Abstract

Artificial intelligence (AI) technologies offer promising opportunities for detecting market movements from news and textual data. However, many organizations fail to create or meet any data quality standards that are key to AI development. In this research, we developed and applied data-centric neural network (NN) language models to extracting business intelligence (BI) factors automatically from textual news articles of high-tech companies. Our methodology iteratively improves the datasets for use in building NN language models. Experimental results confirm that our approach helped to increase the predictive performance across different BI categories. The approach makes the NN language models more transparent to managers, BI specialists, and users through the iterative refinement and query search process. The research produces new information systems artifacts in the form of new methodology, new NN models, and new empirical findings of using the models to extract BI from textual news.

Suppress Author Email on Cover Page

1

Share

COinS