Evaluation method for interoperability of weapon systems applying natural language processing techniques

Abstract

The current weapon system is operated as a complex weapon system with various standards and protocols applied, so there is a risk of failure in smooth information exchange during combined and joint operations on the battlefield. The interoperability of weapon systems to carry out precise strikes on key targets through rapid situational judgment between weapon systems is a key element in the conduct of war. Since the Korean military went into service, there has been a need to change the configuration and improve performance of a large number of software and hardware, but there is no verification system for the impact on interoperability, and there are no related test tools and facilities. In addition, during combined and joint training, errors frequently occur during use after arbitrarily changing the detailed operation method and software of the weapon/power support system. Therefore, periodic verification of interoperability between weapon systems is necessary. To solve this problem, rather than having people schedule an evaluation period and conduct the evaluation once, AI should continuously evaluate the interoperability between weapons and power support systems 24 hours a day to advance warfighting capabilities. To solve these problems, To this end, preliminary research was conducted to improve defense interoperability capabilities by applying natural language processing techniques (①Word2Vec model, ②FastText model, ③Swivel model) (using published algorithms and source code). Based on the results of this experiment, we would like to present a methodology (automated evaluation of interoperability requirements evaluation / level measurement through natural language processing model) to implement an automated defense interoperability evaluation tool without relying on humans.