Social  Web:  Where  are  the  Semantics?

Tutorial within the ESWC 2014

logo

Venue, Date

The tutorial has a full day length and it takes place in Anissaras, Hersonissou (Greece), the 25th May 2014

Programme

  1. Part  1:  Introduction
    Theory:
    • What  are  social  networking  sites  (SNS)?
    • What  are  the  different  types  of  SNS?
      • Business  vs.  Public  domain  SNS
    • SNS  used  in  this  tutorial:  
      • Twitter
      • Facebook
      • IBM  Connections
      • SAP  Community  Network
    • Use  case  scenarios  used  in  this  tutorial
      • Education:  gathering  feedback  about  OU  courses  in  Facebook  open groups
      • Government:  analysing  Twitter  to  monitor  policy  related  topics
      • Knowledge  exchange  and  product  support:  maintaining  the  health of  business  SNS   (SAP  &  IBM)
    Hands  on  session:  Data  extraction:  scripts  will  be  provided  to  gather  data  from Twitter  and  Facebook.
  2. Part  2:  Using  semantics  to  represents  data  from  Social  Networking  Sites

  3. Theory:
    • Using  semantics  to  represent  data  from  SNS:  motivation  
      • Data  unification
      • Portable  analysis  models
      • The  Social  Semantic  Web  Stack
        • Representing  content
        • Representing  people
        • Representing  relations
      • SIOC:  the  Semantic  Interlinked  Online  Communities  model
        • Applying  and  extending  SIOC
    • Part  3:  Using  semantics  to  understand  social  media  conversations
      Theory:
      • Topic  analysis  using  explicit  semantics:  exploiting  entities  and  topics extracted  via  TextRazor  to  understand  the  feedback  of  Open  University students  in  Facebook  open  groups
      • Topic  analysis  using  implicit  semantics:   elicitation  of  latent  topics  from Twitter  data  in  the  entertainment  domain
      • Sentiment  analysis  using  explicit  semantics:  using  semantic  concepts  to reduce  Twitter  data  sparsity  and  enhance  the  accuracy  of  sentiment classifiers  
      • Emotion  analysis  using  a  rule-­based  system.
      • Hands  on  session:  Data  annotation  and  analysis
        • Using  DBPedia  to  extract  entities  from  Twitter  data
        • Using   Latent  Dirichlet  Allocation  to  extract  topics  from  Twitter  data
        • Topic  Analysis,  which  key  topics  emerge  from  the  data?
        Hands  on  session:  Emotion  Analysis
        • Extracting  emotions  from  Twitter  data
    • Part  4:  Using  semantics  to  understand  user  behaviour
      Theory:
      • User  roles  in  online  communities:  who  are  the  leaders,  the  celebrities,  the daily  contributors  and  the  ignored  users?
      • The  Open  University  Behaviour  Analysis  Ontology
      • Monitoring  behaviour  over  time  using  SPIN  rules

Abstract

Social  networks  generate  major  economic  value  and  form  pivotal  parts  of commercial  services,  advertisement,  entertainment,  etc.  Multiple  tools  and technologies  have  emerged  in  the  last  few  years  that  aim  to  monitor  and  analyse data  from  these  networks  in  order  to  exploit  their  value.  This  tutorial  aims  to provide  a  comprehensive  overview  of  where  and  how  semantic  information  has been  used  to  represent  and  analyse  social  networking  data.  The  presented research  and  development  covers  different  use  cases  and  applications  including education,  e-­‐government  and  business  cooperation.  With  the  help  of  hands  on session  attendees  will  make  use  of  some  of  these  technologies  to  collect  and analyse  data  from  popular  social  networking  sites,  such  as  Facebook  or  Twitter. Semantic  technologies  will  be  applied  to  perform  topic  and  sentiment  analysis over  these  data.

Contents

The  explosive  growth  of  social  networking  sites  continues  in  all  areas  of  society, and  their  use  is  now  widespread  in  social,  business,  scientific  and  public  service domains,  enabling  community  members  to  share  ideas,  knowledge  and  opinions. Beyond  high  profile  public  social  networking  sites,  e.g.  Facebook  or  Twitter, social  networks  now  generate  major  economic  value  to  businesses  and  can  form pivotal  parts  of  corporate  expertise  management,  corporate  marketing,  product support  and  targeted  advertising.  A  recent  report  by  McKinsey   1 estimates  that between  900  and  1,300  billion  dollars  in  annual  value  could  be  obtained  if  social media  is  fully  exploited.

