The project JOCL consists of two parts: (1) GraphConstruction; (2) LearnAndInference.
- Unzip the files FreebaseEntityInWiki2.part01 (02, 03, 04).rar and combine them as a file named FreebaseEntityInWiki2. Put the file FreebaseEntityInWiki2 in the folder "GraphConstruction\intermedia_data".
- Unzip all the files of the folder "GraphConstruction\intermedia_data".
- Unzip the files original_data.rar and external_data.rar and put them in the folder "GraphConstruction".
- Download crawl-300d-2M.vec (https://fasttext.cc/docs/en/english-vectors.html) and put it in the folder "GraphConstruction".
- Run the file GraphConstruction\GraphGen.py to construct the factor graph model of the framework JOCL.
- Run the file LearnAndInference\JOCL.cpp to learn parameters of JOCL and inference.
- Run the file GraphConstruction\Evaluation\Final-Evaluation.py to test the performance of JOCL.
We store some intermediate data files generated by each step in the corresponding position.
If you want to obtain the results of JOCL quickly, please start step 7 derectly.
More content will be published soon.