基于知识图谱的大学生就业竞争力动态评估与成长路径推荐方法研究

Knowledge Graph-Based Dynamic Assessment of University Students’ Employability and Growth Path Recommendation

摘要: 目的：针对高校就业指导中证据分散、岗位能力难转译和建议缺少反馈的问题，提出知识图谱驱动的动态评估与路径推荐方法。方法：以证据事件、能力节点、岗位要求和成长任务为对象，规定字段、值域、版本和审核状态；采用MacBERT-CRF抽取实体，利用SimCSE归并同义表达，以TransR表示图关系，通过时间衰减评分、混合匹配和约束束搜索生成差距诊断与任务序列；关键建议实行双教师复核、第三人裁决及申诉回写。结果：面向两专业构建原型系统，实现标准化、画像更新、岗位匹配和证据回写。结论：该方法将判断依据、能力缺口、成长任务和复核责任纳入闭环，支持可追溯、可解释、可申诉的就业指导。

Abstract: Objective: This study addresses fragmented student evidence, weak job translation, and guidance without feedback. Methods: Evidence events, competency nodes, job requirements, and growth tasks are defined by fields, values, versions, and review states. MacBERT-CRF extracts job entities, SimCSE normalizes synonyms, and TransR represents graph relations. Time-decayed scoring, hybrid matching, and constrained beam search generate gap diagnoses and task sequences. Low-confidence mappings and high-impact recommendations undergo review by two teachers, adjudication, and appeal-based versioning. Results: Developing a prototype system for Data Science and Big Data Technology and Statistics supports standardization, profile updates, diagnosis, recommendation, and feedback. Conclusion: The method forms a traceable and contestable guidance loop.