arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2005.11223cs.IRcs.CL
收藏

L2R2:利用排名进行溯因推理

L2R2: Leveraging Ranking for Abductive Reasoning

CAS Key Lab of Network Data Science and Technology, Institute of Computing Technology, Chinese Academy of Sciences(中国科学院计算技术研究所网络数据科学与技术重点实验室);University of Chinese Academy of Sciences(中国科学院大学)

Yunchang Zhu, Liang Pang, Yanyan Lan, Xueqi Cheng

摘要:

溯因自然语言推理任务($α$ NLI)被提出来评估学习系统的溯因推理能力。在$α$ NLI任务中,给出两个观察结果,并要求从候选者中挑选出最合理的假设。现有的方法简单地将其表述为分类问题,因此在训练期间使用交叉熵对数损失目标。然而,区分真假并不能衡量一个假设的可行性,因为所有的假设都有可能发生,只是概率不同。为了填补这一空白,我们切换到一个排名的角度,排序的假设,他们的排名顺序。基于这一新的视角,在学习排序框架下提出了一种新的$L2R^2$方法。首先,训练样本被重新组织成一个排名形式,其中两个观察和他们的假设被视为查询和一组候选文件。然后,获得ESIM模型或预先训练的语言模型(例如,BERT或RoBERTa)作为评分函数。最后,用于排序任务的损失函数可以是成对的或列表式的以用于训练。ART数据集上的实验结果达到了公共排行榜的最新水平。

英文摘要:

The abductive natural language inference task ($α$NLI) is proposed to evaluate the abductive reasoning ability of a learning system. In the $α$NLI task, two observations are given and the most plausible hypothesis is asked to pick out from the candidates. Existing methods simply formulate it as a classification problem, thus a cross-entropy log-loss objective is used during training. However, discriminating true from false does not measure the plausibility of a hypothesis, for all the hypotheses have a chance to happen, only the probabilities are different. To fill this gap, we switch to a ranking perspective that sorts the hypotheses in order of their plausibilities. With this new perspective, a novel $L2R^2$ approach is proposed under the learning-to-rank framework. Firstly, training samples are reorganized into a ranking form, where two observations and their hypotheses are treated as the query and a set of candidate documents respectively. Then, an ESIM model or pre-trained language model, e.g. BERT or RoBERTa, is obtained as the scoring function. Finally, the loss functions for the ranking task can be either pair-wise or list-wise for training. The experimental results on the ART dataset reach the state-of-the-art in the public leaderboard.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