soul-mirror
Version:
🔮 融合千年易学智慧与现代AI技术，为程序员量身打造的命理情绪分析工具。94.1%情绪识别准确率，秒级响应，支持五行人格分析与实时运势指导。
297 lines (266 loc) • 8.47 kB
JavaScript
/**
* 时间与语境修正模块
* @module ContextEmotionCorrector
* @since 0.8.0
*/
// 昼夜节律对情绪的影响系数
const TIME_EMOTION_WEIGHTS = {
'morning': { // 6:00-11:00
'高兴': 1.2,
'愤怒': 0.8,
'悲伤': 0.7,
'焦虑': 0.9,
'平静': 1.1,
'疲惫': 0.6,
'惊讶': 1.0,
'无聊': 0.8
},
// 下午 (11:00-17:00) - 情绪相对平衡
'afternoon': {
'高兴': 1.0,
'愤怒': 1.0,
'悲伤': 1.0,
'焦虑': 1.1,
'平静': 1.0,
'疲惫': 1.2,
'惊讶': 1.0,
'无聊': 1.1
},
// 傍晚 (17:00-22:00) - 更容易疲惫和放松
'evening': {
'高兴': 1.1,
'愤怒': 1.1,
'悲伤': 1.0,
'焦虑': 1.2,
'平静': 1.2,
'疲惫': 1.3,
'惊讶': 0.9,
'无聊': 1.0
},
// 深夜 (22:00-6:00) - 更容易负面情绪
'night': {
'高兴': 0.8,
'愤怒': 1.2,
'悲伤': 1.3,
'焦虑': 1.4,
'平静': 0.9,
'疲惫': 1.5,
'惊讶': 0.8,
'无聊': 1.2
}
};
// 上下文关键词修正规则
const CONTEXT_CORRECTION_RULES = {
// 工作相关上下文
'work': {
keywords: ['代码', '项目', '工作', 'bug', '加班', '开发', '程序', '系统', '功能', 'deadline', 'review'],
adjustments: {
'高兴': { bonus: 0.1, reason: '工作成就感加成' },
'愤怒': { bonus: 0.2, reason: '工作压力放大' },
'焦虑': { bonus: 0.2, reason: '工作焦虑加强' },
'疲惫': { bonus: 0.3, reason: '工作疲劳明显' }
}
},
// 学习相关上下文
'learning': {
keywords: ['学习', '掌握', '理解', '搞懂', '弄明白', '学会', '教程', '文档', '知识'],
adjustments: {
'高兴': { bonus: 0.2, reason: '学习成就感' },
'焦虑': { bonus: 0.1, reason: '学习压力' },
'惊讶': { bonus: 0.2, reason: '学习发现' }
}
},
// 社交相关上下文
'social': {
keywords: ['团队', '同事', '朋友', '聊天', '交流', '讨论', '分享', '合作'],
adjustments: {
'高兴': { bonus: 0.1, reason: '社交正面情绪' },
'愤怒': { bonus: 0.1, reason: '人际冲突' },
'悲伤': { bonus: 0.1, reason: '社交压力' }
}
},
// 成就相关上下文
'achievement': {
keywords: ['完成', '成功', '搞定', '解决', '实现', '达成', '通过', '拿下'],
adjustments: {
'高兴': { bonus: 0.3, reason: '成就感强烈' },
'平静': { bonus: 0.1, reason: '完成后的平静' }
}
},
// 问题相关上下文
'problem': {
keywords: ['问题', '错误', '失败', '崩溃', '报错', '异常', '故障', '卡住'],
adjustments: {
'愤怒': { bonus: 0.2, reason: '问题带来挫折' },
'焦虑': { bonus: 0.2, reason: '问题引发焦虑' },
'疲惫': { bonus: 0.1, reason: '解决问题的疲劳' }
}
}
};
// 情绪强度修正词汇
const INTENSITY_MODIFIERS = {
'extreme': {
keywords: ['超级', '特别', '非常', '巨', '极', '死了', '疯了', '炸了', '爆了', '翻了'],
multiplier: 1.5
},
'high': {
keywords: ['很', '太', '好', '挺', '蛮', '相当', '真的'],
multiplier: 1.2
},
'mild': {
keywords: ['有点', '稍微', '还行', '一般', '不太', '略'],
multiplier: 0.8
}
};
class ContextEmotionCorrector {
constructor() {
this.debugMode = false;
}
/**
* 对情绪识别结果进行上下文修正
* @param {string} originalText - 原始文本
* @param {Object} emotionResult - 原始情绪识别结果
* @returns {Object} 修正后的情绪结果
*/
correctWithContext(originalText, emotionResult) {
const corrections = {
timeWeight: 1.0,
contextBonus: 0,
intensityMultiplier: 1.0,
appliedRules: []
};
// 1. 时间段修正
const timeWeight = this._getTimeWeight(emotionResult.main);
corrections.timeWeight = timeWeight;
// 2. 上下文修正
const contextCorrection = this._getContextCorrection(originalText, emotionResult.main);
corrections.contextBonus = contextCorrection.bonus;
corrections.appliedRules.push(...contextCorrection.rules);
// 3. 强度修正
const intensityMultiplier = this._getIntensityMultiplier(originalText);
corrections.intensityMultiplier = intensityMultiplier;
// 4. 重新计算所有情绪得分
const correctedScores = {};