With  the  purpose  of  exploiting  the  value  of  these  networks  multiple  technologies and  analysis  methods  have  been  developed  in  the  last  few  years.  In  this  tutorial we  focus  on  those  methods  that  exploit  semantic  information,  not  only  to  better model  that  data  from  these  networks,  but  also  to  provide  more  comprehensive and  flexible  analyses.  The  studied  semantic  analysis  methods  make  use  of  both, implicit  and  explicit  semantic  information.  While  explicit  semantic  analysis methods  make  use  of  ontologies,  knowledge  bases  and  linked  data,  implicit semantic  analysis  methods  exploit  natural  language  features  to  identify  the entities  and  relations  emerging  from  the  data.  Once  semantic  information  is identified  and  selected  it  is  used  to  enhance  the  flexibility  and  performance  of  a variety  of  analyses.

In  this  tutorial  we  will  explore  how  semantic  information,  both  implicit  and explicit,  has  been  used  in  the  literature  to  model  and  integrate  data  across communities,   as  well  as  to  analyse  this  data.  Some  of  these  analyses  include understanding  the  users’  behaviour  and  their  influence  within  the  communities, detecting  users’  sentiment  and  emotions  and  identifying  the  topics  that  emerge from  the  users’  conversations.

Examples  of  the  application  of  these  analyses  over  different  practical  uses  cases will  be  presented.  These  uses  cases  include  (i)  the  analysis  of  Facebook  data  to identify  the  issues  and  concerns  of  Open  University  (OU)  students  regarding particular  OU  courses,  (ii)  the  analysis  of  Twitter  data  to  identify  the  key  topics and  opinions  of  citizens  regarding  policy  related  topics,  (iii)  the  elicitation  of latent  topics  from  Twitter  data  in  the  entertainment  domain,  and  (iv)  the analysis  of  private  intranets,  such  as  IBM  connections  and  SAP  Community Network,  to  monitor  the  behaviour  of  their  users.  

Main related papers

  • Fernandez,  M.,  Alani  H.,  and  Brown  S.  OU  Social:  Reaching  Students  in  Social Media.  ISWC  2013.  Sydney,  Australia,  October  2013
  • Rowe,  M.,  Fernandez,  M.,  Angeletou,  S.  and  Alani,  H.  (2012)  Community  Analysis through  Semantic  Rules  and  Role  Composition  Derivation,  Journal  of  Web Semantics
  • Rowe,  M.,  Fernandez,  M.  and  Alani,  H.  (2013)  Modelling  and  Analysing  User Behaviour  in  Online  Communities,  IEEE  Special  Technical  Community  on  Social Networks  e-­letter,  1,  (2)
  • Saif,  H.,  He,  Y.  and  Alani,  H.  (2012)  Semantic  Sentiment  Analysis  of  Twitter. International  Semantic  Web  Conference,  Boston,  US
  • García-­Silva,  A.  Rodríguez-­Doncel  V.  and  Corcho,  O.  Semantic  Characterization  of Tweets  Using  Topic  Models:  A  Use  Case  in  the  Entertainment  Domain. International  Journal  on  Semantic  Web  and  Information  Systems  (In  press)
  • Wandhöfer,  Timo;  Taylor,  Steve;  Alani,  Harith;  Joshi,  Somya;  Sizov,  Sergej; Walland,  Paul;  Thamm,  Mark;  Bleier,  Arnim;  Mutschke,  Peter  (2012):  Engaging Politicians  with  Citizens  on  Social  Networking  Sites:  The  WeGov  Toolbox.  In: International  Journal  of  E  lectronic  Government  Research,  July-­September  2012, Vol.  8,  No.  3,  p.  22-­43.
  • Bontcheva  K.,  Rout,  D.  2012.  Making  Sense  of  Social  Media  Streams  through semantics:  a  Survey.  Semantic  Web  Journal

Required Knowledge

 Basic  knowledge  in  Semantic  Web  and  Social  Networks may  allow  better  following  the  tutorial  and  gaining  more  benefits  from  it.  A Twitter  and  Facebook  accounts  are  required  for  the  first  hands-­on  session.

Presenters

Photo used under Creative Commons CC-BY license from youasamachine