Paddlenlp general information extraction model: UIE [information extraction {entity relationship extraction, Chinese word segmentation, accurate entity markers, emotion analysis, etc.}, text error cor

0. PaddleNLP One click predictive power Taskflow API Three characteristics
   Comprehensive function

Full scene support ： Cover NLU and NLG Eleven classic tasks in the field .
Document level input ： Support document level input , Solve the length limitation of the input text in the pre training model , Greatly save the amount of code development when users enter long text .
Customized training ： Support users to use their own data sets for customized training , Use the customized training model with one click through the user-defined path .
Easy to use

Open the box , The cost of learning is low , A few lines of code will complete the call .
Industrial effect

Gather many self-developed algorithms in Baidu naturallanguageprocessing field and excellent open source models in the community , Model effect is leading .

---

Reference material ：
PaddleNLP General information extraction technology UIE【 One 】 Industrial application examples ： Information extraction { Entity relation extraction 、 Chinese word segmentation 、 Accurate physical mark . Emotional analysis, etc }、 Text correction 、 Question answering system 、 Chat robots 、 Custom training
PaddleNLP–UIE（ Two ）– Fast performance improvement with small samples （ contain doccona mark ）
uie Model tuning personal summary
Information extraction new SOTA: UIE– The first structured generative information extraction pre training language model , Four tasks of unified information extraction
Information extraction is unified ： Baidu Chinese Academy of Sciences released a general extraction model UIE, Refresh 13 individual IE Data sets SOTA！