关于知识产权 知识产权培训 知识产权外联 部门知识产权 知识产权和热点议题 特定领域知识产权 专利和技术信息 商标信息 工业品外观设计信息 地理标志信息 植物品种信息(UPOV) 知识产权法律、条约和判决 知识产权资源 知识产权报告 专利保护 商标保护 工业品外观设计保护 地理标志保护 植物品种保护(UPOV) 知识产权争议解决 知识产权局业务解决方案 知识产权服务缴费 谈判与决策 发展合作 创新支持 公私伙伴关系 组织简介 与产权组织合作 问责制 专利 商标 工业品外观设计 地理标志 版权 商业秘密 WIPO学院 讲习班和研讨会 世界知识产权日 WIPO杂志 宣传 案例研究和成功故事 知识产权新闻 产权组织奖 企业 高校 土著人民 司法机构 遗传资源、传统知识和传统文化表现形式 经济学 性别平等 全球卫生 气候变化 竞争政策 可持续发展目标 执法 前沿技术 移动应用 体育 旅游 PATENTSCOPE 专利分析 国际专利分类 ARDI - 研究促进创新 ASPI - 专业化专利信息 全球品牌数据库 马德里监视器 Article 6ter Express数据库 尼斯分类 维也纳分类 全球外观设计数据库 国际外观设计公报 Hague Express数据库 洛迦诺分类 Lisbon Express数据库 全球品牌数据库地理标志信息 PLUTO植物品种数据库 GENIE数据库 产权组织管理的条约 WIPO Lex - 知识产权法律、条约和判决 产权组织标准 知识产权统计 WIPO Pearl(术语) 产权组织出版物 国家知识产权概况 产权组织知识中心 产权组织技术趋势 全球创新指数 世界知识产权报告 PCT - 国际专利体系 ePCT 布达佩斯 - 国际微生物保藏体系 马德里 - 国际商标体系 eMadrid 第六条之三(徽章、旗帜、国徽) 海牙 - 国际外观设计体系 eHague 里斯本 - 国际地理标志体系 eLisbon UPOV PRISMA 调解 仲裁 专家裁决 域名争议 检索和审查集中式接入(CASE) 数字查询服务(DAS) WIPO Pay 产权组织往来账户 产权组织各大会 常设委员会 会议日历 产权组织正式文件 发展议程 技术援助 知识产权培训机构 COVID-19支持 国家知识产权战略 政策和立法咨询 合作枢纽 技术与创新支持中心(TISC) 技术转移 发明人援助计划(IAP) WIPO GREEN 产权组织的PAT-INFORMED 无障碍图书联合会 产权组织服务创作者 WIPO ALERT 成员国 观察员 总干事 部门活动 驻外办事处 职位空缺 采购 成果和预算 财务报告 监督

知识产权局人工智能举措指数

国家/地区 机构名称业务应用说明
中国State Administration for Industry and Commerce (SAIC)专利分类

For invention and utility model patents, CNIPA has developed IPC automatic classification system, which carries out batch pre classification for newly applied patents. The automatic classification system could give precise results in subclass level. For design patents, CNIPA has also developed LOC automatic classification system based on text information, which can give precise results in subclass level.

中国State Administration for Industry and Commerce (SAIC)机器翻译

At present, CNIPA has translated part of foreign patent data into Chinese by using machine translation technology, so that the examiners can search and browse foreign patent data in Chinese.

中国State Administration for Industry and Commerce (SAIC)专利现有技术检索

In 2021, CNIPA launched the new intelligent search system and the intelligent semantic search technology was applied in the system. The intelligent search function mainly provides 2 search modes. The first one is the automatic search mode. When the examiner clicks on the "Semantic Search" button,the system will push similar documents, which are sorted by similarity.

The second search mode is semantic sort to Boolean search results. The examiners could do Boolean search and get the results in the first step, and then conduct the semantic sort based on the results. This may help the examiner to view the closest documents, and improve the efficiency of finding reference documents.

中国State Administration for Industry and Commerce (SAIC)商标分类(商品和服务)

中国国家工商行政管理总局(工商总局)利用“标准商品系统”把商品分入近似的小组,以建立“商品关系辞典”。有了这个辞典,系统自动将新提供的商品分入相应的近似组。对于第一次提供的商品,将指定一项上一级商品。

中国State Administration for Industry and Commerce (SAIC)图形检索(商标、外观设计)

2018年初,中国工商总局在开发一套能够提供相对准确可靠结果的图形检索系统。 该系统可以检索已有图形要素,经审查员确认后,结果将会被输入系统。 如此一来,系统就可以实现自主创新和自主学习,提高检索效率。

中国State Administration for Industry and Commerce (SAIC)数据分析

中国工商总局使用“行政区域自动匹配系统”锁定行政区域,以便为未来的区域统计分析提供数据支持。

乌拉圭National Directorate of Industrial Property帮助台服务

国家总局使用内部开发的通知系统,并与其在线申请系统相连接。2018年初,国家总局在开发一种更复杂的通知算法,目的是确认特定用户不再使用或已有一段时间未使用该系统。在这种情况下,新算法会触发增发通知。

新算法将发送电子邮件、测试通知、反馈调查和更新个人信息的请求,以确保与用户保持沟通,并在截止日期之前捕获数据变更信息。

俄罗斯联邦Federal Service for Intellectual Property (Rospatent) / Federal Institute of Industrial Property (FIPS)商标分类(商品和服务)

