In this project, we have implemented various keywords/keyphrases extaction methods, including supervised methods and unsupervised methods. Since these two category methods will accept different number of parameters, we have two run files:

for run.py: Run as follows: $ python run.py method_name dataset_dir dataset_name For example, run.py nlm graph_closeness method_name: choose from 'NB', 'graph_closeness', 'text_rank', 'svm', 'svm_ranking' dataset_dir: the actual data directory dataset_name: choose from 'nlm', 'js', should match the data directory you mentioned in the previous argument

run2.py is for unsupervised methods To run this program, two command line arguments are needed: The first one is method name used to extract keywords, vaild method names are listed below: tf : term frequency based extraction td : term distribution based extraction tfd : assemble term frequency and term distribution closeness : closeness centrality based graph method textrank : textrank centrality based graph method

The second argument can be a directory or a filename if it's directory (containg docs and corresponding keywords), the program will output the average accuracy and recall for all the files in the directory; if it's filename, the program will output the extracted keywords/keyphrases