Thursday, February 11, 2021 at 4:00pm to 5:00pm
Virtual EventAn Automated Pipeline for Narrative Extraction and Conspiracy Theory Detection
Join Shadi Shahsavari and of UCLA Samueli School of Engineering and Applied Science as they detail their research on the semantic structure of internet spread conspiracy theories.
Summary
Given the rise of social media and the clear impact that the circulation of stories and story bits are having on real-world behavior, it is of interest to us to characterize the semantic structure of conspiracy theories. Such a characterization, if successful and mature, may provide a more holistic and well-connected understanding (and visualization) of these theories and in the process, may better inform the end user of disparate contexts and entities that drive the continued proliferation of these beliefs.
The resulting model is a pipeline of interlocking machine learning (ML) methods that allow us to jointly estimate the parameters of the underlying narrative framework graph. Given a corpus of text, this pipeline generates a complex network of the super/subnodes and their interrelationships. Such a modeling effort at the very least, adds semantic structure to multi-context conspiracy theories circulating the internet and in time, may yield a predictive model for theory generation. In this talk, we will shed light on the inner workings on the pipeline and interesting results.
Dial-In Information
Register in advance for this meeting:
https://uri-edu.zoom.us/meeting/register/tJMkf-6oqTMtG9UOFsdGVksG0DsYsPwP0dGX
After registering, you will receive a confirmation email containing information about joining the meeting.
Free. Please register.
#ai, ml, artificial intelligence, machine learning, politics, current, social media, data science
No recent activity