Component ranking based on hierarchy structure
Abstract
Many component retrieval models lack of an efficient and accurate method to rank components because most of them ignore information hidden in components and the impact of component structure. In this paper, we propose a hierarchy ranking model combining the importance of hierarchical structure of component with the frequency of keywords retrieved from components. Experiments prove all keywords from different hierarchy level of components are necessary to retrieval and the importance of keywords can be weighted by their hierarchy structure automatically. Moreover, performance of the model is improved by combining with vector space. The dominance of this model can be shown in experiments compared with other similar ranking models.