Towards End-to-End Open Conversational Machine Reading
This work addresses error propagation and optimization issues in conversational AI for multi-turn question answering, representing an incremental advance by unifying sub-tasks into a single model.
The paper tackled the open-retrieval conversational machine reading task by proposing a unified end-to-end text-to-text framework to replace existing cascaded methods, achieving new state-of-the-art results on the ShARC and OR-ShARC datasets with significant performance improvements.
In open-retrieval conversational machine reading (OR-CMR) task, machines are required to do multi-turn question answering given dialogue history and a textual knowledge base. Existing works generally utilize two independent modules to approach this problem's two successive sub-tasks: first with a hard-label decision making and second with a question generation aided by various entailment reasoning methods. Such usual cascaded modeling is vulnerable to error propagation and prevents the two sub-tasks from being consistently optimized. In this work, we instead model OR-CMR as a unified text-to-text task in a fully end-to-end style. Experiments on the ShARC and OR-ShARC dataset show the effectiveness of our proposed end-to-end framework on both sub-tasks by a large margin, achieving new state-of-the-art results. Further ablation studies support that our framework can generalize to different backbone models.