基于潜在语义分析的领域知识地图构建技术

LSA-Based Domain Knowledge Map Construction Method

  • 摘要: 信息过载现象普遍存在于产品设计过程中,而提升信息和知识检索的准确性是解决该问题的主要手段. 领域知识地图能够为知识检索引擎提供领域知识关联信息,以提升知识检索的准确性. 针对领域知识地图的构建,提出了一种半自动化的领域知识地图构建过程. 该过程采用领域专家定义领域特征集的方法来提升领域特征集的全面性和独立性,进一步采用潜在语义分析技术(LSA)消除领域特征集中存在的语义交叉. 通过分析领域知识相似度的分布,采用优先去除中距离的策略构建最终的知识地图. 最后给出数控领域知识地图的构建过程,证明了该方法的有效性.

     

    Abstract: Information overload is widely exist in product design process. To alleviate the information overload, the main approach is to improve the accuracy of the search engine. The knowledge map can be used to achieve the goal by providing the knowledge association information. A semi-automation process of constructing domain knowledge map was proposed. In the process, the domain feature set was defined by domain experts to guarantee its comprehensiveness and independence, and the latent semantic analysis was adopted to eliminate the semantic overlapping in the feature set. Analyzed the distribution of the domain knowledge similarity, an ultimate domain knowledge map was built with a middle-distance strategy. At last, a knowledge map for computer numerical control (CNC) domain was constructed by the proposed method to demonstrate the feasibility.

     

/

返回文章
返回
Baidu
map