Educational MCP Server

"""认知诊断模块（规则版占位），可平滑替换为 IRT/CDM。""" from __future__ import annotations from statistics import mean from typing import List from schemas import ( CognitiveDiagnosisRequest, CognitiveDiagnosisResponse, ConceptDiagnosis, ConceptSnapshot, ) import database def _level_from_mastery(mastery: float) -> str: if mastery >= 0.85: return "稳定掌握" if mastery >= 0.65: return "发展中" return "高风险" def _recommendation(snapshot: ConceptSnapshot, mastery: float) -> str: if mastery >= 0.85: return "通过挑战性任务保持迁移练习。" if mastery >= 0.65: focus = snapshot.misconceptions[0] if snapshot.misconceptions else "易错点" return f"安排变式练习，突出对比 {focus}。" return "回到概念本源，结合具体例子重新建模。" def diagnose(payload: CognitiveDiagnosisRequest) -> CognitiveDiagnosisResponse: concepts: List[ConceptDiagnosis] = [] strengths: List[str] = [] risks: List[str] = [] for snapshot in payload.concept_snapshots: mastery = round(snapshot.mastery, 3) level = _level_from_mastery(mastery) if level == "稳定掌握": strengths.append(snapshot.concept_name) elif level == "高风险": risks.append(snapshot.concept_name) # 同步入“数据库” database.set_mastery(payload.student_id, snapshot.concept_name, mastery) concepts.append( ConceptDiagnosis( concept_name=snapshot.concept_name, mastery=mastery, level=level, misconceptions=snapshot.misconceptions, recommendation=_recommendation(snapshot, mastery), ) ) overall_mastery = round( mean([concept.mastery for concept in payload.concept_snapshots]) if payload.concept_snapshots else 0.0, 3 ) behavior_note = "" if payload.recent_behaviors: behavior_note = f"行为观察：{'、'.join(payload.recent_behaviors)}。" summary = ( f"{payload.student_id} 在 {payload.subject} 中整体掌握度约为 {overall_mastery:.0%}。" f"优势概念：{', '.join(strengths) or '暂未形成亮点'}；" f"风险概念：{', '.join(risks) or '暂无'}。" f"{behavior_note}" ) return CognitiveDiagnosisResponse( request_id=payload.request_id, student_id=payload.student_id, subject=payload.subject, overall_mastery=overall_mastery, strengths=strengths, risks=risks, concepts=concepts, summary=summary, model_version="rule-0.1", )