lichunnan / marktool

这是一款基于web、面向领域的通用文本标注工具，支持大规模实体标注、关系标注、事件标注、文本分类、基于字典匹配和正则匹配的自动标注以及用于实现归一化的标准名标注，同时也支持文本的迭代标注和实体的嵌套标注。标注规范可自定义且同类型任务中可“一次创建多次复用”。通过分级实体集合扩大了实体类型的规模，并设计了全新高效的标注方式，提升了用户体验和标注效率。此外，本工具增加了审核环节，可对多人的标注结果进行一致性检验、自动合并和手动调整，提高了标注结果的准确率。

DoTAT: A Domain-oriented Text Annotation Tool

East China University of Science and Technology - NLP

[华东理工大学-自然语言处理与医疗大数据实验室]

Attention

该工具已于2020年获得软件著作权，证书号：软著登字第5885316号，如需二次开发使用则要在项目中着重标明来源ECUST-NLP！

Notification

在线试用版网站（a live demo website）：http://175.24.69.95/new/

A typical annotation process using DoTAT may include the following five steps:

(1) Define annotation specifications: The administrator selects the annotation type andvisually defines event types, entity types, relation types or text categories in annotation specifications.

(2) Create and assign tasks: Administrator cre-ates and assigns tasks. Each task contains anannotation specification and several raw texts.It is recommended that two annotators andone reviewer are assigned to each task.

(3) Annotate: Before the annotators interactively annotate events or entities, they can use automatic batch annotation to accelerate the speed.

(4) Merge and Review: The reviewer starts consistency checking and automatic merging ofthe annotated content by multiple annotators. The reviewer can visually analyze the errors according to the merged events list. When there are too many similar errors, the reviewer can give feedback for administrator to re-definethe annotation specification. With iterative annotation function, all exsisting annotationscan be reused.

(5) Export results: After the review process, theannotated content can be exported by admin-istrator to a result file and saved in JSON format.

During the trial period, the user's task management rights and annotation specification management rights will be opened. You will need to contact us to apply for the administrator account in the future.

更多最新消息请加QQ群1054082857 (MarkTool in NLP) 第一时间了解，加群时需要验证身份，已收藏该项目的GitHub帐号才会通过申请。 欢迎进行学术交流, 如需合作请私聊。

Code

后端代码地址[暂不开放]：https://github.com/sxthunder/ecust_annotation

clone the project

git clone https://github.com/FXLP/MarkTool.git

enter the project directory

cd MarkTool

(You may need nodejs/npm v8. You can use the nvm to control the version of nodejs/npm.)

install dependency

npm install

develop

npm run dev