Using artificial intelligent (AI) to re-design and enhance the current wireless communication system is a promising pathway for the future sixth-generation (6G) wireless network. The performance of AI-enabled wireless communication depends heavily on the quality of wireless air-interface data. Although there are various approaches to data quality assessment (DQA) for different applications, none has been designed for wireless air-interface data. In this paper, we propose a DQA framework to measure the quality of wireless air-interface data from three aspects: similarity, diversity, and completeness. The similarity measures how close the considered datasets are in terms of their statistical distributions; the diversity measures how well-rounded a dataset is, while the completeness measures to what degree the considered dataset satisfies the required performance metrics in an application scenario. The proposed framework can be applied to various types of wireless air-interface data, such as channel state information (CSI), signal-to-interference-plus-noise ratio (SINR), reference signal received power (RSRP), etc. For simplicity, the validity of our proposed DQA framework is corroborated by applying it to CSI data and using similarity and diversity metrics to improve CSI compression and recovery in Massive MIMO systems.