Genealogical Network Inference and Analysis

Code for the following paper:

Malmi, E., Gionis, A., Solin, A. "Computationally Inferred Genealogical Networks Uncover Long-Term Trends in Assortative Mating". In Proceedings of the Web Conference (WWW 2018), Lyon, France, 2018.

Data

To get access to the data, email [email protected] (first author), and you will receive a link to a download site.

Running Genealogical Network Inference

Once you have downloaded and copied the HisKi database into the data directory, do the following three steps.

First, link the burial records into birth records in, e.g., 50 batches by running

python compute_burial_matches.py --indices=burial_event_ids.txt --num_batches=50 --batch_index=1 \
                                 --output=reconstructions/burial_birth_links_single

for batch_index values from 1 to 50 (can be run in parallel).

Second, store the burial-birth links into the HisKi DB by running

sh serial_burial_db_write.sh

Third, infer the parents of each person in the HisKi DB in 50 batches by running

python compute_matches.py --indices=birth_event_ids.txt --num_batches=50 --batch_index=1 \
                          --output=reconstructions/all_parent_edges

for batch_index values from 1 to 50 (can be run in parallel).

Running Assortative Mating Analysis

cd src/assortative_mating/
python assortative_mating_analysis.py