. Dialogue Earth’s Work • Moving beyond +/-, to answer questions Harris Interactive • Conducted traditional online survey of 2,500 adults • PLUS, teamed up with Clarabridge to analyze 240,000 blog and Twitter text items 44
  • 45. Dialogue Earth’s Work • Moving beyond +/-, to answer questions • Current work: Analyzing tweets to infer answers to questions about environmental topics • Focusing on levers contributing to public concern about issues like global warming: • Does global warming exist? • Will it have negative impacts? • Are we accountable? • Are there effective solutions? 45
  • 46. Dialogue Earth’s Work • Early results • Eight rounds of coding • First iteration of a codebook • Increased intercoder reliability scores • Summary analysis on thousands of tweets 46
  • 47. Dialogue Earth’s Work • Producing summaries and visualizations • Future Work: Create white papers, data tables, graphs and multimedia to produce a “Pulse of the Planet” Drought Severity Twitter Maps NOAA / Weather Channel Bing Maps Time Lapsed Tweet Animation 47 blprnt.com
  • 48. What You Can Do Today
  • 49. What You Can Do Today 1. Install Google Analytics 2. Search the buzz for your industry at Social Mention 3. Determine your influencers at Klout 4. Search the sentiment of your brand at Twitter Sentiment 5. Set benchmarks based on your competition and your specific ROI needs 49
  • 50. Questions? Tom Masterman Institute on the Environment E-mail: maste078@umn.edu Twitter: @tommasterman 2010 Communicators Forum Conference University of Minnesota 05.13.10