MiNLP

MiNLP

Xiaomi Natural Language Processing Platform (MiNLP) has dozens of functional modules such as lexicography, syntax, and semantic analysis, and has been widely used in the company's business.

After continuous optimization and practical polishing, the MiNLP-Tokenizer Chinese word segmentation tool was officially opened to the public in November 2020.

We plan to complete the open source of all lexical tools (part-of-speech annotation and named entity recognition) in Q2 2021. Starting from Q3 2021, we will gradually open source syntax analysis and some semantic analysis tools, and work with developers to create a powerful and leading NLP platform.

duckling-fork-chinese Chinese structured parsing tool is a JVM fork based on facebook/duckling to meet the conversion from text to structured objects. Currently, tools are widely used in Xiaoai production environment, and the most important application is the analysis of numbers and time.

Module

Chinese participle: MiNLP-Tokenizer (portal)

Part of speech marking: During construction, please stay tuned

Named entity recognition: Stay tuned

Dependence syntax analysis: Stay tuned

Structured analysis: duckling-fork-chinese