Contrastive Opinion Summarization Research Data

• This page has links to the data set and demo page for contrastive opinion summary.
Hyun Duk Kim and ChengXiang Zhai, Generating Comparative Summaries of Contradictory Opinions in Text, CIKM 2009
• Page created by Hyun Duk Kim and ChengXiang Zhai of the TIMan group at UIUC.

[Demo system]

• Demo2: Ooops! - Open Opinion Search engine using Contrastive opinion summary.
• Demo1: Contrastive opinion summary generation with user input.

[Data set]

• Data set used for the experiment of the 'Contrastive Opinoin Summarization'.[link]
• When use this data set, please use the following publication as a reference.
  Hyun Duk Kim and ChengXiang Zhai, Generating Comparative Summaries of Contradictory Opinions in Text, CIKM 2009
  
• Details about how this data set was made and used are on the publication.
• Contacts:
  Hyun Duk Kim (hkim277 at illinois.edu), [HomePage]
  ChengXiang Zhai (czhai at cs.uiuc.edu), [HomePage]
• Data File ('data' folder)
• Filename
  [ProductName]_[feature]
• Format
  [(polarity)] (sentence number) - (sentence)
  
  [1]: positive
  [0]: negative
  
  Ex.
  [1] 1 - the height adjustment knob is a bit flimsy, but it works well and provides good accuracy.
  [1] 2 - i noticed that the micro adjustment knob and collet are well made and work well too.
  [1] 3 - i don't have problems with the depth adjustment, the speed control, the collet ( i haven't tried the 1/4" sleeve yet) .
  [1] 4 - it is a great router for a table and with the depth adjustment knob it is easy to adjust depth prescisely.
  [1] 5 - the depth adjustment and b
  ...
• Human Label File
• Two human labels (cos_label1.txt, cos_label2.txt)
• Format
  For each data file, sentences are clustered into subtopcis.
  For each subtopic, sentence numbers maked in the data file are listed.
  
  ------[data file name]
  ---Pos
  [subtopic1] [positiveSentenceNumber1] [positiveSentenceNumber2] ...
  [subtopic2] [positiveSentenceNumber1] [positiveSentenceNumber2] ...
  ...
  ---Neg
  [subtopic1] [negativeSentenceNumber1] [negativeSentenceNumber2] ...
  [subtopic2] [negativeSentenceNumber1] [negativeSentenceNumber2] ...
  ...