oxyde 0.1.10

AI Agent SDK for Game NPCs
Documentation 
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
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281

#!/usr/bin/env python3
"""
Analysis and visualization script for behavior priority study

Generates plots comparing emotion-modulated vs baseline strategies
"""

import json
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from pathlib import Path

# Set style
sns.set_style("whitegrid")
plt.rcParams['figure.figsize'] = (12, 8)

def load_data():
    """Load experimental results"""
    results_dir = Path("experiments/behavior_priority_study/results")

    # Load detailed results
    with open(results_dir / "detailed_results.json") as f:
        results = json.load(f)

    # Load statistics
    with open(results_dir / "statistics.json") as f:
        stats = json.load(f)

    # Create DataFrame
    df = pd.DataFrame(results)

    return df, stats

def plot_behavior_distribution(df, output_dir):
    """Plot behavior selection distribution by strategy"""
    fig, axes = plt.subplots(1, 3, figsize=(18, 6))

    strategies = df['strategy'].unique()

    for idx, strategy in enumerate(strategies):
        strategy_data = df[df['strategy'] == strategy]
        behavior_counts = strategy_data['selected_behavior'].value_counts()

        axes[idx].bar(range(len(behavior_counts)), behavior_counts.values)
        axes[idx].set_xticks(range(len(behavior_counts)))
        axes[idx].set_xticklabels(behavior_counts.index, rotation=45, ha='right')
        axes[idx].set_title(f'{strategy.replace("_", " ").title()}\nBehavior Distribution')
        axes[idx].set_ylabel('Count')
        axes[idx].set_xlabel('Behavior')

    plt.tight_layout()
    plt.savefig(output_dir / 'behavior_distribution.png', dpi=300, bbox_inches='tight')
    print(f"✓ Saved behavior_distribution.png")
    plt.close()

def plot_variety_comparison(stats, output_dir):
    """Plot behavior variety scores"""
    strategies = [s['strategy'] for s in stats]
    variety_scores = [s['variety_score'] for s in stats]

    plt.figure(figsize=(10, 6))
    bars = plt.bar(strategies, variety_scores, color=['#2ecc71', '#3498db', '#e74c3c'])
    plt.ylabel('Variety Score (Shannon Entropy)', fontsize=12)
    plt.xlabel('Strategy', fontsize=12)
    plt.title('Behavior Variety Comparison\n(Higher = More Diverse Behavior)', fontsize=14, fontweight='bold')
    plt.xticks([s.replace('_', ' ').title() for s in strategies])

    # Add value labels on bars
    for bar in bars:
        height = bar.get_height()
        plt.text(bar.get_x() + bar.get_width()/2., height,
                f'{height:.3f}',
                ha='center', va='bottom', fontweight='bold')

    plt.tight_layout()
    plt.savefig(output_dir / 'variety_comparison.png', dpi=300, bbox_inches='tight')
    print(f"✓ Saved variety_comparison.png")
    plt.close()

def plot_expected_match_rate(stats, output_dir):
    """Plot how well behaviors matched expected categories"""
    strategies = [s['strategy'] for s in stats]
    match_rates = [s['expected_match_rate'] for s in stats]

    plt.figure(figsize=(10, 6))
    bars = plt.bar(strategies, match_rates, color=['#2ecc71', '#3498db', '#e74c3c'])
    plt.ylabel('Match Rate (%)', fontsize=12)
    plt.xlabel('Strategy', fontsize=12)
    plt.title('Expected Behavior Match Rate\n(Higher = More Contextually Appropriate)', fontsize=14, fontweight='bold')
    plt.xticks([s.replace('_', ' ').title() for s in strategies])
    plt.ylim(0, 100)

