关键词: 信息推荐, 堆叠降噪自编码器, 精确欧式局部敏感哈希, 数据降维, 冷启动

Abstract: There are several problems in the domain of recommendation method based on k nearest neighbors, such as the high dimensionality of rating features, the slow speed of searching nearest neighbors and the cold start problem of ratings. To solve these problems, a recommendation method based on k nearest neighbors using data dimensionality reduction and Exact Euclidean Locality-Sensitive Hashing is proposed. Firstly, integrating the rating data, the user attribute data and the item category data, reduce the dimension of the data mentioned above using Stack Denoising Auto-encoder neutral network. Then, use the Exact Euclidean Local-Sensitive Hash algorithm to build the index of the reduced dimension data, and retrieve the target users or the target items to get their similar nearest neighbors. Finally, calculate the similarities between the target and the neighbors, and the target user’s similarity-weighted prediction rating for the target item is obtained. The experimental results on standard data sets show that the proposed method improves the accuracy and efficiency of recommendation, and it outperformed other similar method on precision in cold start scene.

Key words: information recommendation, Stack Denoising Auto-encoder, Exact Euclidean Locality-Sensitive Hashing, data dimensionality reduction, cold start