QA-SRL

QA-SRL 被提议作为语义角色的开放模式，其中论点和谓词之间的关系表示为包含谓词（“某人在哪里受过教育？”）的自然语言问题，其答案是论点（“普林斯顿”）。作者从 3,200 个句子中收集了大约 19,000 个问答对。QA-SRL was proposed as an open paradigm for semantic roles, where the relationship between arguments and predicates is represented as natural language questions containing the predicate (e.g., "Where was someone educated?"), with the answer being the corresponding argument (e.g., "Princeton"). The authors collected approximately 19,000 question-answer pairs from 3,200 sentences.

搜集汇总

背景与挑战

背景概述

QA-SRL是一种语义角色标注的开放模式，通过自然语言问题与答案来捕捉论点和谓词之间的关系。该数据集基于3,200个句子，收集了约19,000个问答对，用于支持自然语言处理任务。