Visualizing Word Categorization of Corporate Annual Reports Using Self- Organizing Maps

An increasing proportion of corporate information takes the form of unstructured or semi-structured text. Annual reports are one of the most important external documents that discuss corporate performance and managerial priorities. This paper is aimed at visualizing textual information in corporate annual reports. Several word categorization schemes are used to extract the tone of text. Self-organizing maps are employed to map the word categories. The results indicate that the annual reports of U.S. companies differ in terms of the tone emphasized. In addition, we show that the tone reflects both current financial performance and future change in the performance.