Embodiments relate to systems and methods that retrieve, from the one or more computer storage memory, dialogue data associated with a plurality of utterances from two or more entities. A target concept in one or more dialogue segments of the transcribed dialogue data can be determined, where the one or more dialogue segments comprising a first dialogue segment associated with a first utterance by a first entity, and where the first dialogue segment includes the target concept. A dialogue goal can be determined for the first dialogue segment comprising the target concept. In addition, a structured link can be generated, associating the extracted concept to the dialogue goal.