Induction of a sentiment dictionary for financial analyst communication: a data-driven approach balancing machine learning and human intuition

While sentiment dictionaries are easy to apply and provide reproducible results, they often exhibit inferior classification performance compared machine learning approaches trained for specific application domains. Nevertheless, both typically require manual data analysis. This paper develops a domain-specific dictionary using regularised linear models drawing from textual reports of financial analysts. The first evaluation step demonstrates that the developed analyst can explain cumulative abnormal stock returns related earnings events more accurately other finance-related classifiers. In second step, manually annotated sentiment. is accurate than dictionary-based approaches, although it cannot compete with pre-trained deep classifier. we show proposed approach suited texts analysts, be applied use cases. realises context specificity while reducing extensive