Mood-Based Micro-Blogging in Distributed Software Engineering

Talk presented at MSR 2013, the 10th Working Conference on Mining Software Repositories.

Paper: http://aspic.nl/publications/DullemondGamerenStoreyDeursenMSR2013.pdf

Transcript

1. Mood-­‐Based  Micro-­‐Blogging   Kevin  Dullemond,    Ben  van  Gameren,     Margaret-­‐Anne  Storey,  Arie  van  Deursen   Mining  So;ware  Repositories  (MSR),  San  Francisco,  May  2013   1  
2. Micro-­‐Blogs   •  “Short  messages  people  use  to  provide  updates   on  their  ac5vi5es,  observa5ons  and  interes5ng   content,  directly  or  indirectly  to  others”   •  Use  in  so;ware  engineering:   – Support  awareness?   – Support  distributed  working?   Ehrlich  et  al:   “Microblogging  inside   and  outside  the   workplace”,  ICWSM,  2010   2  
3. IHomer  Creates  WeHomer   WeHomer  @  iHomer:   Miro-­‐Blogging  with  Moods     >L      L      K      J      >J   3  
4. Today’s  MSR  keynote   4  
5. 5   I’m  going  to  be  a  daddy!    :-­‐D   Will  be  partly  available  today;  Son  is  sick  at  home  >L   Working  on  our  Linux  servers  today  –     some  services  may  experience  outages  J   On  my  way  to  customer  X  for  first  live  test  :-­‐D  
6. 1  year  usage  data;   2500  messages   Mood  is  actually   indicated   6  
7. Approach   •  Try  to  understand:   – Role  of  micro-­‐blogging  in   distributed  work  @  home   – Role  of  “mood”   •  Qualitadve  analysis   – Coding  2500  posts     from  20  users   – Interviews  with  5  users   Content Information about a person Health Sentiment Personal Experience Information about technology Technical Knowledge Information about task articulation work Work Planning Work Assignment Supplies Non-Technical Infrastructure Technical Infrastructure Intern Technical Infrastructure Extern Information about customer relations Relation Project Commissioning Information about entrepreneurial tasks Prospects Company Meeting Applicant Invoicing 7  
8. “Nature”:     Most  Posts  Are  Posidve   8   Manual  sendment  analysis.   Largely  in  line  with  indicated  “happiness”  
9. “Form”:     9%  of  Posts  is  a  Quesdon   9   Ehrlich  et  al  BlueTwit  analysis:  13%  quesdons    
10. “Intendon”:     Personal,  Work,  Coordinadon   10   “Task  ardculadon”  work   Home  /  work  balance  
11. “Content”:     Intra-­‐  And  Inter-­‐Team  Topics   •  Intra-­‐team  (4  teams,  2-­‐7  members)   – Directed  messages:  more  within  team   – Planning  and  coordinadon   •  Full  organizadon  (20  people):   – New  employees,    new  prospects,  new  business   direcdons,  social  events   – Planning  and  coordinadon   •  “Glue”  between  people  working  at  home.   11  
12. The  Tuesday  Effect   12   “There  is  no  halfsies  in  a  distributed  team  …   If  even  one  person  of  the  team  is  remote,  very  single  person  has  to   start  communica5ng  on  line.”  –  Fullerton  /  StackExchange  
13. Discussion   •  Social  interacdons  affect  so;w.  development   – Mining  research  must  be  aware  of  them!   – To  what  extent  should  we  mine  them?   •  Mood  indicators?   – Issues,  feature  requests,  commits,  pull  requests?   •  Sendment  analysis?  –  Anger  versus  Joy   13  
14. Conclusions   •  Micro-­‐blogging  with  mood  indicators:   – Share  knowledge  and  emo5ons  like     you  are  co-­‐located   – Create  and  sustain  a  feeling  of  connectedness   (conform  Zhao  et  al,  ICWSM  2009)   – Beyond  individual  projects   •  All  or  nothing  proposidon  (Tuesday  effect)   14