Design and Implementation of Keyword Extraction Framework in Military Domain

In order to improve the performance of keywords extraction in the military domain, this paper analyzes the charateristics of military texts, and gives five principles of military keyword extraction. By imitating "six-step process" that human follow when extracting keywords of military documents, a new technical framework is put forward. Combined multiple features of term frequency, topic and position, a new military keyword extraction algorithm (TTL) is designed and implemented. By making full use of the role of "human in the loop", this framework can meet the demands of multi-user online and distributed application in the Military Knowledge Engineering System. The experimental results show the self-iterative upgradability, self-adaptability and intelligence of the proposed framework. In addtion, the TTL algorithm demonstrates its better performance and effectiveness than TextRank, TF-IDF, and LDA algorithm.