目前,一种检索商标、地理标志和原产地名称的新检索引擎正在开发中,将于2020年夏季投入使用。新系统运用神经网络进行图像相似性检索及商标词语智能识别(涉及术语的语义相似性)。此外,系统功能还包括商标(图像)索引信息识别,即商标词语智能识别及根据维也纳分类自动分类。现阶段,正在测试图像检索的技术解决方案。


相关链接

俄罗斯联邦Federal Service for Intellectual Property (Rospatent) / Federal Institute of Industrial Property (FIPS)机器翻译

2015年,在PatSearch系统开发框架内推出了实施专利文献机器翻译的倡议。翻译功能的实现得益于采用了俄罗斯PROMPT公司开发的机器混合翻译系统。该系统涵盖全面的语言分析方法。现已通过运用俄英双语专利文献平行文本的机器学习方法,创建了神经网络。该系统除翻译专利文献外,还能够将CPC译成俄文。


相关链接

俄罗斯联邦Federal Service for Intellectual Property (Rospatent) / Federal Institute of Industrial Property (FIPS)专利现有技术检索

实施人工智能检索的倡议于2017年推出,并于2018年开始实施,涉及了发明和实用新型申请审查。PatSearch系统所实施的专利文献相似性检索功能的实现,源于采用了一组人工智能方法和技术,并结合了全球信息检索最佳做法。目前,相似性检索功能正在俄罗斯专利文献数据库中运行。在建立俄罗斯分布式同义词库过程中,采用了一种针对俄罗斯专利文献的神经网络,其中所涉的标准兼顾了审查程序的复杂性和专家经验。


相关链接

全世界World Intellectual Property Organization (WIPO)图形检索(商标、外观设计)

全球品牌数据库中的图片检索让商标所有人能够从数百万计的图片中找出视觉上相似的商标和其他品牌信息记录。


相关链接

全世界World Intellectual Property Organization (WIPO)机器翻译

WIPO Translate是全球领先的即时翻译工具,专门用于专利文献翻译。 它可以通过 PATENTSCOPE数据库访问,也可以应要求纳入知识产权局的系统。


相关链接

全世界World Intellectual Property Organization (WIPO)专利分类

专利自动分类系统(IPCCAT)帮助专利申请人和知识产权局审查员依照国际专利分类(IPC)大类、小类或大组,把专利申请自动分类归入相应技术部门。


相关链接

加拿大Canadian Intellectual Property Office (CIPO)帮助台服务

2018年初,加拿大知识产权局在探索如何利用国际商用机器公司(IBM)Watson工具套件通过社交媒体宣传和分析功能与客户进行互动。

加拿大Canadian Intellectual Property Office (CIPO)专利现有技术检索

加拿大知识产权局的专利部门使用人工智能语义检索引擎的商业服务(Questel、STN和Clarivate Analytics)帮助开展现有技术和援引方面的检索。这些工具依赖机器学习算法,更好地查找出援引、申请和现有技术之间的联系。

专利审查员还利用谷歌的算法,特别是“翻译”、“专利”和“学术搜索”工具中的算法用于机器翻译,并实时获取国际上参与的专利局所提供的文献全文和权利要求书表格以及援引情况统计和相关学术出版物。

对于数据操作,加拿大局使用Vantage Point文本挖掘工具从专利和文献数据库的检索结果中发现知识,同时提供优化、自动化、导入等处理所得原始数据的方法。

加拿大Canadian Intellectual Property Office (CIPO)数据分析

加拿大知识产权局的经济研究与战略分析股使用人工智能帮助进行语义检索并收集、清理和分析大型数据集。

在正在开展的经济研究背景下,加拿大局计划在2018年初探索利用机器学习解决知识产权政策与研究问题的可行性。

加拿大Canadian Intellectual Property Office (CIPO)商标分类(商品和服务)

In April 2022, CIPO began issuing pre-assessment letters on all national, unexamined trademark applications that inform applicants of the results of the automated analysis of goods and services, as well as the advantage of submitting an amended application using the pre-approved list of goods or services.

The letter provides information that the application contains either:

• acceptable goods or services;

• goods or services not classed;

• an improper Nice class;

• unacceptable goods or services, including terms retired from the pre-approved list;

or a combination of the above.

加拿大Canadian Intellectual Property Office (CIPO)帮助台服务

CIPO uses a chatbot called “The ISED Business Assistant” which provides information on IP to users who visit the CIPO website. It is available 24/7. CIPO is also analysing the potential implementation of a floating chatbot on CIPO’s most visited web pages as well as a website crawling feature (when the chatbot fails to provide an answer).

