Today, more people use social networking platforms to convey their desires and recent needs. Actually, there are numerous daily posts carrying commercial intention. The detection of these kinds of user intention would be quite valuable, especially for the platform itself. Firstly, it could help the platform provide precise and instant recommendations to users for its own business interests. Secondly, intention mining works may help link users' needs by detecting potential buyers and sellers and their specific intentions which can benefit users by optimizing the resources in their hand and increase functional richness. The whole intention mining process generally includes three main stages: user commercial intention filtering, intention domain identification and specific intention words extraction. In this work, the first stage was simplified using keywords-based automatic filter followed by a manual screening. The main focus of this paper is the second stage, assigning the intention-holding posts into their own single domain. Three machine learning models and two deep learning models were proposed to solve this text classification problem. The proposed methods have been evaluated on a dataset containing 5500 real-time intention-holding tweets collected from Twitter. In general, the experimental results showed impressive performance with the highest classification accuracy of 96% achieved by Long short-term memory.