Create a text clustering algorithm that identifies both broad categories and subtle subcategories within document collections. Include methods for determining optimal cluster numbers and handling overlapping concepts. All text in it should be assigned to appropriate levels within the hierarchy.