塞尔维亚Intellectual Property Office of the Republic of Serbia数字化与过程自动化

塞尔维亚知识产权局使用产权组织专利OCR校对平台。产权组织 OCR平台可能能够利用机器学习改进OCR校对。由于ABBYYOCR中的本地语言资源有限(字典和语法规则不足),机器训练对 OCR校对质量的影响仍然很小。根据塞尔维亚知识产权局的经验,导致 OCR正确性降低的主要问题是在文档中使用多种文字(塞尔维亚西里尔文、塞尔维亚拉丁文、英文、化学和数学公式)。

2018年初,塞尔维亚知识产权局计划在手动 OCR校对(由产权组织提供)中利用机器学习扩充词典,并为塞尔维亚语专利文档制定特别处理规则。

塞尔维亚Intellectual Property Office of the Republic of Serbia机器翻译

在欧洲专利局的专利机器翻译项目范围内,塞尔维亚知识产权局为机器翻译学习这一具体目的提供了专利说明书全文双语文件(塞尔维亚语 /英语)语料库。2018年初,Espacenet数据库中的塞尔维亚语机器翻译工具尚未产生令人满意的结果。

大韩民国Korean Intellectual Property Office (KIPO)专利现有技术检索

2017年4月,韩国特许厅与韩国电子和电信研究所(ETRI)建立了人工智能协议,共同努力建立用于人工智能学习的专利知识库,并合作研究如何将他们开发的人工智能系统应用于知识产权行政管理工作。

智能专利检索的试点模型正在开发,计划于2019年完成。该模型通过从关键字检索转向语法和语义检索系统来提高现有技术检索的质量。


相关链接

大韩民国Korean Intellectual Property Office (KIPO)帮助台服务

2018年初,韩国特许厅计划在未来三年内开发和完善基于文本和语音识别的人工智能专利客户服务系统试点模型。


相关链接

大韩民国Korean Intellectual Property Office (KIPO)机器翻译

韩国特许厅利用国际专利分类(IPC)H部的专利公告数据建立了一个用于机器学习的数据库。该数据库将编入10万项专利技术术语条目和100万条专利语言分析和附图标记信息。


相关链接

奥地利The Austrian Patent Office专利现有技术检索

The Austrian Patent Office (APO) is committed to the responsible and trustworthy use of AI in line with the national government's official strategic plan “Artificial Intelligence Mission Austria 2030” (https://www.ki-strategie.at/) and the European Union Artificial Intelligence Act (https://artificialintelligenceact.eu).

Since early 2018, APO has been actively evaluating the state of the art of advanced tools for intellectual property search and management. In this context, APO has conducted extensive trials of various commercial providers’ semantic and artificial intelligence enhanced search tools (IPRally, IPscreener, Predori, Intergator, Orbit Intelligence, Minesoft). The evaluated tools show promising results and interesting capabilities, the implementation within internal processes will still take some time, though.

Some notable use-cases for APO are AI-assisted patent pre-classification for internal routing of new applications, and pre-search for the preparatory stages of prior art search. APO plans further exploration of AI applications in areas such as trademark image search.

奥地利The Austrian Patent Office专利分类

The Austrian Patent Office (APO) is committed to the responsible and trustworthy use of AI in line with the national government's official strategic plan “Artificial Intelligence Mission Austria 2030” (https://www.ki-strategie.at/) and the European Union Artificial Intelligence Act (https://artificialintelligenceact.eu).

Since early 2018, APO has been actively evaluating the state of the art of advanced tools for intellectual property search and management. In this context, APO has conducted extensive trials of various commercial providers’ semantic and artificial intelligence enhanced search tools (IPRally, IPscreener, Predori, Intergator, Orbit Intelligence, Minesoft). The evaluated tools show promising results and interesting capabilities, the implementation within internal processes will still take some time, though.

Some notable use-cases for APO are AI-assisted patent pre-classification for internal routing of new applications, and pre-search for the preparatory stages of prior art search. APO plans further exploration of AI applications in areas such as trademark image search.(Updated November 2024)

巴西National Institute of Industrial Property (INPI)专利分类

在2018年初,巴西国家工业产权局在开展一项计划,根据国际专利分类和/或合作专利分类(CPC)开发用于专利申请内部自动预先分类的神经网络,以便之后在技术部门中分配申请。巴西国家局考虑将数学实验室作为开展此计划的最佳工具。

德国German Patent and Trade Mark Office (DPMA)数字化与过程自动化

德国专利商标局的许多商标申请完全通过自动化的方式进行分类。

德国German Patent and Trade Mark Office (DPMA)专利现有技术检索

2016年,德国专利商标局启动了一个项目,对本局不同数据来源(如电子文件、专家数据库等)的现有技术检索实施集中式服务。集中式服务利用算法来改进相似文本的检索。

德国German Patent and Trade Mark Office (DPMA)专利分类

2011年,德国专利商标局推出了国际专利分类(IPC)自动化工具,作为电子专利和实用新型管理系统的一部分。这个分类器基于启发式算法,对进入系统的专利申请进行初步的IPC分类,从而帮助将申请分派给适当的专利审查员。这种暗箱式的工具主要缺点是不灵活,不能通过参数表达,因此不能用于德国专利商标局所有的使用情况。

于是,开发了一种新的专利分类自动化工具,作为新专利检索系统的一部分。在对不同技术进行评价后,选择了基于带有“分散词语表达”的神经网络的方法。第一步是使用“上限预测”和“三项猜测”的质量措施进行IPC小类一级的自动分类分析。使用不同的训练数据集进行实验,这些数据集由2010-2015年公布的德国专利申请、获得授权的专利和实用新型经过挑选汇集而成。

分类器的落实包括一项管道式机制,包含数据准备、训练和评价。每一步都可以配置参数并查看部分结果,从而使整个分类过程灵活透明。分类器在IPC分类上具有升级空间,而且训练过程中体现出的绩效尚可接受。它可以迅速地对未知文本进行在线分类。

