BS ISO 21636-3:2024
语言编码 语言变体框架 框架的应用

Language coding. A framework for language varieties - Application of the framework


标准号
BS ISO 21636-3:2024
发布
2024年
发布单位
英国标准学会
当前最新
BS ISO 21636-3:2024
 
 
适用范围
1   Scope The ISO 21636 series provides a framework for the identification and description of varieties of all individual human languages (see ISO 639). It is applicable to sign languages. It does not apply to: — artificial means of communication with or between machines (such as programming languages); — those means of human communication which are neither fully nor largely equivalent to human language (such as sets of individual symbols or gestures that each carry isolated meanings but cannot be freely combined into complex expressions). This document gives guidance on how to apply the framework to identify basic dimensions and sub-dimensions of linguistic variation and the resulting varieties, including major modalities of human communication. It does not include any code or individual identifiers. This document is structured strictly analogously to ISO/TR 21636‑2. For a general description of the dimension and varieties dealt with in each clause, the user can refer to the corresponding clause in that document. This document focuses only on the identification and description of language varieties, not on the general, formal or technical aspects of the description of human language resources (LRs), which are covered by general metadata frameworks. NOTE 1 For the...

BS ISO 21636-3:2024相似标准


推荐

Nature | 谷歌推出医学“Chatgpt”,基于大语言模型AI医生开始临床测试

为了评估LLM对临床知识编码程度并评估其在医学中潜力,研究者策划了MultiMedQA,这是一个由七个医学问答数据集组成基准,包括六个现有数据集:MedQA、MedMCQA、PubMedQA、LiveQA、MedicationQA和MMLU临床主题。为了使用MultiMedQA评估LLM,研究者建立在PaLM、5400亿参数(540B)LLM及其指令调优变体Flan-PaLM基础上。...

自动化所提出三大“人机交互”新模型

最近,自动化所自然语言处理团队构建了一种同步双向序列生成模型(SBSG),不仅刷新了计算机“说话”极限,也挑战了人类说话极限:同时顺着说和倒着说,到中间点结束。该算法可应用于机器翻译和自动摘要,不仅显著提升了文本生成效率,还改善了生成文本质量。  当前,基于编码器-解码器结构序列生成模型广泛应用于自然语言生成任务,例如神经机器翻译、摘要生成等。...

量子计算机与超级计算机“协同学习”框架发布

据安徽省量子计算工程研究中心副主任窦猛汉介绍,自动驾驶研究者可使用新框架,开发在自动驾驶领域新算法,加快训练速度;自然语言处理研究者也可用其开发新算法,提高算法对语言理解能力。新框架使中国量子计算机和经典计算机“协同学习”能力再进一步,将有力推动各行业经典算力和量子新算力协同应用落地。...

大模型向人类思维方式“对齐”将颠覆各行业生态

“首先是对语言模型基座进行架构优化,这面临输入长度、中文编码、多模态接入等挑战;其次是指令微调,其难度远高于预训练,而让模型回答‘对齐’人类思维习惯也非常困难;最后才是能力强化。”邱锡鹏说,“当语言模型向人类价值观和思维方式对齐,并应用于真实世界,无疑将颠覆各个行业生态。 ”...





Copyright ©2007-2022 ANTPEDIA, All Rights Reserved
京ICP备07018254号 京公网安备1101085018 电信与信息服务业务经营许可证:京ICP证110310号