patent-parsing-tools is a library providing tools for generating training and test set from Google's USPTO data helpful with for testing machine learning algorithms
Project description
## System requirements:
`Bash sudo yum install python-devel libxslt-devel libxml2-devel `
## Python requirements:
`Bash pip install -r requirements.txt `
## Running:
Collecting and serializing data: `Bash python -m patent_parsing_tools.supervisor [working_directory] [train_destination] [test_destination] [year_from] [year_to] `
Eg. `Bash python -m patent_parsing_tools.supervisor patents/working_directory patents/train_destination patents/test_destination 2014 2015 `
Generating dictionary with train set: `Bash python -m patent_parsing_tools.bow.dictionary_maker [train_directory] [max_parsed_patents] [dict_max_size] [dictionary_name] `
Eg. `Bash python -m patent_parsing_tools.bow.dictionary_maker patents/train_destination 1000000000 4096 dictionary.txt `
Generate bag of words with train set and test set: `Bash python -m patent_parsing_tools.bow.bag_of_words [directory_with_serialized_patents] [destination_directory] [dictionary.txt] [package_size > 1024] `
Eg. `Bash python -m patent_parsing_tools.bow.bag_of_words patents/train_destination patents/final_dataset_train dictionary.txt 1048576 python -m patent_parsing_tools.bow.bag_of_words patents/test_destination patents/final_dataset_test dictionary.txt 1048576 `
## Running tests
`Bash python -m unittest discover . `