Object.keys(emotionResult.scores).forEach(emotion => {
let score = emotionResult.scores[emotion];
// 应用时间权重
const emotionTimeWeight = this._getTimeWeight(emotion);
score *= emotionTimeWeight;
// 应用上下文加成
const emotionContext = this._getContextCorrection(originalText, emotion);
score += emotionContext.bonus;
// 应用强度修正
score *= intensityMultiplier;
correctedScores[emotion] = Math.max(0, score);
});
// 5. 重新确定主情绪
const sortedEmotions = Object.keys(correctedScores)
.sort((a, b) => correctedScores[b] - correctedScores[a]);
const newMainEmotion = sortedEmotions[0];
const newSecondaryEmotions = sortedEmotions.slice(1).filter(e => correctedScores[e] > 0);
// 6. 重新计算置信度
const maxScore = correctedScores[newMainEmotion];
const totalScore = Object.values(correctedScores).reduce((sum, score) => sum + score, 0);
const newConfidence = totalScore > 0 ?
Math.min(95, Math.max(15, Math.round((maxScore / totalScore) * 100))) :
emotionResult.confidence;
return {
...emotionResult,
main: newMainEmotion,
secondary: newSecondaryEmotions.slice(0, 2),
scores: correctedScores,
confidence: newConfidence,
corrections: corrections,
_originalResult: emotionResult
};
}
/**
* 获取当前时间段的情绪权重
* @private
*/
_getTimeWeight(emotion) {
const hour = new Date().getHours();
let timeSlot;
if (hour >= 6 && hour < 11) {
timeSlot = 'morning';
} else if (hour >= 11 && hour < 17) {
timeSlot = 'afternoon';
} else if (hour >= 17 && hour < 22) {
timeSlot = 'evening';
} else {
timeSlot = 'night';
}
return TIME_EMOTION_WEIGHTS[timeSlot][emotion] || 1.0;
}
/**
* 获取上下文修正
* @private
*/
_getContextCorrection(text, emotion) {
let totalBonus = 0;
const appliedRules = [];
Object.entries(CONTEXT_CORRECTION_RULES).forEach(([contextType, rule]) => {
const matchedKeywords = rule.keywords.filter(keyword =>
text.toLowerCase().includes(keyword.toLowerCase())
);
if (matchedKeywords.length > 0) {
const adjustment = rule.adjustments[emotion];
if (adjustment) {
// 匹配关键词越多，加成越大，但有上限
const keywordBonus = Math.min(matchedKeywords.length * adjustment.bonus, adjustment.bonus * 2);
totalBonus += keywordBonus;
appliedRules.push({
context: contextType,
keywords: matchedKeywords,
bonus: keywordBonus,
reason: adjustment.reason
});
}
}
});
return { bonus: totalBonus, rules: appliedRules };
}
/**
* 获取强度修正倍数
* @private
*/
_getIntensityMultiplier(text) {
const lowerText = text.toLowerCase();
// 检查极强强度词汇
for (const keyword of INTENSITY_MODIFIERS.extreme.keywords) {
if (lowerText.includes(keyword)) {
return INTENSITY_MODIFIERS.extreme.multiplier;
}
}
// 检查高强度词汇
for (const keyword of INTENSITY_MODIFIERS.high.keywords) {
if (lowerText.includes(keyword)) {
return INTENSITY_MODIFIERS.high.multiplier;
}
}
// 检查轻度强度词汇
for (const keyword of INTENSITY_MODIFIERS.mild.keywords) {
if (lowerText.includes(keyword)) {
return INTENSITY_MODIFIERS.mild.multiplier;
}
}
return 1.0;
}
/**
* 设置调试模式
*/
setDebugMode(enabled) {
this.debugMode = enabled;
}
/**
* 获取当前时间段描述
*/
getCurrentTimeSlot() {
const hour = new Date().getHours();
if (hour >= 6 && hour < 11) return '早晨';
if (hour >= 11 && hour < 17) return '下午';
if (hour >= 17 && hour < 22) return '傍晚';
return '深夜';
}
}
module.exports = ContextEmotionCorrector;