Financial sentiment analysis in the machine learning era

What is it about?

An important question to many practitioners and researchers is: "how should I build a neuro-symbolic AI model?" Using financial sentiment analysis as an example, where both general language intelligence and expert knowledge are needed to excel, this research shows that "parallel knowledge integration" is a promising method.

Featured Image

Photo by Firmbee.com on Unsplash

Photo by Firmbee.com on Unsplash

Why is it important?

In the machine learning era, knowledge bases are much overlooked. However, for domain-specific tasks, such as financial sentiment analysis, combining trained based models and existing knowledge sources accumulated over decades is currently still the best practice. This research shows how to build such hybrid systems, and provides information on which step (in the whole neural network architecture) to integrate the knowledge is more effective.