[Webinar] The Black Box of Sentiment Analysis: What's in it, and how to do it better
Webinar presented by Normand Péladeau on May 19, 2021
Sentiment analysis, or the analysis of opinions expressed by Internet users is of increasing interest to academic researchers as well as companies. With the growing popularity of this approach, commercial offers have multiplied very quickly. The often spectacular (and sometimes abusive) promises of these ready-to-use solutions reflect an expectation expressed by the user, but do not guarantee quality.
The controversy fuelled around this "black box" approach that aims to deliberately hide any internal parameterization deserves a careful examination to better specify what it is reasonable to expect from sentiment analysis and opinion mining methods and how to improve their performance.
In this webinar, we will insist on the importance of adapting the analysis tool to the application domains, but also to the context of the interaction in which it occurs. A benchmark test involving a comparison of open-source lexicons, commercial solutions as well as our own WordStat sentiment dictionary will be used to illustrate current issues with all those tools as well as the required efforts needed to develop better sentiment measurements.
Normand Péladeau is the president and CEO of Provalis Research, a software company based in Montreal. Normand Péladeau has a doctorate degree in psychology and more than 35 years of experience as a social science researcher and as a consultant in research methodology for large corporations, governmental agencies, and international organization. Dr. Péladeau has trained thousands of people in text analysis techniques in a wide range of applications, such as business intelligence, media analysis, market research, urban planning, aviation safety, survey research, and international crime analysis.