Detection of Bias in Media Outlets with Statistical Learning Methods - Text Mining: Classification, Clustering, and Applications - page 26

Chapter 2

Detection of Bias in Media Outlets
with Statistical Learning Methods

Blaz Fortuna, Carolina Galleguillos, and Nello Cristianini

2.1 Introduction ............................................................. 27

2.2 Overview of the Experiments ............................................ 29

2.3 Data Collection and Preparation ........................................ 30

2.4 News Outlet Identification .............................................. 35

2.5 Topic-Wise Comparison of Term Bias ................................... 38

2.6 News Outlets Map ....................................................... 40

2.7 Related Work ............................................................ 44

...............................................................

2.8 Conclusion 45

2.9 Appendix A: Support Vector Machines ................................. 48

2.10 Appendix B: Bag of Words and Vector Space Models .................. 48

2.11 Appendix C: Kernel Canonical Correlation Analysis ................... 49

2.12 Appendix D: Multidimensional Scaling ................................. 50

2.1 Introduction
The international media system plays a crucial role both in reflecting pub-
lic opinion and events, and in shaping them. Understanding the workings of
this complex system is of crucial importance for society, business and democ-
racy, and is one of the research foci of media analysts, anthropologists, and
increasingly also of computer scientists.
The traditional way in which the contents of the media system are analyzed
has been by human analysts reading each news item, filling a pre-designed
questionnaire in the process. This phase of the analysis is termed 'coding.'
The questions addressed often have some subjective component, such as the
detection of opinions, emphasis, or perspectives. Great care is paid in the
design phase of the questionnaire, to make the coding phase as objective as
possible. One key limitation of this approach, of course, is that it is very labor
intensive, and hence it cannot be scaled up to a global / constant monitoring
of the entire system. Indeed, the systems perspective of this analysis cannot
be pursued, as long as only small portions of the system can be analyzed at
each given time.