德国专利商标局的新分类器可以投入使用的范围包括:对新进入系统的专利申请进行自动预先分类(以改善专利审查员中的专利申请分配);交互式分类(在既定的IPC级别作出若干预测,协助专利审查员的工作);重新分类(支持新版IPC的使用);持续改进现有技术专利文献IPC的质量。基于分类器建立的网络服务将就专利文献的既定部分(如摘要、权利要求书或说明书)作出即时的IPC预测。

挪威Norwegian Industrial Property Office (NIPO)Generative AI initiatives

NIPO uses a commercially available tool: Acsepto for trademark, by Coexya SA, Paris, for trademark image search. The search results (hit list) are prioritized based on AI-assisted search on image property coding. It is trained on images/logos, coding and previous selection by examiners from a number of IP offices.(Updated October 2024)

挪威Norwegian Industrial Property Office (NIPO)图形检索(商标、外观设计)

In 2024 NIPO initiated a project to explore value in use of Generative AI where own data is utilized in a RAG solution. The project aim to provide capacity and competence, an infrastructure and capacity for safe exploration of cases, and hence a playground/sandbox for future projects. In end of 2024 we have a chatbot on subset of register (DS, PT, TM), all regulations, rules, routines, leading documents and trademark decisions. The solution is also tested on trademark list of goods and service comparisons. (Updated October 2024)

捷克共和国Industrial Property Office of the Czech Republic (IPO CZ)专利审查管理

Starting from 2024, the Industrial Property Office of the Czech Republic plans to launch internal AI examination support tool in pilot phase, which will help patent examiners with the pre-classification of patent applications.

捷克共和国Industrial Property Office of the Czech Republic (IPO CZ)帮助台服务

The Industrial Property Office of the Czech Republic is currently working on creation of automated IP helpdesk which will be nonstop available to the users. The idea is to start with the provision of IP related advice in general. Later the Office wishes to enlarge this service and provide also the information dedicated to procedures of different IP applications. In this context, the Office would like to take advantage of the cooperation with Czech universities and introduce a chatbot to improve the helpdesk service. For enhancing this service, the Office will also analyze the use of voice recognition.

捷克共和国Industrial Property Office of the Czech Republic (IPO CZ)专利分类
图形检索(商标、外观设计)

In terms of the introduction of automated search and classification system, the Industrial Property Office of the Czech Republic has run the proof of concept. It confirmed that such a project is helpful and needed. The preparatory phase of this project has been finished and the development work will be finished by the end of 2023. The service should be ready for public use starting from 2024.

摩洛哥Moroccan Industrial and Commercial Property Office (OMPIC)专利现有技术检索

自2011年以来,摩洛哥知识产权局使用人工智能专利分析商业工具Orbite Intelligence,通过技术领域或关键字检索全球专利申请。推出这项基于地图的工具,是为满足摩洛哥技术与创新支持中心网络的需求,检索现有技术和专利先例。

摩洛哥Moroccan Industrial and Commercial Property Office (OMPIC)数据分析

摩洛哥知识产权局使用Qlikview系统管理来自各种数据库的大数据,无论这些数据存储在何处,该局创建了用于报告和质量控制的统计数据库。该解决方案即时生成信息、压缩数据并存储,从而确保可供多个用户即时搜索,而不受预定义的层次结构路线或预配置的控制面板限制。这一解决方案很好地解决了该局及其客户的需求。它可靠并且易于使用,自动生成不同的控制面板,并以图形或表格形式呈现。这一工具用于生成工业产权统计指标,为公众服务,可通过以下链接获取:www.barometreompic.ma。

摩洛哥Moroccan Industrial and Commercial Property Office (OMPIC)数字化与过程自动化

摩洛哥知识产权局采用ABBYY公司的人工智能辅助光学字符识别(OCR)技术将文件图像转换为机器编码文本。该技术从PDF文件中获取信息,并按照明确定义的结构(模板)将其导入该局数据库。然后运用审查规则确保准确性,并对错误数据进行视频编码。OCR技术减少了提取该局所管理数据时的延迟,并降低了手动输入逾100万份文件的成本。这一积极经验也扩展到专利文件的处理工作。

新加坡Intellectual Property Office of Singapore (IPOS)专利审查管理

In early 2018 IPOS was exploring the feasibility of implementing a patents auto checker that uses Natural Language Processing (NLP) and other machine learning technologies to perform formalities check automatically.

新加坡Intellectual Property Office of Singapore (IPOS)专利分类

From 2019, IPOS was exploring the feasibility of implementing a patents auto classification tool that uses Natural Language Processing to understand patent documents and aid the examiners in classifying incoming patent applications.

新加坡Intellectual Property Office of Singapore (IPOS)图形检索(商标、外观设计)

IPOS has implemented a commercial AI-powered image-based search solution on both our e-services web portal and our mobile app (IPOS GO). The solution enables the public and examiners to efficiently search for both visually similar trademarks as well as the conceptually similar trademarks.

日本Japan Patent Office商标分类(商品和服务)

The JPO is validating its systems to verify possible uses for AI to assign trademark classifications of designated goods and services. Using reference materials, such as the Examination Guidelines for Similar Goods and Services (which include many examples of specific goods and/or services and their appropriate similar-group codes), the JPO is testing functions to assign tentative similar-group codes to unclear designated goods and services in trademark applications.

(Updated October 2024)


相关链接

日本Japan Patent Office图形检索(商标、外观设计)

The JPO is validating its systems to verify possible uses for AI to conduct prior searches of figurative trademarks. Using past results in prior searches of figurative trademarks, the JPO is validating functions to retrieve prior figurative trademarks by inputting image data of claimed figurative trademarks, which might be identical with, or similar to, the claimed trademarks.

