LiveQA: A Question Answering Dataset over Sports Live

TL;DR

LiveQA is a large, challenging dataset for question answering based on live sports broadcasts, requiring timeline understanding, event tracking, and calculations, with models currently performing poorly.

Contribution

We introduce LiveQA, a novel dataset from live NBA broadcasts that tests reasoning over timeline-based sports events, filling a gap in existing QA datasets.

Findings

Baseline models achieve only 53.1% accuracy.

The dataset challenges current question answering models.

LiveQA enables future research in reasoning over live sports data.

Abstract

In this paper, we introduce LiveQA, a new question answering dataset constructed from play-by-play live broadcast. It contains 117k multiple-choice questions written by human commentators for over 1,670 NBA games, which are collected from the Chinese Hupu (https://nba.hupu.com/games) website. Derived from the characteristics of sports games, LiveQA can potentially test the reasoning ability across timeline-based live broadcasts, which is challenging compared to the existing datasets. In LiveQA, the questions require understanding the timeline, tracking events or doing mathematical computations. Our preliminary experiments show that the dataset introduces a challenging problem for question answering models, and a strong baseline model only achieves the accuracy of 53.1\% and cannot beat the dominant option rule. We release the code and data of this paper for future research.