The expansion of digital information has led to a need for more flexible and scalable data storage options, where NoSQL databases excel. Leveraging question-answering (QA) systems enables organizations to efficiently extract valuable information and insights through natural language questions from databases without requiring specialized knowledge of query languages. This capability is precious in the movie domain, where users may wish to access specific information about movies, actors, directors, and other related entities quickly and intuitively. QA for NoSQL databases is more challenging than QA for relational databases, mainly due to the need to manage data consistency and integrity, and the limited availability of datasets for text-to-NoSQL conversion. To address this issue, we present a new text-to-MQL dataset, M2Q2 (Movies in MongoDB Question-Query) for question answering over the NoSQL MongoDB document-oriented database in the Movies domain. The M2Q2 construction process consists of three steps: (1) Creating templates of question and MongoDB query pairs, (2) data augmentation, which is based on three methods (paraphrasing, back-translation, and entity replacement), and (3) data revision. The created dataset was evaluated for translation tasks and used in the QA system. The results demonstrate that the M2Q2 dataset facilitates more accurate and efficient information retrieval in QA systems.