This release contains RNNSharp source code and a demo package for organization inner-structure parsing in Chinese.
The source code contains all projects related with RNNSharp, such as core algorithm, console tool, feature generator and so on. You can open it by Visual Studio. Note that, to get the latest source code or previous versions, please access SOURCE CODE section and download them.
For the demo package, please following the below instruction step by step to set up and use it
Assuming you have already created a path named "D:\RNNSharp" where you play with RNNSharp in your local machine
1. Visit RNNSharp project home page http://rnnsharp.codeplex.com
and download "RNNSharp organization inner-structure parser in Chinese" from DOWNLOADS
section to your local machine.
2. Uncompress "RNNSharp organization inner-structure parser in Chinese" to D:\RNNSharp
3. Download "RNNSharp source code and binary files" from above same page to your local machine
4. Uncompress "RNNSharp source code and binary files" to a temporary directory and copy all files in "bin" directory to D:\RNNSharp
5. Visit WSDSharp project home page http://wsdsharp.codeplex.com
and download "WSDSharp demo package for Chinese" from DOWNLOADS
section to your local machine. For using word embedding features, we need an existed encoded word2vec model.
6. Uncompress "WSDSharp demo package for Chinese" to a temporary directory and copy word_vector.bin file from the directory to D:\RNNSharp
7. Run train.bat to train the model and then generate a new model named model.bin
8. Run test.bat to test the model