Many NLP tasks require to automatically identify the most significant words in a text. In this work, we derive word significance from models trained to solve semantic task: Natural Language Inference and Paraphrase Identification. Using an attribution method aimed to explain the predictions of these models, we derive importance scores for each input token. We evaluate their relevance using a so-called cross-task evaluation: Analyzing the performance of one model on an input masked according to the other model's weight, we show that our method is robust with respect to the choice of the initial task. Additionally, we investigate the scores from the syntax point of view and observe interesting patterns, e.g. words closer to the root of a syntactic tree receive higher importance scores. Altogether, these observations suggest that our method can be used to identify important words in sentences without any explicit word importance labeling in training.

翻译：许多自然语言处理任务需要自动识别文本中最关键的词语。本研究从解决语义任务（自然语言推理与复述识别）的模型中推导词语重要性。通过采用旨在解释模型预测的归因方法，我们为每个输入词元推导出重要性分数。利用所谓的跨任务评估方法验证其相关性：分析一个模型在处理根据另一模型权重遮蔽的输入时的表现，结果表明我们的方法对初始任务的选择具有稳健性。此外，我们从语法角度研究分数并观察到有趣模式，例如距离句法树根节点更近的词获得更高的重要性分数。这些观察表明，无需在训练中使用显式的词语重要性标注，我们的方法即可用于识别句子中的重要词语。