The JPO is also validating its systems to verify possible uses for AI to conduct prior design searches.

(Updated October 2024)


相关链接

日本Japan Patent Office专利分类

The JPO is validating its systems to verify possible uses for AI to assign patent classifications. Using text data of already filed documents to which patent classifications were assigned, the JPO is verifying a function to assign patent classifications (FI/F-terms) to foreign patent document.

(Updated October 2024)


相关链接

日本Japan Patent Office专利审查管理

The JPO is considering possible uses for AI to implement examination management tasks such as appropriate distribution of applications effectively and efficiently.

(Updated October 2024)


相关链接

日本Japan Patent Office专利现有技术检索

The JPO is validating its systems to verify possible uses for AI to support conducting prior art searches. (1) Using text data of examined patent documents and the retrieval history of search queries used in the examinations, the JPO is validating a function to suggest keywords and patent classifications that should be included in search queries. (2) Using image data of already filed documents, the JPO is validating functions to retrieve (i) images similar to designated patent image and (ii) images which are of specific types (like flowchart or circuit diagram). (3) Using past results in prior searches of patent, the JPO is validating a function to sort retrieved documents so that examiners can check the most related document first.

(Updated October 2024)


相关链接

智利National Institute of Industrial Property (INAPI)图形检索(商标、外观设计)

2018年初,智利国家工业产权局与智利大学工程学院在工程学院开发的算法基础上,合作开发了图形检索系统。

系统使用智利国家工业产权局的图形数据库进行培训,由商标审查员作出评价。

欧洲European Patent Office (EPO)专利审查管理

欧洲专利局(欧专局)一直积极致力开发利用机器学习和人工智能的业务解决方案,以便在不同实施程度管理专利文件:自动生成查询;对专利文献自动注释;自动检测专利文献中的问题或解决方案;自动检测对可专利性予以排除。

欧专局已经开发了一个专利文档模型(PDM),并在知识与信息管理环境(KIME)中予以实施。它们共同实现了对用于机器学习的专利和其他数据的强化管理。


相关链接

欧洲European Patent Office (EPO)机器翻译

欧专局在机器翻译方面使用“专利翻译”工具,但是也在开发自己的机器学习翻译。欧专局在其专利数据库中向公众提供这一工具,瑞典专利注册局和联合王国知识产权局经过专门培训的专利审查员也使用这一工具。


相关链接

欧洲European Patent Office (EPO)数据分析

欧专局借助其DataScience团队,主要基于适用的开源软件库,开发自己的人工智能系统。欧专局凭借其审查员和数据集(即历史保存的检索数据和该局现有技术语料库),将DataScience团队的专业知识与对业务的理解相结合。

欧专局一直积极开展确认其他技术领域中特定技术(计算机实施的发明)的迁移/渗透趋势的工作。


相关链接

欧洲European Patent Office (EPO)专利分类

欧专局已经开发利用机器学习和人工智能的业务解决方案,用于不同实施程度的专利分类:对收到的专利申请进行自动预分类,以分配给负责检索和审查的相应单位;根据合作专利分类(CPC)方案对专利文档进行自动分类;根据 CPC方案中的变更对专利文档进行自动重新分类。


相关链接

欧洲European Patent Office (EPO)专利现有技术检索

欧专局一直积极致力开发利用机器学习和人工智能的业务解决方案,以便在不同实施程度进行专利检索:自动检索收到的专利申请的现有技术;自动生成查询。

欧专局已经生成了自己的参考数据(黄金标准)和用于衡量自动检索工具表现的系统。

欧专局还通过不同项目的软件提供商,对自动注释领域中的商业产品加以利用。


相关链接

欧洲European Patent Office (EPO)图形检索(商标、外观设计)

欧专局一直积极致力开发利用机器学习和人工智能的业务解决方案,以便在不同实施程度进行专利检索,包括对专利附图进行自动图形和图片检索。


相关链接

欧洲联盟European Union Intellectual Property Office (EUIPO)商标分类(商品和服务)

EUIPO has developed AI based tools to extract relevant information from letters and make decisions based on this information. EUIPO has applied this technique to analyse Classification, formalities and AG deficiencies in trademark applications and to analyse the deficiency rate and grounds in Design applications.

欧洲联盟European Union Intellectual Property Office (EUIPO)机器翻译

EUIPO is making use of machine translation for Case Law documents through eSearch Case Law. The Office provides automatic translations in the website for EUIPO decisions. This allows the user to grasp the main idea of the content of the decision.


相关链接

欧洲联盟European Union Intellectual Property Office (EUIPO)商标分类(商品和服务)

EUIPO is making use of AI based semantic search for Goods and Services in Easy Filing helping users to find the right protection for their trade marks.


相关链接

欧洲联盟European Union Intellectual Property Office (EUIPO)图形检索(商标、外观设计)

The EUIPO has developed an in-house image search system that is integrated in eSearch plus to search Trademarks and designs using images.


相关链接

欧洲联盟European Union Intellectual Property Office (EUIPO)帮助台服务

EUIPO has its first Chatbot included in Easy Filing helping users to ask trademark related questions using standard responses with a possibility to go back to an Human agent.


