Abstract: In some implementations, a device may identify a text from a document that includes structured information, wherein the structured information is a set of patent claims. The device may extract a first one or more parts and a second one or more parts, wherein the first one or more parts are one or more grammatical parts and the second one or more parts are one or more legal parts. The device may insert the first one or more parts and the second one or more parts into a structured model data object, wherein the structured model data object is a claim model data object. The device may generate structured output text from the structured model data object, wherein the structured output text corresponds to one or more patent document parts. The device may provide the structured output text.