I participated in the natural language group of TAL's winter camp this year. I was very happy, and I got good grades. I got to know a lot of big guys. Special thanks to my teammate Liu Zhiqiang's children's shoes. The good teammates are really reliable. It's a pity that I can't go there for some reasons, very, very regrettable. Let's update the second week first, because we will talk about it soon. . . The sentences in the first week are similar and let's do it after a while. . . I don’t want to write specific things, just put the GitHub address directly. https://github.com/DDigimon/TCAMP-WEEK2
This time I am mainly responsible for the deep learning part and part of the feature selection. Just take a screenshot of the ppt. . .
Additional supplement. In general, the methods of all the students are similar, but the results of the top few in the end depend to a large extent on how well the parameters are adjusted. . . Before deep learning, when doing named entity recognition, the effect was greatly improved when the word feature was also used as a feature for training, but this time I don't know why the effect did not appear. . . The code is implemented in GitHub. .