相关链接

欧洲联盟European Union Intellectual Property Office (EUIPO)专利审查管理

EUIPO has created an algorithm assess a given pair of goods and/or services and provide a prediction as to the outcome of the comparison based on the historical data together with finding the closest semantically relevant matches. The tool is only available for examiners at the moment

澳大利亚IP Australia商标分类(商品和服务)

IP Australia has implemented a Trade Mark International Classification Service (TMICS) Application Programming Interface (API) to assist users when searching the Madrid Goods and Service (MGS) database. Leveraging Natural Language Processing (NLP) models (sentence-transformers), TMICS helps finding goods and services that are semantically related to the search terms entered. This reduces the number of queries that need to be performed to identify relevant goods and services. It also improves the quality of trade mark applications when filing overseas.

Additionally, TMICS delivers business intelligence benefits to IP Australia through semantic comparison of the Australian Picklist with the MGS database to determine gaps in coverage.

(Updated September 2024)


相关链接

澳大利亚IP Australia商标检索

IP Australia implemented the Trade Mark Precedent Identification (TMPI) tool in November 2023. TMPI has been built to retrieve, rank, and display in order of relevance, substantially identical text trade marks from the Australian trade marks register during examination.

It uses a combination of well-defined, automated business rules and Natural Language Processing (NLP) techniques including key word extraction, segmentation, lemmatisation, spelling correction and character replacement to ensure the automated search query includes relevant variations of the trade mark being searched.

The new search function intends to improve quality and consistency, when searching for substantially identical marks and increase the decision-making capability of trade mark examiners, by providing them ready access to highly relevant information.

(Updated September 2024)


相关链接

澳大利亚IP Australia专利审查管理

IP Australia’s Automated Preliminary Search tool (APST) executes an automated search query at the start of the search and examination process, looking for potential prior publication by the applicants or inventors named in a patent application. The tool searches IP Australia’s non-open to public inspection (OPI) database, as well as an external OPI data sources. The default query is based on the applicant and inventor names, as well as Cooperative Patent Classification (CPC) and International Patent Classification (IPC) symbols and can be further refined by users. Natural Language Processing (NLP) is used to compare the potential citations as well as claims with the input application and provide a relevance ranking to the user.

(Updated September 2024)


相关链接

澳大利亚IP Australia专利审查管理

IP Australia’s Family Member Analyser (FMA) tool provides patent examiners with direct links to family members and documents from their electronic dossiers (where available) during patent examination. Examiners will often consider observations made in Foreign Examination Reports (FERs) of closely related patent family members to improve examination quality and to avoid duplication of work where appropriate.

To assist with this process, FMA is used to retrieve and identify the most relevant family members suitable for examination purposes. Natural Language Processing (NLP) is used to perform a pairwise comparison between the claims of the family member and those of the queried application. A relevance score, which can be viewed by the user, is then assigned to each family member based on the comparison. FMA also facilitates a deep dive into FERs using the FER Feature Analyser (FFA) function that searches examination reports for novelty and inventive step objections and presents these specifically to examiners for their assessment. (Updated September 2024)


相关链接

澳大利亚IP Australia专利分类

IP Australia's Patent Auto Classification (PAC) service analyses the contents of a patent specification and predicts relevant technology groups enabling prioritisation and allocation to appropriate patent examiner sections.

The service is an internally developed machine learning-based system that performs technology sorting of patent applications using a sophisticated hierarchy classification model. The system allocates relevant International Patent Classification (IPC) marks based on the extracted abstract, descriptions and claim texts form the patent specifications. This in turn allows for the distribution of patent applications to the appropriate patent examination section for further processing, based on the top IPC mark allocated by the model.

PAC replaced the manual process previously in place that required a patent examiner to perform a technology sorting function: to read a patent specification, decide on the most appropriate IPC classification, and assign that application to the appropriate examination section for search and examination. The PAC model is retrained annually when new versions of the IPC become available.

(Updated September 2024)


相关链接

澳大利亚IP Australia专利审查管理

IP Australia continues to investigate the use of machine learning to determine the risk and complexity level of patent applications waiting in the exam requested stockpile. AI models are used to predict the effort required to conduct a high-quality examination. This approach facilitates the allocation of appropriate resourcing for each examination task thereby improving examination efficiency. Several proof of concepts (POC) are being refined to determine the viability of such tools in providing the required information.

(Updated September 2024)


相关链接

澳大利亚IP Australia专利审查管理

In Australia, the Commissioner of Patents may direct an applicant to request examination for a patent application. This process is known as a ‘direction’ and is used to manage the inventory of applications and examination requests within IP Australia.

IP Australia’s Outcome Based Directions service uses a machine learning model to identify applications that are ‘ready’ and ‘interested’ in pursuing examination, and issues directions to request examination in the order determined by the model. This system offers improved inventory management flexibility when compared to the process of the Commissioner of Patents issuing directions based entirely on chronological order.

(Updated September 2024)


相关链接

澳大利亚IP Australia商标检索

TM Checker is a free AI-assisted trade mark availability check. TM Checker is aimed at educating small to medium enterprises, who otherwise do not engage with the IP system, about trade marks and help them navigate the application process quickly and easily. A user can enter a brand name or logo and TM Checker provides general observations about eligibility for a trade mark, using a tool with AI assisted algorithms to assist searches of the trade mark register. TM Checker identifies potentially similar trade marks and highlights potential distinctiveness and offensiveness issues with the user’s proposed trade mark.

