Parallel Meaning Bank

在格罗宁根大学开发并建立在格罗宁根意义银行基础上的平行意义银行 (PMB) 包括原始和标记格式的句子和文本，句法分析，词义，主题角色，参考分辨率和形式含义表示。PMB的主要目的是为单词，句子和文本提供细粒度的含义表示。隔离，句子往往模棱两可。目的是为句子提供最可能的解释，而最少使用不足的说明。 PMB注释包括完全手动校正的金标准数据，以及银 (部分手动校正) 和青铜 (没有手动校正) 数据。到目前为止，这些版本包含英语，德语，意大利语和荷兰语的文档，但对于将来的版本，计划包括中文和日语。The Parallel Meaning Bank (PMB), developed at the University of Groningen and built upon the Groningen Meaning Bank, includes sentences and texts in both raw and annotated formats, along with syntactic parses, word senses, thematic roles, coreference resolution, and formal meaning representations. The primary goal of PMB is to provide fine-grained meaning representations for words, sentences, and texts. When taken in isolation, sentences are often ambiguous; the aim is to offer the most plausible interpretations for sentences while minimizing under-specification. PMB annotations include fully manually curated gold-standard data, as well as silver (partially manually corrected) and bronze (no manual correction) data. To date, these releases cover documents in English, German, Italian, and Dutch, while future releases are planned to include Chinese and Japanese.

搜集汇总

背景与挑战

背景概述

Parallel Meaning Bank是由格罗宁根大学开发的多语言语义标注数据集，基于格罗宁根意义银行构建，包含英语、德语、意大利语和荷兰语的句子与文本，提供词义、句法分析等细粒度含义表示。数据分为金、银、青铜三个质量等级，未来计划扩展至中文和日语。