What Can We Do With Words?

What Can We Do with Words?

Mini-conference with Nello Cristianini and Roberto Franzosi

Friday, February 22, 2013

2:00 - 4:00 pm, White Hall room 110

Reception in the lobby following the event.


Automatic Discovery of Patterns in Media Content

Nello Cristianini, University of Bristol, UK

What can we learn about the world (and the media system) by analyzing millions of news articles or tweets? Media content analysis has historically been the domain of the social sciences, but recently we are witnessing a strong trend towards the automation of many tasks, paving the way for a new ¿ computational ¿ approach to social science and the humanities. In this talk, I will survey the results obtained over the past 5 years at the Intelligent Systems Laboratory of Bristol, in the area of automating the analysis of news media content. By combining techniques from machine translation, pattern recognition, statistical learning, information retrieval, I will analyze patterns connected to the past US Presidential Elections, to UK public opinion, and to EU cultural biases.


The Social Scientist, the Word, and the World: 

What the Words Tell Us about Italian Fascism(1919¿1922) and Georgia Lynchings (1875¿1930)

Roberto Franzosi, Emory University

The talk illustrates a quantitative social science approach to texts developed by the author, Quantitative Narrative Analysis (QNA). QNA relies on computer-assisted story grammars to analyze narrative, where a story grammar is the simple Subject-Verb-Object (SVO) structure. In narrative, Subjects are typically social actors ¿ individuals, groups, organizations ¿ Verbs are actions, and Objects are both social actors and physical and abstract objects. To each of the three SVO components one can add several attributes to capture the complexity of stories (e.g., name of an individual, number of actors in a group, time and space of action). The talk will illustrate the power of the approach using data collected by the author from newspapers on the rise of Italian fascism (1919¿1922) (300,000 SVOs) and Georgia lynchings (1875¿1930) (7,000 SVOs). It will show how narrative data lend themselves to cutting-edge tools of data visualization and analysis as network graphs and maps in Google Earth and other GIS software. It will also show how QNA data provide the basis for fascinating digital humanities projects.


Conference Co-sponsors:

The Hightower Fund | The Department of Sociology | The Department of Mathematics and Computer Science

Digital Scholarship Commons |  The Institute for Quantitative Theory and Methods | The Emory College Language Center

The Graduate School of Liberal Arts  |  The Fox Center for Humanistic Inquiry