Verbalizing time series data from a macroscopic viewpoint

Most data observed in our lives are time series data. So, we need a method to be able to access and easily utilize the data. As a representative method to realize it, visualization is widely used. On the other hand, as we see that there are many texts, e.g., newspaper articles reporting the trends on stock prices, foreign exchange rates, weather conditions, etc. Explaining time-series data with words is also widely used. With this background, in this study, we focus on explaining time series data with words and propose a method to verbalize time series data from a macroscopic viewpoint -which is that we aim to verbalize time series data by visually recognizing the shapes of a line chart of time series data. We apply our proposed method to verbalize stock price time series data and evaluate the results of generated texts.