# Rethinking the Objectives of Extractive Question Answering

arXiv: https://arxiv.org/abs/2008.12804 · HTML: https://mlsanity.com/paper/2008.12804/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 28 Aug 2020.

## Built on by (most cited first, 6 of 6)

- [NeurIPS 2020 EfficientQA Competition: Systems, Analyses and Lessons Learned](https://mlsanity.com/paper/2101.00133/) (2021)
- [R2-D2: A Modular Baseline for Open-Domain Question Answering](https://mlsanity.com/paper/2109.03502/) (2021)
- [Pruning the Index Contents for Memory Efficient Open-Domain QA](https://mlsanity.com/paper/2102.10697/) (2021)
- [Cascaded Span Extraction and Response Generation for Document-Grounded Dialog](https://mlsanity.com/paper/2106.07275/) (2021)
- [NAPG: Non-Autoregressive Program Generation for Hybrid Tabular-Textual Question Answering](https://mlsanity.com/paper/2211.03462/) (2022)
- [DyREx: Dynamic Query Representation for Extractive Question Answering](https://mlsanity.com/paper/2210.15048/) (2022)

## Abstract (arXiv)

This work demonstrates that using the objective with independence assumption for modelling the span probability P(as,ae) = P(as)P(ae) of span starting at position as and ending at position ae has adverse effects. Therefore we propose multiple approaches to modelling joint probability P(as,ae) directly. Among those, we propose a compound objective, composed from the joint probability while still keeping the objective with independence assumption as an auxiliary objective. We find that the compound objective is consistently superior or equal to other assumptions in exact match. Additionally, we identified common errors caused by the assumption of independence and manually checked the counterpart predictions, demonstrating the impact of the compound objective on the real examples. Our findings are supported via experiments with three extractive QA models (BIDAF, BERT, ALBERT) over six datasets and our code, individual results and manual analysis are available online.