数据集概览
| 数据集名称 | 企业项目合同数据集 |
| 数据总量 | 42,000 份（含合同文本 + 结构化条款字段） |
| 合同类型 | 工程总承包 / 技术服务 / 采购 / 租赁 / 投融资 / 委托开发 / 战略合作 |
| 覆盖行业 | 建筑工程 / IT互联网 / 制造业 / 能源 / 金融 / 医疗 / 教育 |
| 时间跨度 | 2016 - 2026 年 |
| 数据来源 | 上市公司公告、政府采购平台、企业ERP系统（经授权脱敏） |
| 标注方式 | 法务专业人员 + NLP预标注辅助 |
数据维度详细说明
| 字段名称 | 数据类型 | 取值范围/说明 |
|---|---|---|
| contract_id | String(32) | 脱敏合同ID |
| contract_type | Enum | EPC | TECH_SERVICE | PROCUREMENT | LEASE | INVESTMENT | RND | STRATEGIC |
| industry | Enum | CONSTRUCTION | IT | MANUFACTURING | ENERGY | FINANCE | HEALTHCARE | EDUCATION |
| contract_amount | Float | 合同金额（万元） |
| signing_date | Date | 签订日期（仅保留年月） |
| duration_months | Integer | 合同期限（月） |
| parties | Object | {party_a: {type, region}, party_b: {type, region}} |
| key_clauses | Array | [{clause_type, risk_level, summary, original_text_start, original_text_end}] |
| risk_indicators | Object | {payment_risk, delivery_risk, legal_risk, overall_risk} 0-100 |
| payment_terms | Object | {milestones, percentages, conditions} |
脱敏 JSON 数据样例
[
{
"contract_id": "CTR-2024-A0008234",
"contract_type": "EPC",
"industry": "CONSTRUCTION",
"contract_amount": 18500.0,
"signing_date": "2024-03",
"duration_months": 24,
"parties": {"party_a": {"type": "GOV", "region": "华南"}, "party_b": {"type": "SOE", "region": "华东"}},
"key_clauses": [
{"clause_type": "付款条款", "risk_level": "MEDIUM", "summary": "按工程进度分5期支付，每期20%"},
{"clause_type": "违约责任", "risk_level": "HIGH", "summary": "逾期交付每日罚合同总额0.05%"}
],
"risk_indicators": {"payment_risk": 35, "delivery_risk": 60, "legal_risk": 25, "overall_risk": 42},
"payment_terms": {"milestones": 5, "percentages": [20, 20, 20, 20, 20], "conditions": ["基础验收", "主体封顶", "竣工预验收", "竣工验收", "质保期满"]}
},
{
"contract_id": "CTR-2023-B0015621",
"contract_type": "TECH_SERVICE",
"industry": "IT",
"contract_amount": 480.0,
"signing_date": "2023-08",
"duration_months": 12,
"parties": {"party_a": {"type": "PRIVATE", "region": "华北"}, "party_b": {"type": "PRIVATE", "region": "华中"}},
"key_clauses": [
{"clause_type": "知识产权", "risk_level": "HIGH", "summary": "开发成果知识产权归甲方所有，乙方保留署名权"}
],
"risk_indicators": {"payment_risk": 15, "delivery_risk": 30, "legal_risk": 50, "overall_risk": 32},
"payment_terms": {"milestones": 3, "percentages": [30, 40, 30], "conditions": ["合同签订", "中期验收", "终验"]}
},
{
"contract_id": "CTR-2024-C0032891",
"contract_type": "PROCUREMENT",
"industry": "MANUFACTURING",
"contract_amount": 3200.0,
"signing_date": "2024-06",
"duration_months": 6,
"parties": {"party_a": {"type": "SOE", "region": "华东"}, "party_b": {"type": "FOREIGN", "region": "海外"}},
"key_clauses": [
{"clause_type": "质量保证", "risk_level": "LOW", "summary": "质保期24个月，故障响应4小时内"}
],
"risk_indicators": {"payment_risk": 20, "delivery_risk": 25, "legal_risk": 15, "overall_risk": 18},
"payment_terms": {"milestones": 2, "percentages": [70, 30], "conditions": ["发货", "验收合格"]}
},
{
"contract_id": "CTR-2025-D0074532",
"contract_type": "INVESTMENT",
"industry": "ENERGY",
"contract_amount": 52000.0,
"signing_date": "2025-01",
"duration_months": 120,
"parties": {"party_a": {"type": "GOV", "region": "西南"}, "party_b": {"type": "SOE", "region": "华北"}},
"key_clauses": [
{"clause_type": "投资回报", "risk_level": "HIGH", "summary": "特许经营期25年，内部收益率不低于8%"}
],
"risk_indicators": {"payment_risk": 55, "delivery_risk": 40, "legal_risk": 30, "overall_risk": 48},
"payment_terms": {"milestones": 4, "percentages": [25, 30, 30, 15], "conditions": ["项目核准", "开工", "并网", "稳定运行"]}
},
{
"contract_id": "CTR-2024-E0021456",
"contract_type": "STRATEGIC",
"industry": "HEALTHCARE",
"contract_amount": 0.0,
"signing_date": "2024-11",
"duration_months": 36,
"parties": {"party_a": {"type": "LISTED", "region": "华东"}, "party_b": {"type": "PRIVATE", "region": "华中"}},
"key_clauses": [
{"clause_type": "合作范围", "risk_level": "LOW", "summary": "AI医疗影像产品联合研发与市场推广"}
],
"risk_indicators": {"payment_risk": 10, "delivery_risk": 20, "legal_risk": 35, "overall_risk": 22},
"payment_terms": null
}
]
AI 训练适用场景
合同智能审查
适用模型：Legal-BERT / ChatLaw + 多标签分类。自动识别合同中的风险条款、缺失条款和不公平条款。
合同条款自动生成
适用模型：LLM（GPT-4o / DeepSeek-V3）+ Few-shot Prompting。根据合同类型和参数自动生成标准条款文本。
合同风险评估
适用模型：XGBoost / TabNet。基于历史合同的条款特征和履约结果训练合同风险评分模型。
智能合同管理（CLM）
适用模型：NER + RE + 规则引擎。自动提取合同关键日期、金额、履约节点实现合同全生命周期管理。