mForms : Multimodal Form-Filling with Question Answering
This work provides a zero-shot, data-efficient method for multimodal form-filling, which is significant for developers and users interacting with digital forms, especially in scenarios with limited training data.
This paper redefines form-filling as a multimodal Question Answering (QA) task, translating GUI elements into natural language questions and using a pre-trained extractive QA system for zero-shot form completion. The approach achieves a state-of-the-art F1 score of 0.97 on the ATIS dataset using approximately 1/10th of the training data.
This paper presents a new approach to form-filling by reformulating the task as multimodal natural language Question Answering (QA). The reformulation is achieved by first translating the elements on the GUI form (text fields, buttons, icons, etc.) to natural language questions, where these questions capture the element's multimodal semantics. After a match is determined between the form element (Question) and the user utterance (Answer), the form element is filled through a pre-trained extractive QA system. By leveraging pre-trained QA models and not requiring form-specific training, this approach to form-filling is zero-shot. The paper also presents an approach to further refine the form-filling by using multi-task training to incorporate a potentially large number of successive tasks. Finally, the paper introduces a multimodal natural language form-filling dataset Multimodal Forms (mForms), as well as a multimodal extension of the popular ATIS dataset to support future research and experimentation. Results show the new approach not only maintains robust accuracy for sparse training conditions but achieves state-of-the-art F1 of 0.97 on ATIS with approximately 1/10th of the training data.