    # Add value labels
    for bar in bars:
        height = bar.get_height()
        plt.text(bar.get_x() + bar.get_width()/2., height,
                f'{height:.1f}%',
                ha='center', va='bottom', fontweight='bold')

    plt.tight_layout()
    plt.savefig(output_dir / 'match_rate_comparison.png', dpi=300, bbox_inches='tight')
    print(f"✓ Saved match_rate_comparison.png")
    plt.close()

def plot_priority_modulation(df, output_dir):
    """Plot priority modulation over scenarios"""
    emotion_mod = df[df['strategy'] == 'emotion_modulated']

    if emotion_mod.empty:
        print("! No emotion-modulated data found")
        return

    fig, axes = plt.subplots(2, 1, figsize=(14, 10))

    # Plot 1: Base vs Final Priority
    scenarios = emotion_mod['scenario'].unique()
    for scenario in scenarios:
        scenario_data = emotion_mod[emotion_mod['scenario'] == scenario]
        axes[0].plot(scenario_data['step'], scenario_data['base_priority'],
                    'o--', alpha=0.6, label=f'{scenario} (base)')
        axes[0].plot(scenario_data['step'], scenario_data['final_priority'],
                    's-', alpha=0.8, label=f'{scenario} (final)')

    axes[0].set_xlabel('Interaction Step', fontsize=11)
    axes[0].set_ylabel('Priority', fontsize=11)
    axes[0].set_title('Emotional Priority Modulation\nBase vs Final Priority', fontsize=13, fontweight='bold')
    axes[0].legend(bbox_to_anchor=(1.05, 1), loc='upper left', fontsize=8)
    axes[0].grid(True, alpha=0.3)

    # Plot 2: Emotional Modifier Impact
    for scenario in scenarios:
        scenario_data = emotion_mod[emotion_mod['scenario'] == scenario]
        axes[1].plot(scenario_data['step'], scenario_data['emotional_modifier'],
                    'o-', alpha=0.7, label=scenario)

    axes[1].set_xlabel('Interaction Step', fontsize=11)
    axes[1].set_ylabel('Emotional Modifier', fontsize=11)
    axes[1].set_title('Emotional Modifier Over Time\n(Positive = Priority Boost, Negative = Priority Reduction)',
                     fontsize=13, fontweight='bold')
    axes[1].legend(bbox_to_anchor=(1.05, 1), loc='upper left', fontsize=9)
    axes[1].axhline(y=0, color='k', linestyle='--', alpha=0.3)
    axes[1].grid(True, alpha=0.3)

    plt.tight_layout()
    plt.savefig(output_dir / 'priority_modulation.png', dpi=300, bbox_inches='tight')
    print(f"✓ Saved priority_modulation.png")
    plt.close()

def plot_emotion_behavior_heatmap(df, output_dir):
    """Heatmap showing which emotions trigger which behaviors"""
    emotion_mod = df[df['strategy'] == 'emotion_modulated']

    if emotion_mod.empty:
        return

    # Create pivot table
    pivot = emotion_mod.groupby(['dominant_emotion', 'selected_behavior']).size().unstack(fill_value=0)

    plt.figure(figsize=(12, 8))
    sns.heatmap(pivot, annot=True, fmt='d', cmap='YlOrRd', cbar_kws={'label': 'Count'})
    plt.title('Emotion-Behavior Association\n(Emotion-Modulated Strategy)', fontsize=14, fontweight='bold')
    plt.xlabel('Selected Behavior', fontsize=12)
    plt.ylabel('Dominant Emotion', fontsize=12)
    plt.tight_layout()
    plt.savefig(output_dir / 'emotion_behavior_heatmap.png', dpi=300, bbox_inches='tight')
    print(f"✓ Saved emotion_behavior_heatmap.png")
    plt.close()

def plot_scenario_comparison(df, output_dir):
    """Compare behavior selections across scenarios"""
    fig, axes = plt.subplots(5, 1, figsize=(14, 18))

    scenarios = sorted(df['scenario'].unique())
    strategies = df['strategy'].unique()

    for idx, scenario in enumerate(scenarios):
        scenario_data = df[df['scenario'] == scenario]

