Abstract: 【Purpose/significance】This study is to help businesses and platforms obtain user requirements in mobile commerce online reviews, and solve the problem of user requirements extraction caused by online review overload.【Method/process】Firstly, this paper obtains the original online review data set for text preprocessing and cleaning. Then, it goes deep into semantic level to achieve user needs aggregation based on the improved Canopy-Kmeans algorithm; finally, the Kano questionnaire is designed with the aggregation results as the hierarchical index, and the classification standard of user needs is optimized with the importance discrimination method and the user satisfaction index to achieve the efficient aggregation and accurate mining of user needs.【Result/conclusion】Through
the comparative analysis of the experimental results, it is found that compared with the traditional semantic-based clustering method,the clustering results of the mobile commerce user demand aggregation and mining method designed in this paper are more clear and reasonable, and can obtain more accurate and refined user needs.【Innovation/limitation】The Word2Vec model is used to analyze user needs from a semantic perspective, and the optimized Canopy-Kmeans algorithm is used to aggregate user needs, but the selected re‐search objects and data size are relatively limited, then the research scope and experimental data size of online product comments will be expanded in the future.