Doc2Dial: a Framework for Dialogue Composition Grounded in Business Documents

Song Feng, Kshitij Fadni, Q. Vera Liao, Luis A. Lastras

Keywords: business documents, dialog, data set

Abstract: We introduce Doc2Dial, an end-to-end framework for generating conversational data grounded in business documents via crowdsourcing. Such data can be used to train automated dialogue agents performing customer care tasks for the enterprises or organizations. In particular, the framework takes the documents as input and generates the tasks for obtaining the annotations for simulating dialog flows. The dialog flows are used to guide the collection of utterances produced by crowd workers. The outcomes include dialogue data grounded in the given documents, as well as various types of annotations that help ensure the quality of the data and the flexibility to (re)composite dialogues.

TL;DR: We introduce Doc2Dial, an end-to-end framework for generating conversational data grounded in business documents via crowdsourcing for train automated dialogue agents