        for strategy in strategies:
            strategy_data = scenario_data[scenario_data['strategy'] == strategy]
            steps = strategy_data['step'].values
            # Encode behaviors as numbers for plotting
            behavior_map = {b: i for i, b in enumerate(df['selected_behavior'].unique())}
            behaviors = [behavior_map[b] for b in strategy_data['selected_behavior'].values]

            axes[idx].plot(steps, behaviors, 'o-', label=strategy, alpha=0.7, markersize=8)

        axes[idx].set_title(f'Scenario: {scenario.replace("_", " ").title()}', fontsize=12, fontweight='bold')
        axes[idx].set_xlabel('Step')
        axes[idx].set_ylabel('Behavior')
        axes[idx].legend()
        axes[idx].grid(True, alpha=0.3)

    plt.tight_layout()
    plt.savefig(output_dir / 'scenario_comparison.png', dpi=300, bbox_inches='tight')
    print(f"✓ Saved scenario_comparison.png")
    plt.close()

def generate_summary_report(df, stats, output_dir):
    """Generate markdown summary report"""
    report = """# Behavior Priority Study - Results

## Overview
This experiment compares emotion-modulated behavior selection against traditional approaches.

## Key Findings

"""

    # Add statistics comparison
    for stat in stats:
        report += f"### {stat['strategy'].replace('_', ' ').title()}\n\n"
        report += f"- **Variety Score**: {stat['variety_score']:.3f}\n"
        report += f"- **Expected Match Rate**: {stat['expected_match_rate']:.1f}%\n"
        report += f"- **Avg Priority Override**: {stat['avg_priority_override']:.2f}\n"
        report += f"- **Total Interactions**: {stat['total_interactions']}\n\n"

        report += "**Behavior Distribution**:\n"
        for behavior, count in sorted(stat['behavior_counts'].items(), key=lambda x: x[1], reverse=True):
            pct = (count / stat['total_interactions']) * 100
            report += f"- {behavior}: {count} ({pct:.1f}%)\n"
        report += "\n"

    # Conclusions
    emotion_mod_stats = next(s for s in stats if s['strategy'] == 'emotion_modulated')
    fixed_stats = next(s for s in stats if s['strategy'] == 'fixed_priority')

    variety_improvement = ((emotion_mod_stats['variety_score'] - fixed_stats['variety_score'])
                          / fixed_stats['variety_score'] * 100)

    report += f"""## Conclusions

1. **Behavior Variety**: Emotion-modulated approach shows {variety_improvement:+.1f}% difference in variety compared to fixed priority
2. **Context Appropriateness**: {emotion_mod_stats['expected_match_rate']:.1f}% of behaviors matched expected categories
3. **Emotional Influence**: Average priority modification of {emotion_mod_stats['avg_priority_override']:.2f}

## Visualizations

See generated PNG files for detailed visualizations.
"""

    with open(output_dir / 'REPORT.md', 'w') as f:
        f.write(report)

    print(f"✓ Saved REPORT.md")

def main():
    print("==================================================")
    print("Analyzing Behavior Priority Study Results")
    print("==================================================\n")

    # Load data
    print("Loading data...")
    df, stats = load_data()
    print(f"✓ Loaded {len(df)} interactions\n")

    # Create output directory
    output_dir = Path("experiments/behavior_priority_study/results")
    output_dir.mkdir(parents=True, exist_ok=True)

    # Generate plots
    print("Generating visualizations...")
    plot_behavior_distribution(df, output_dir)
    plot_variety_comparison(stats, output_dir)
    plot_expected_match_rate(stats, output_dir)
    plot_priority_modulation(df, output_dir)
    plot_emotion_behavior_heatmap(df, output_dir)
    plot_scenario_comparison(df, output_dir)

    # Generate report
    print("\nGenerating summary report...")
    generate_summary_report(df, stats, output_dir)

    print("\n==================================================")
    print("✓ Analysis complete!")
    print(f"✓ Results saved to {output_dir}")
    print("==================================================")

if __name__ == "__main__":
    main()