DeepAI
Log In Sign Up

Do not let the history haunt you – Mitigating Compounding Errors in Conversational Question Answering

05/12/2020
by   Angrosh Mandya, et al.
0

The Conversational Question Answering (CoQA) task involves answering a sequence of inter-related conversational questions about a contextual paragraph. Although existing approaches employ human-written ground-truth answers for answering conversational questions at test time, in a realistic scenario, the CoQA model will not have any access to ground-truth answers for the previous questions, compelling the model to rely upon its own previously predicted answers for answering the subsequent questions. In this paper, we find that compounding errors occur when using previously predicted answers at test time, significantly lowering the performance of CoQA systems. To solve this problem, we propose a sampling strategy that dynamically selects between target answers and model predictions during training, thereby closely simulating the situation at test time. Further, we analyse the severity of this phenomena as a function of the question type, conversation length and domain type.

READ FULL TEXT

page 1

page 2

page 3

page 4

12/16/2021

Ditch the Gold Standard: Re-evaluating Conversational Question Answering

Conversational question answering (CQA) systems aim to provide natural-l...
11/17/2022

Open-Domain Conversational Question Answering with Historical Answers

Open-domain conversational question answering can be viewed as two tasks...
10/17/2022

Adversarial and Safely Scaled Question Generation

Question generation has recently gained a lot of research interest, espe...
11/26/2019

A Vietnamese Text-Based Conversational Agent

This paper introduces a Vietnamese text-based conversational agent archi...
08/24/2020

DiverseNet: When One Right Answer is not Enough

Many structured prediction tasks in machine vision have a collection of ...
03/18/2022

Simulating Bandit Learning from User Feedback for Extractive Question Answering

We study learning from user feedback for extractive question answering b...