The military domain knowledge question answering system based on Retrieval-Augmented Generation （RAG） has become an important tool for modern intelligence personnel to collect and analyze intelligence gradually. Focusing on the issue that the application strategies of RAG methods currently suffer from poor portability in hybrid retrieval as well as the problem of semantic drift caused by unnecessary query rewriting easily， a Multi-Strategy Retrieval-Augmented Generation （MSRAG） method was proposed. Firstly， the retrieval model was matched adaptively to recall relevant text based on query characteristics of the user input. Secondly， a text filter was utilized to extract the key text fragments that can answer the question. Thirdly， the content validity was assessed by the text filter to trigger query rewriting based on synonym expansion， and the initial query was merged with the rewritten information and used as input of the retrieval controller for more targeted re-retrieval. Finally， the key text fragments that can answer the question were merged with the question， prompt engineering input was used to generate answer model， and the response generated by the model was returned to the user. Experimental results show that compared to the convex linear combination RAG method， MSRAG method improves the ROUGE-L （Recall-Oriented Understudy for Gisting Evaluation Longest common subsequence） by 14.35 percentage points on the Military domain dataset （Military） and by 5.83 percentage points on the Medical dataset. It can be seen that MSRAG method has strong universality and portability， enables the reduction of the semantic drift caused by unnecessary query rewriting， and effectively helps large language models generate more accurate answers.

Latent Factor Model （LFM） have been widely used in recommendation field due to their excellent performance. In addition to interactive data， auxiliary information is also introduced to solve the problem of data sparsity， thereby improving the performance of recommendations. However， most LFMs still have some problems. First， when modeling users by LFM， how users make decisions on items based on their feature preferences is ignored. Second， the feature interaction using inner product assumes that the feature dimensions are independent to each other， without considering the correlation between the feature dimensions. In order to solve the above problems， a recommendation model for User Attribute preference Modeling based on Convolutional Neural Network （CNN） interaction （UAMC） was proposed. In this model， the general preferences of users， user attributes and item embeddings were firstly obtained， and then the user attributes and item embeddings were interacted to explore the preferences of different attributes of users to different items. After that， the interacted user preference attributes were sent to the CNN layer to explore the correlation between different dimensions of different preference attributes and thus obtain the users’ attribute preference vectors. Next， the attention mechanism was used to combine the general preferences of the users with the attribute preferences obtained from CNN layer to obtain the vector representations of the users. Finally， the dot product was used to calculate the users’ ratings of the items. Experiments were conducted on three real datasets： Movielens-100K， Movielens-1M and Book-crossing. The results show that the proposed algorithm decreases the Root Mean Square Error （RMSE） by 1.75%， 2.78% and 0.25% respectively compared with the model of Neural Factorization Machine for sparse predictive analytics （NFM）， which verifies the effectiveness of UAMC model in improving the accuracy of recommendation in the rating prediction recommendation of LFM.