TM Checker continues to implement improvements to make the engine more accurate and improve usability of the tool.

IP Australia also provides customers the Australian Trade Mark Search to search for existing trade mark phrases and images. Australian Trade Mark Search uses the commercially available Clarivate Image Recognition software for the search functionality.

(Updated September 2024)


相关链接

瑞典Swedish Patent and Registration Office (PRV)机器翻译

瑞典专利注册局的专利审查员使用欧洲专利局在EpoQueNet和Espacenet中提供的机器翻译服务。

瑞士Swiss Federal Institute of Intellectual Property (IPI)数字化与过程自动化

瑞士联邦知识产权局使用基于经典规则的人工智能(Bosch SI Visual Rules和Camunda BPM)进行过程自动化(例如,带有费用或截止日期的裁决/决定的申请)。据瑞士局表示,基于规则的过程自动化在减少重复性的管理工作方面潜力最大。基本上,所有由费用或截止日期(包括文档创建)触发的决定都可能实现自动化,并且自动化工作可靠。瑞士局设有中央实时监控功能,严格使用业务过程建模符号(BPMN)流程进行自动化。

2018年初,瑞士局将推出使用ABBYY 智能分类器进行文件分类的自学人工智能。瑞士局持续使用手动分类的文件训练自学人工智能。瑞士局自动分析自学人工智能的结果质量,然后决定是否需要手动确认。手动确认的结果随后用于增强人工智能的训练集。

2018年初,瑞士局计划推出用于信息提取的自学人工智能,使用ABBY InfoExtractor进行高级企业检索。


相关链接

美利坚合众国United States Patent and Trademark Office (USPTO)专利分类

2018年初,美国专利商标局在研究深度机器学习“优质聊天机器人”(Quality Chat Bots),以便为美国专利商标局《专利审查程序手册》(MPEP)以及其他使用算法和权利要求用语的权利要求分析和分类分析提供“概念提问”(而不是关键词),从而更好地了解权利要求用语和分类的趋势。


相关链接

美利坚合众国United States Patent and Trademark Office (USPTO)专利审查管理

美国专利商标局有一套应用于多领域的、结合人工智能与大数据和机器学习的方案。其中包括:为审查员提供最实用相关的信息,助其判定申请的专利性;对专利申请以及专利商标局的后续行动进行文本分析,进而分析专利申请历史;改进应用程序接口,令公众访问美国专利商标局数据更加便利。

这套方案是美国专利商标局使用按每个应用程序和每个系统定制的开源技术(Java和Python)自主开发。

美国专利商标局的人工智能方案在以下领域对商标业务作了改进:1)使用分析功能开发了质量审核智能表格;2)利用包括使用和描述性统计在内的高级分析功能,对大数据库实施办公职能;3)确定深度机器学习对商标图形检索的效果。


相关链接

美利坚合众国United States Patent and Trademark Office (USPTO)专利现有技术检索

2018年初,美国专利商标局正在交付概念验证“Sigma”,它使用机器学习/人工智能算法在文档数据库中检索整篇文档。这一版Sigma在已授权专利和授权前公告(仅限美国)中检索专利申请。


相关链接

美利坚合众国United States Patent and Trademark Office (USPTO)专利现有技术检索

United States Patent and Trademark Office (USPTO) recently added the new artificial intelligence (AI)-based “Similarity Search” feature to examiners’ Patents End-to-End (PE2E) search suite. This new tool will support the USPTO’s goal to grant more robust and reliable patents by further assisting patent examiners as they search for prior art during the examination process.

联合王国Intellectual Property Office (UKIPO)商标检索

Trademark Application Checker "Check if you could register your trademark" is a free AI-assisted trademark application acceptance checker. It is aimed at assisting and educating novice users with their application; it helps identify goods and services to protect, any aspects of a trademark that might not be appropriate (such as offensive words or protected symbols e.g. a crown or crest), and if there are any similar trademarks that could cause conflict. It does not provide legal advice, and does not prevent a user submitting an application, as a trademark examiner always examines the submitted trademark. (Updated November 2024)

联合王国Intellectual Property Office (UKIPO)专利现有技术检索

Enhancing Key word searching of Rights for the public

As part of the development of a new public facing webpage for the public to search rights, we have included the ability to perform key word searching.

This uses a hybrid search system combining BM25 search with document similarity search using sentence embeddings. Base models have been finetuned using Patent Examiner Epoquenet searches as well as searches generated using LLMs fed patent documents. Once live our service will use public searches as a feedback loop to validate and if possible train future iterations of our search model.

We are also exploring options for query term expansion where we add additional terms into the BM25 search, either by enhancing the user query or potentially expanding the documents themselves similar to methodology applied in splade v2.

Currently this covers patents, but future phases of development will include search enhancements for trademarks and designs.

(Updated November 2024)

联合王国Intellectual Property Office (UKIPO)专利分类

The UKIPO has been developing a tool for the automatic allocation of new patent applications according to the International Patent Classification (IPC). After the exploration of different technologies, a methodology based on an ensemble of neural networks trained using the extracted textual components of the patent (claims, description and abstract) was settled on. Our solution makes use of the flexibility of neural networks by generating the probabilities of the primary mark and additional marks separately for each textual type. These are then combined to produce a top 5 ranking result.

