Predicting Persuasiveness of Online Reviewers
In this project we intendeded to predict the degree of persuasiveness of a movie reviewer (either highly persuasive or non-persuasive) by using Multimodal Signal Processing techniques (techniques to combine audio, visual or textual representations), aided with appropriate Machine Learning tools (various feature selection and classification algorithms). We simultaneously looked at visual, acoustic and linguistic features individually as well as by merging them to do the prediction. The above figure shows an overview of the recognition pipeline.
The project was conducted in a team of three. The results were published in several top conferences/workshops.
Publications:
Journals:
S. Park, H. S. Shim, M. Chatterjee, K. Sagae, L. P. Morency, "Multimodal Analysis and Prediction of Persuasiveness in Online Social Multimedia", ACM Transactions on Interactive Intelligent Systems, (ACM TiiS) 2016.
Conferences:
H. S. Shim, M. Chatterjee*, S. Park*, S. Scherer, K. Sagae, L. P. Morency, "Acoustic and Paraverbal Indicators of Persuasiveness in Social Multimedia”, IEEE Int’l Conf. on Acoustics Speech and Signal Processing 2015 (IEEE ICASSP 2015) (Oral).
S. Park, M. Chatterjee*, H. S. Shim*, K. Sagae, L. P. Morency, "Computational Analysis of Persuasiveness in Social Multimedia: A Novel Dataset and Multimodal Prediction Approach”, ACM Int'l Conf. on Multimodal Interfaces 2014 (ACM ICMI 2014) (Oral).
Workshops:
M. Chatterjee, S. Park*, H. S. Shim*, K. Sagae, L. P. Morency, "Verbal Behaviors and Persuasiveness in Online Multimedia Content”, Workshop on SocialNLP at COLING 2014 (COLING-SocialNLP 2014) (Oral).
[*- indicates an equal contribution]
Project Funding Support:
We are extremely grateful to the US NSF for funding this research project.