基于标签树的生长干部招生计划聚类方法

生长干部招生计划是规定军队干部补充来源渠道及数量规模的重要依据,对干部队伍建设与发展具有重要的意义。随着招生计划每年拟制与积累,分析与评估计划将会起到辅助决策的作用,对招生工作的实施带来深远影响。本文从数据挖掘的角度,以聚类的方法研究分析招生计划,这为分析历年计划的波动性并理解评价计划与政策的贴合度提供了定量分析的手段。首先分析了招生计划的特点,进而提出以标签树量化招生计划的解决思路,通过抽取标签树中的特征子树作为聚类中的度量特征,并采用共现的方法实施"先形成核心,再依次分类"的步骤完成聚类。实验表明该方法在合成数据集和真实数据集上聚类效果较好、效率较高,对分析招生计划具有一定理论意义。

Clustering cadet recruiting plans based on labeled trees

Cadets recruiting plan is a kind of important rule to prescribe source and quantity of officer. It is significant for construction and development of officer. With accumulation of recruiting plans, it is meaningful to analyze and evaluate the plans. This paper utilized clustering technic to analyze recruiting plans, which will provide a quantitive approach to analyze fluctuation of plans and to understand and evaluate closeness of plans with policies. Firstly, the paper summarized characteristics of recruiting plans. And then it proposed a solution which utilized labeled tree to represent recruiting plan and extracted representative subtrees to participate in clustering. Finally, it uses co-occurrence idea to form clustering cores then classify the plans to finish the clustering process. Experimental results reveal that the method can provide better clustering results and be efficient.