Numeric Fused-Heads Identification and Resolution Dataset

数值融合头部识别与解析数据集是由巴伊兰大学计算机科学系和Allen人工智能研究所合作创建的，专注于处理英语文本中数值融合头部（NFH）现象。该数据集包含约10,000个示例，总计100万个令牌，用于训练和评估NFH的识别和解析模型。创建过程涉及从大规模英语文本语料库中提取NFH现象，并通过众包方式进行数据标注。数据集的应用领域包括文本蕴含、解决算术问题、数值推理和语言模型等，旨在提高自然语言处理系统对隐含信息的理解和处理能力。

The Numerical Fusion Head (NFH) Recognition and Parsing Dataset was collaboratively developed by the Department of Computer Science of Bar-Ilan University and the Allen Institute for AI, focusing on the NFH phenomenon in English texts. This dataset contains approximately 10,000 examples with a total of 1 million tokens, and is used for training and evaluating models for NFH recognition and parsing. The dataset construction process involves extracting NFH instances from large-scale English text corpora and performing data annotation via crowdsourcing. The application scenarios of this dataset include textual entailment, arithmetic problem solving, numerical reasoning, and language modeling, among others, aiming to enhance the ability of natural language processing systems to understand and process implicit information.

原始信息汇总

Numeric Fused-Head 数据集

概述

该数据集与论文 "Where’s My Head? Definition, Dataset and Models for Numeric Fused-Heads Identification and Resolution" 相关，作者为 Yanai Elazar 和 Yoav Goldberg，将发表于 TACL。

子任务

数据集涉及两个子任务：

• Identification：识别任务
• Resolution：解析任务

这两个子任务在论文的第3、4节和第5、6节中分别讨论，并在代码中独立处理。

示例用法

python import spacy from num_fh import NFH nlp = spacy.load(en_core_web_sm) nfh = NFH(nlp) nlp.add_pipe(nfh, first=False)

doc = nlp("I told you two, that only one of them is the one who will get 2 or 3 icecreams") assert doc[16]..is_nfh == True assert doc[18]..is_nfh == False assert doc[3]..is_deter_nfh == True assert doc[16]..is_deter_nfh == False assert len(doc._.nfh) == 4

引用

如果该工作与您的研究相关，请考虑引用：

@article{elazar_head, author = {Elazar, Yanai and Goldberg, Yoav}, title = {Where’s My Head? Definition, Data Set, and Models for Numeric Fused-Head Identification and Resolution}, journal = {Transactions of the Association for Computational Linguistics}, volume = {7}, number = {}, pages = {519-535}, year = {2019}, doi = {10.1162/tacl_a_00280}, URL = {https://doi.org/10.1162/tacl_a_00280}, }

搜集汇总

背景与挑战

背景概述

该数据集用于自然语言处理中的数字融合头识别和解析任务，专注于识别和解析文本中缺失的数字头部元素。它基于学术论文构建，包含代码和模型，支持spaCy集成，适用于机器学习和NLP研究。