File size: 3,712 Bytes
8117ca9 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 | #!/usr/bin/env python3
"""Visualize LAS F1 scores across time periods for three parsing models."""
import re
import matplotlib.pyplot as plt
import numpy as np
def parse_scores_file(filepath):
"""Extract average LAS F1 scores per time period from a scores file."""
with open(filepath, 'r') as f:
content = f.read()
scores = {}
# Pattern for time period averages
period_pattern = r'Average scores for time period (\d{4}-\d{4}).*?\nMetric\t.*?\nUPOS\t.*?\nUAS\t.*?\nLAS\t[\d.]+\t[\d.]+\t([\d.]+)'
for match in re.finditer(period_pattern, content, re.DOTALL):
period = match.group(1)
las_f1 = float(match.group(2))
scores[period] = las_f1
# Pattern for overall average
overall_pattern = r'Overall scores across all time periods.*?\nMetric\t.*?\nUPOS\t.*?\nUAS\t.*?\nLAS\t[\d.]+\t[\d.]+\t([\d.]+)'
match = re.search(overall_pattern, content, re.DOTALL)
if match:
scores['Overall'] = float(match.group(1))
return scores
def main():
# Parse scores from all three files
scores_dir = 'eval/scores'
talbanken = parse_scores_file(f'{scores_dir}/scores_talbanken.txt')
transformer_silver = parse_scores_file(f'{scores_dir}/scores_transformer_silver.txt')
transformer_no_silver = parse_scores_file(f'{scores_dir}/scores_transformer_no_silver.txt')
# Define time periods in order
time_periods = ['1700-1750', '1750-1800', '1800-1850', '1850-1900', '1900-1950', 'Overall']
# Extract values for each model
talbanken_vals = [talbanken.get(p, 0) for p in time_periods]
silver_vals = [transformer_silver.get(p, 0) for p in time_periods]
no_silver_vals = [transformer_no_silver.get(p, 0) for p in time_periods]
# Create grouped bar chart
x = np.arange(len(time_periods))
width = 0.25
fig, ax = plt.subplots(figsize=(12, 6))
bars1 = ax.bar(x - width, talbanken_vals, width, label='Talbanken', color='#2ecc71')
bars2 = ax.bar(x, silver_vals, width, label='Transformer Silver', color='#3498db')
bars3 = ax.bar(x + width, no_silver_vals, width, label='Transformer No Silver', color='#e74c3c')
# Customize the plot
ax.set_xlabel('Time Period', fontsize=12)
ax.set_ylabel('LAS F1 Score', fontsize=12)
ax.set_title('LAS F1 Scores by Time Period and Model', fontsize=14)
ax.set_xticks(x)
ax.set_xticklabels(time_periods, rotation=45, ha='right')
ax.legend(loc='upper left')
ax.set_ylim(0.4, 0.9)
# Add value labels on bars, with bold for winners
def add_labels(bars, all_vals, model_idx):
"""Add labels to bars, bold if this model is the winner for that category."""
for i, bar in enumerate(bars):
height = bar.get_height()
# Check if this model is the winner for this time period
period_vals = [all_vals[0][i], all_vals[1][i], all_vals[2][i]]
is_winner = (model_idx == period_vals.index(max(period_vals)))
fontweight = 'bold' if is_winner else 'normal'
ax.annotate(f'{height:.3f}',
xy=(bar.get_x() + bar.get_width() / 2, height),
xytext=(0, 3),
textcoords="offset points",
ha='center', va='bottom', fontsize=8, rotation=90,
fontweight=fontweight)
all_vals = [talbanken_vals, silver_vals, no_silver_vals]
add_labels(bars1, all_vals, 0)
add_labels(bars2, all_vals, 1)
add_labels(bars3, all_vals, 2)
plt.tight_layout()
plt.savefig('plot/las_f1_scores_comparison.png', dpi=400)
print("Saved plot to plot/las_f1_scores_comparison.png and .pdf")
plt.show()
if __name__ == '__main__':
main()
|