Before you begin training your model, it is crucial to understand and outline the objectives you are trying to achieve by creating a model in Re:infer. It is likely that you will have multiple objectives, and you should think about how these would best align to how you then create taxonomies on your datasets.
You may be able to meet several objectives with one taxonomy, but remember that you can always create a separate taxonomy on a copy of the same dataset to suit another objective.
It’s best not to try and achieve absolutely everything at once within one sprawling multi-purpose taxonomy, as this can become very difficult to train and maintain. It is much easier to start with a more limited taxonomy for a specific purpose. For example, analysing in-app customer feedback data for product feature requests and product bugs, or monitoring client servicing in an operations team inbox.
It's very important that taxonomies are suitable for the data sources to which they are applied. When creating labels and structuring your taxonomy, you should make sure that they provide value by helping meet the objectives of the model. Guidance on structuring your taxonomy and creating appropriate labels can be found here.