################################################################################
# 2014-06-11
# This file contains the instructions to generate the files with
# Python 2.7 and Mathematica
################################################################################

################################################################################
# Mathematica
################################################################################
# Put the files 
# - dblp_top_clean.nb
# - dblp_bias_stat_clean.nb
# - A_D_Proof.nb
# - Vcolor_simple_v1.m  
# as well as 
# - dblp_edges_maxAuthor20_mentee_frequent_known_degree_v1.csv
# - dblp_edges_maxAuthor20_simple_degree_1000_edges_v1.csv
# - dblp_nodes_maxAuthor20_simple_degree_1000_nodes_v1.csv
# - dblp_id_gender_dict_v1.csv
# in the same directory and open the notebooks.

################################################################################
# Python
################################################################################
# Unzip ./data/dblp-2013-12-23.zip to ./data/dblp-2013-12-23.xml and
# run the following command line instructions from this directory:
# "Parsing xml, create (multi) collaboration graph for all publications with at most 20 co-authors"
python ./code/parse_DBLP_xml.py './data/dblp-2013-12-23.xml' './output/dblp_edges_maxAuthor20.csv' './output/dblp_id_name.csv' 20

# "Compute (simple) co-author graph from collaboration (multi) graph"
python ./code/convert_to_simple_graph.py './output/dblp_edges_maxAuthor20.csv'

# "Create (sorted) id degree list from (simple) co-author graph"
python ./code/create_id_degree_file.py './output/dblp_edges_maxAuthor20_simple.csv'

# "Extract top node list of authors with known gender and induced subgraph of co-authors"
python ./code/extract_top_graph.py './data/dblp_id_gender_dict_v0.dict' './output/dblp_edges_maxAuthor20_simple.csv' './output/dblp_edges_maxAuthor20_simple_id_degree.csv'
python ./code/extract_top_graph.py './data/dblp_id_gender_dict_v1.dict' './output/dblp_edges_maxAuthor20_simple.csv' './output/dblp_edges_maxAuthor20_simple_id_degree.csv'

# "Compute mentor/mentee"
python ./code/create_mentor_file.py './output/dblp_edges_maxAuthor20.csv' './output/dblp_edges_maxAuthor20_mentee_frequent.csv'

# "Compute known mentoring degree"
python ./code/compute_known_degree.py './data/dblp_id_gender_dict_v0.dict' './output/dblp_edges_maxAuthor20_mentee_frequent.csv'
python ./code/compute_known_degree.py './data/dblp_id_gender_dict_v1.dict' './output/dblp_edges_maxAuthor20_mentee_frequent.csv'