The accuracy of the result (currently 70% for single mark allocation, and 90% for top 5) allows for the distribution of patent applications to the appropriate patent examination section for further processing, based on the top IPC mark allocated by the models. Note that the model is retrained annually when new versions of the IPC become available. (Updated November 2024)

联合王国Intellectual Property Office (UKIPO)专利现有技术检索

Patent summarisation

The UKIPO has been exploring the use of ML and AI in summarising patent applications. Both abstractive and extractive methods have been explored, using cutting edge Large Language Models (LLMs) and modern Natural Language Processing (NLP) techniques. Our eventual solution was an NLP based extractive method - sentences are split and scored, clustered according to their context, then ranked and the top N output as the summary (according to length)

This tool has many uses: selecting the most relevant parts of the description (based on entered criteria); generating novel text for use in model training; creating clusters of similar sentences and the automatic generation of abstracts. (Updated November 2024)

芬兰Finnish Patent and Registration Office (PRH)专利现有技术检索

A patent search system by Finnish startup IPRally Technologies Ltd. was deployed for all patent examiners at the Finnish Patent and Registration Office in March 2020. We chose the system after comparing it to other AI based patent search systems that could be installed on our own servers, such as IPScreener and Teqmine. The search system by Teqmine Analytics Ltd. was installed on our servers previously (from late 2017 until early 2020), but it was never deployed for all examiners. In addition, we have previously tested other systems, such as InnovationQ Plus.

IPRally's system builds a graph from an input text that typically includes the claims and description of a patent application, where the dependencies between concepts are modeled by the structure of the graph. This graph is input into a Graph Neural Network (GNN), which produces a high-dimensional vector that can be compared to vectors produced from prior art patent documents. The system is trained using a large amount of search report data for patent publications. The output format from the system was customized so that it can be easily transferred into our existing search tools. An examiner can thus analyze whether the prior art publications represented by the closest vectors, i.e. top ranked documents, are actually relevant prior art.

The system performs better than the systems we have previously tested. In real world testing, a prior art document that was eventually used by an examiner to deny novelty or inventive step in the first office action was found among the top 20 ranked documents in more than 40 per cent of the cases. At least in some cases, using the system may thus allow for a faster search or for a better quality search.

The system can also be used to find possible classifications for a patent application by determining the most common classification symbols for the top ranked documents.

芬兰Finnish Patent and Registration Office (PRH)专利分类

A patent search system by Finnish startup IPRally Technologies Ltd. was deployed for all patent examiners at the Finnish Patent and Registration Office in March 2020. We chose the system after comparing it to other AI based patent search systems that could be installed on our own servers, such as IPScreener and Teqmine. The search system by Teqmine Analytics Ltd. was installed on our servers previously (from late 2017 until early 2020), but it was never deployed for all examiners. In addition, we have previously tested other systems, such as InnovationQ Plus.

IPRally's system builds a graph from an input text that typically includes the claims and description of a patent application, where the dependencies between concepts are modeled by the structure of the graph. This graph is input into a Graph Neural Network (GNN), which produces a high-dimensional vector that can be compared to vectors produced from prior art patent documents. The system is trained using a large amount of search report data for patent publications. The output format from the system was customized so that it can be easily transferred into our existing search tools. An examiner can thus analyze whether the prior art publications represented by the closest vectors, i.e. top ranked documents, are actually relevant prior art.

The system performs better than the systems we have previously tested. In real world testing, a prior art document that was eventually used by an examiner to deny novelty or inventive step in the first office action was found among the top 20 ranked documents in more than 40 per cent of the cases. At least in some cases, using the system may thus allow for a faster search or for a better quality search.

The system can also be used to find possible classifications for a patent application by determining the most common classification symbols for the top ranked documents.

菲律宾Intellectual Property Office of the Philippines (IPOPHL)数字化与过程自动化

菲律宾局使用商用的业务智能软件 COGNOS来为该局的管理报告要求提供支持。在使用该系统时,菲律宾知识产权局从工业产权自动化系统(IPAS)数据库向COGNOS可读包执行ETL(提取-传输-加载)程序。

菲律宾Intellectual Property Office of the Philippines (IPOPHL)专利现有技术检索

菲律宾知识产权局(IPOPHL)使用名为 DTSearch的第三方搜索引擎进行专利检索。该系统与所有其他搜索引擎相似,能够执行增量索引、模糊搜索和其他功能。尽管该系统是一种低端人工智能,但是比传统数据库检索更为强大。

西班牙Spanish Patent and Trademark Office (OEPM)图形检索(商标、外观设计)

The OEPM is currently in a process of continuous analysis of state-of-the-art tools and services, which may be useful in different areas of the OEPM. In this context, contacts are being made with cloud service providers of recognized prestige to study the possibility of applying their capabilities, either directly or through specific training and development, to the needs of the OEPM in relation to image comparison. This functionality would be of direct application in the field of Detection of Priorities of Distinctive Signs.

西班牙Spanish Patent and Trademark Office (OEPM)数字化与过程自动化

In recent years, a process of discovery, analysis and implementation of possible robotizable processes, applying RPA technology, has been carried out continuously at the OEPM. All departments of the OEPM have been involved in this process. As a result, the following three procedures have been automated by using pay-per-use technology, installed in the cloud:

- Process 1: Implementation of the process of sending the original copy to WIPO.

- Process 2: Implementation of a process of checking that filed sequence lists are in accordance with ST-26

- Process 3: Implementation of the process of sending priority documents to WIPO (work in progress).(Updated October 2024)