With the rapid development of information technology and the Internet, the existing data query system suffers from poor performance, single structure and slow efficiency in processing data. In view of this, the study introduces the GmmCube multidimensional data intelligent query system and combines column-to-block structure and single-dimension matrix for structure optimization. On the basis of guaranteeing specific dimensions, it continues to introduce the sparse prefix sum for efficiency optimization. The experimental results showed that the maximum storage space consumption was 1050 MB in the training set and 1040 MB in the test set after comparing with three commonly used algorithms: prefix sum array, binary index tree, and segment tree. The maximum deviation in query time was 0.1 hours. When querying Weibo big data (10 million records), the lowest measured absolute error was 2%, the precision rate was 93.56%, and the recall rate was 93.74%. It can be concluded that the novel data query technique proposed by the study is superior in data query performance and efficiency, and can provide an effective solution for big data query analysis.