An Attention-Based Deep learning Situation Information Recommendation Model

Massive battlefield data, fast combat rhythm and complex combat system pose great challenges to commanders’ limited cognitive ability. While increasing available information and enhancing situational awareness, it will also lead to information overload, which will seriously interfere with commanders’ extraction and effective utilization of useful information. This paper puts forward a deep-learning situation information recommendation model based on attention mechanism in view of overload of situation information. It introduces a double-layer attention mechanism, and uses a multi-layer neural network to learn the attention weights at item-level and component-level respectively in order to explore the potential relationship between commander and situation information, build a commander preference prediction model and realize personalized recommendation.