Tagged / Sentiment Analysis

Presenting Studies on LLMs Reasoning Capabilities in Sentiment Analysis of Mass-Media texts at NLPSummit-2024

 

As a part of reseach studies in Natural Langauge Processing (NLP) field, this year I am delighted to present the most recent advances of Generative AI in it at NLPSummit-2024. The 5’th summit represents a free online conference September 24-26, hosted by JohnSnowLabs. The conference is dedicated to showcase the best practices, real-world case studies and challanges in Generative AI for Natural Language Processing.

By joining to my talk you become aware of how Large Language Models (LLMs) could be applied for retrieving implicit information from non-structured texts. Sentiment Analysis represent one of such problems, and as a task aimed at extraction of the hidden opinion of the author towards objects mentioned in text. We start by discovering reasoning capabilities of the most popular Large Language Models (ChatGPT, Mistral, Gemma, Microsoft-Phi, and more) out-of-the-box to show their limitations in retrieving authors opinion from Mass-media texts. To address the existed limitations in models reasoning capabilities 🧠 , we cover Chain-of-Thought technique and explore the way of its proper adaptiation in Sentiment Analysis. It is worth to note that the techniques, to be covered, could be distributed and adapted in the other domains that go beyond Mass-media. Such domains include but are not limited to: medical (adverse drug reaction), literature (fictional chatbot development), conversational (emotion extaction / empathy mapping).

These advances were achieved while at Centre for Applied Creative Technologies CfACTs+ by working on “Marking Medical Image Reports Automatically with Natural Language Processing (NLP-MMI)” project.

The keylinks realted to the event and presentation in particular, are as follows:

📍 Event page: https://www.nlpsummit.org/nlp-summit-2024/
When: 24-26 September 2024 (Online)
 Project: https://github.com/nicolay-r/Reasoning-for-Sentiment-Analysis-Framework

Dr. Nicolay Rusnachenko
Research Fellow at Centre For Applied Creative Technologies PLUS (CFACT+)
Bournemouth University