Modeling Distortion In News Aggregation Page

1
Modeling Distortion In News Aggregation Page

• Research Report

2008 07 16 Soyoon Won Data Mining Lab. SNU.
2
Contents

• Background
• Formulation
• Related Work
• Expected Problems
• References

3
Background

• Web portals have media power as gatekeepers of
  news articles

articles
Editors choice
4
Background

• Disputes about intentional distortion
• 2007 Hostage Crisis in Afghanistan
• 2007 Presidential Election
• Typical aspects of distortion
• Hiding specific topics
• Exaggerating specific topics

5
Formulation

• Definition of Distortion
• Article space

topic 1
Event 1
Event 2
Event 3
6
Formulation

• Definition of Distortion
• True distribution Q(x)
• All articles from all media sources
• Subset distribution P(x)
• Articles on the aggregation page is a subset of
  whole article space
• Measure of distortion
• KullbackLeibler (KL) divergence Distance
  between two distributions
• How the subset distribution reflect the true
  distribution

7
Formulation

• Main sub-problems
• Topic extraction
• Event extraction
• Attitude extraction

8
Related work

• Topic Mining
• By article level clustering 1
• Discovering events in news articles
• By sentence clustering 2
• Extracting opinions in documents
• By Sentence level classification using Naïve
  Bayes Classifier 3
• By Language analysis 4
• Opinion polarity analysis
• By Language analysis 3, 4

9
Expected Problems - Keep going or Not?

• Accuracy
• For each sub-problem, accuracy is under 90
  (5080)
• Portal regulation law has proposed in congress
  (2008-07-14)
• If this proposal is approved, web portal cannot
  service news aggregation pages anymore
• Identity as a thesis subject
• Data Mining? Natural Language Processing problem?

10
References

• 1 Seokkyung Chung and Dennis McLeod (2003).
  Dynamic Topic Mining from News Stream Data. In
  Proc. Of the CoopIS/DOA/ODBASE 2003
• 2 Martina Naughton, Nicholas Kushmerick and
  Joe Carthy (2006). Clustering Sentences for
  Discovering Events in News Articles. In Proc. of
  the 28th European Conference on Information
  Retrieval Research (ECIR 2006)
• 3 Hong Yu and Vasileios Hatzivassiloglou
  (2003). Towards Answering Opinion Questions
  Separating Facts from Opinions and Identifying
  the Polarity of Opinion Sentences. In Proc. of
  the 2003 Conference on Empirical methods in
  natural language processing(EMNLP 2003)
• 4 Jiahui Liu and Larry Birnbaum(2008).
  LocalSavvy Aggregating Local Points of View
  about News Issues. In Proc. of the first
  international workshop on Location and the web
  (LocWeb 2008)