If you want to define your own custom logic of what terms are to be considered possible duplicates and then to plug this custom logic into all Term Merge machinery, then this page is for you. Duplicate detection heuristics are implemented through cTools plugins. If you have not worked with cTools plugins before, we highly recommend you to read documentation about them, such as this one. Term Merge module defines the cTools plugin type of duplicate_suggestion. This plugin type contains the following properties: Having that said, core of your duplicate suggestion plugin should be the hash function. You can look at the examples of hash function inside of Term Merge module, just study the term_merge/plugins/duplicate_suggestion/*.inc files. In fact your hash function does not necessarily have to be complicated to yield reasonable duplicate suggestions. Good luck coding! And if you feel like you have coded a duplicate suggestion that others might benefit from, please, be kind to open an issue against Term Merge module and share your duplicate suggestion there. I thank you beforehand on behalf of all the community!