sysrev-seed-collection
March 30, 2022 ยท View on GitHub
This is a collection repository for systematic review topics with seed studies. The file is separated into two sections:
- collection_data: contains organised data, including basic data, search data and snowballing data.
- experiments: contains the result and data usage for three experiments, including seed-driven query formulation, seed-driven document ranking, and seed driven snowballing document ranking.
For folder collection_data:
- overall_collection.jsonl: with each line referring to one topic, every topic includes following attributes:
| Attributes | Description |
|---|---|
| id | correpsonds to ID in paper |
| link_to_review | corresponds to Link to Review in paper |
| title | Corresponds to Title in paper |
| search_name | corresponds to Description in paper |
| Date_from | Corresponds to Date restriction in paper |
| Date_to | Corresponds to Date restriction in paper |
| query | Corresponds to PubMed query in paper |
| seed_studies | Corresponds to Seed studies in paper |
| included_studies | Corresponds to Included studies in paper |
| edited_search | Stating the edited search queries (with standard format) |
- search: a folder containing candidate_documents.res, which is a trec format file for candidate documents retrieved by systematic review boolean queries, this file corresponds to Retrieved studies in paper
- snowballing: a folder containing seed_snowballing_document.tsv and screened_snowballing_documeent.tsv, these two files are for snowballing candidate document, every line corresponds to one topic, topic and documents' list is separated by \t, documents' ids are separated by '|', these two files correspond to Snowballed studies in paper
For folder experiment:
Three experiments in the paper are included in this folder, including:
- query_formulation: Automatic query formulation experiment.
- sdr_document_ranking: SDR-driven document ranking.
- sdr_snowballing: SDR-driven snowballing document ranking.
Instructions on how to run these experiments are inside each experiment folder.
For sample data extraction processing, please run:
python3 sample_data_processing.py
This is a sample data extraction work, with an input of topic id; the python script will output all the information for this topic.
Please refer to the paper "From Little Things Big Things Grow: A Collection with Seed Studies for Medical Systematic Review Literature Search"