Fig 1.
Overall process framework.
Fig 2.
Model structure.
Table 1.
Financial and textual indicators.
Fig 3.
Parameter Sensitivity Analysis.
Table 2.
Arameter Settings.
Table 3.
Comparison of Results from Traditional Machine Learning Methods.
Fig 4.
Radar Chart of the Comparison of the Traditional Machine Learning Method Results.
Table 4.
T-test results(Values in bold indicate significant differences at 90% confidence).
Table 5.
Comparison of Text Feature Extraction between Longformer and BERT.
Fig 5.
Radar Chart Comparing Longformer and BERT Text Feature Extraction.
Table 6.
Prediction performance of single-period and multiperiod annual report texts.
Fig 6.
Prediction performance of single-period and multiperiod annual report texts.
Table 7.
Comparison of the prediction results of the BiLSTM and TextCNN single models.
Fig 7.
Prediction Results of the BiLSTM and TextCNN Single Models – Radar Chart.
Table 8.
Information Gain Effects of Multi-Period Text and Multi-Period Financial Data.
Fig 8.
Information Gain Effect of Multi-Period Text and Multi-Period Financial Data.