Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Mining Questions Asked by Web Developers

395 views

Published on

MSR'14

Published in: Engineering
  • Be the first to comment

  • Be the first to like this

Mining Questions Asked by Web Developers

  1. 1. Mining Ques$ons Asked by Web Developers Kar$k Bajaj, Karthik Pa.abiraman, Ali Mesbah Electrical and Computer Engineering University of Bri$sh Columbia Vancouver, BC, Canada {kbajaj, karthikp, amesbah}@ece.ubc.ca MSR 2014
  2. 2. Problem Statement n Web applica$ons are becoming popular over $me n [Jazayeri et al., FOSE 2007] n Complex integra$on of JavaScript, CSS, and HTML n JavaScript code oen experiences errors n [Ocariza et al., ISSRE 2011] MSR 2014 2
  3. 3. Exis$ng Work n Sta$c Checkers n Dynamic Checkers n Console Messages n Bug Reports MSR 2014 3
  4. 4. Exis$ng Work n Sta$c Checkers n Dynamic Checkers n Console Messages n Bug Reports No a.empt to understand developer’s misconcep$on MSR 2014 4
  5. 5. Goal n Understand the common challenges and/or misconcep$ons among web developers MSR 2014 5
  6. 6. Background n Stack Overflow n QA website for programmers n Started in 2008 n 4,125,638 ques$ons asked from Jan’09 to Dec’12 n 500,000+ ques$ons related to web development n Ques$ons directly asked/answered by developers n Followed by discussion in comments MSR 2014 6
  7. 7. Background n Stack Overflow n QA website for programmers n Started in 2008 n 4,125,638 ques$ons asked from Jan’09 to Dec’12 n 500,000+ ques$ons related to web development n Ques$ons directly asked/answered by developers n Followed by discussion in comments StackOverflow provides sufficient data for analysis MSR 2014 7
  8. 8. Straw man Approach n Categoriza$on based on tags a.ached to each ques$on n Can provide an overview of major topics of discussion MSR 2014 8
  9. 9. Straw man Approach n Categoriza$on based on tags a.ached to each ques$on n Can provide an overview of major topics of discussion How appropriate are tags? MSR 2014 9
  10. 10. Tags n What is known to developer while tagging? n Plamorm n Target Device n What is unknown to developer? n Possible ways to solve the problem n Correct topic of discussion MSR 2014 10
  11. 11. Example MSR 2014 11
  12. 12. Natural Language Analysis n Analyzing the text provided in the ques$ons and answers provided by developers MSR 2014 12 Data Collec$on Data Cleaning Data Processing
  13. 13. Datasets MSR 2014 13
  14. 14. RQ1: Categoriza$on of topics of discussion
  15. 15. ! #$
  16. 16. %
  17. 17. '
  18. 18. ('
  19. 19. $
  20. 20. MSR 2014 14
  21. 21. RQ1: Categoriza$on of topics of discussion
  22. 22. Cross Browser related discussions have gained maximum
  23. 23. ! #$
  24. 24. %
  25. 25. '
  26. 26. ('
  27. 27. $
  28. 28. a.en$on from web developers. MSR 2014 15
  29. 29. RQ2: Hot topics of discussion Technology Hot Topics JavaScript Document Structure, File Handling, Cross-­‐Browser, jQuery, DOM HTML5 Media, Browser Support, HTML5 Elements, Canvas API, Offline Web CSS CSS3, Fonts, JavaScript, Box-­‐Model, Layout MSR 2014 16
  30. 30. RQ2: Hot topics of discussion Technology Hot Topics JavaScript Document Structure, File Handling, Cross-­‐Browser, jQuery, DOM View counts provide a hint towards recurrent issues faced by HTML5 Media, Browser Support, HTML5 Elements, Canvas API, Offline Web web developers. CSS CSS3, Fonts, JavaScript, Box-­‐Model, Layout MSR 2014 17
  31. 31. RQ3: Temporal trends over $me
  32. 32. MSR 2014 18
  33. 33. RQ3: Temporal trends over $me Cross-­‐browser compa$bility issues have seen a sharp decline
  34. 34. in the recent past. Further, CSS3 and HTML5 discussions are gaining popularity. MSR 2014 19
  35. 35. RQ4: Prevalence of web in mobile development
  36. 36. MSR 2014 20
  37. 37. RQ4: Prevalence of web in mobile development
  38. 38. Discussions related to Mobile development are seeing an
  39. 39. increasing share of web technologies. MSR 2014 21
  40. 40. Popularity n Ques$ons that gained maximum a.en$on from the community n Post Score n Up votes n Down votes MSR 2014 22
  41. 41. Popularity n Ques$ons that gained maximum a.en$on from the community n Post Score n Up votes n Down votes Is post score accurate representa$on of user par$cipa$on? MSR 2014 23
  42. 42. StackOverflow Post Score n Missing details n How many users up voted or down voted the ques$on? n How many users tried to answer the ques$on? n Is there an accepted answer for this ques$on? n How many users are following the ques$on? n How many users are par$cipa$ng in the discussion? MSR 2014 24
  43. 43. StackOverflow Post Score n Missing details n How many users up voted or down voted the ques$on? n How many users tried to answer the ques$on? n Is there an accepted answer for this ques$on? n How many users are following the ques$on? n How many users are par$cipa$ng in the discussion? MSR 2014 We devised a new metric to rank Stack Overflow ques$ons based on these factors. 25
  44. 44. RQ5: Technical challenges MSR 2014 26
  45. 45. RQ5: Technical challenges Programmers get confused about new features, sugges$ng that the available API resources for these features is far from ideal. Maintaining web code, such as CSS, is complex without proper tool support. MSR 2014 27
  46. 46. Implica$ons n Finding 1, 3 (CategorizaFon and temporal trends) n Developers can shi their focus away from cross browser issues. n Finding 4 (Prevalence of mobile applicaFons) n Can guide tool developers to build be.er mobile development tools. n Finding 5 (Technical Challenges) n Can guide standardiza0on communi0es to focus on areas that need improvement. MSR 2014 28
  47. 47. Thank You!! Contribu$ons 1. Categorized web related discussions. 2. Iden$fied temporal trends. 3. Devised a metric to rank StackOverflow ques$ons. Analyzed Data: h.p://www.ece.ubc.ca/~kbajaj/so/data.zip MSR 2014 29

×