The particularity of aeronautic products results in the difficulty of prevailing product recommendation algorithm in the industry to be applied to civil aviation e-commerce. This paper, in combination of big data technology on the basis of clustering technology and collaborative filtering algorithm, proposes a multi-level abstract computing method of similarity based on the scene. It has solved the difficulty that aeronautic products have encountered in terms of precise recommendation in the civil aviation industry. As indicated by the precise marketing of an aviation company, this method can promote ROI by three times.