| Summary: | 碩士 === 國立成功大學 === 會計學系 === 107 === In recent years, the development of machine learning and deep learning have been remarkably advanced due to the rapid growth of computer hardware. In addition, many studies have shown that deep learning techniques have achieved good results in the theme of Natural Language Processing. In this study, we apply the deep learning-based textual analysis to convert texts into sentiment scores, which are then used measure CEO's sentiment to predict over-investment. By doing so, we can not only quantify texts into meaningful numbers but also make decision based on the analyzed results. We found that the sentiment scores are positively related to over-investment. Negative sentiment tends to have a greater impact on over-investment than positive sentiment. Finally, when multiplied by arousal, valence would show more explanatory power.
|
|---|
