Input data is a JSON file in which each item is a document. For example:
{
"doc_id": "doc0001", # unique ID for each document
"title": "",
"body": "I was diagnosed with bipolar 2 years ago and started taking Lamictal. Last April, my psych switched me to Abilify because of issues with my blood tests. I didn't have any side effects with the lamictal but now I feel just super antsy and restless-- has anyone else had this issue?",
"dct": "2019-06-22", # Document Creation Time
"meds": { # list of medication entities and their respective mentions
"196502": [
{
"mention_id": "mention_0_196502"
"string": "Lamictal",
"entity_id": "196502", # RxCUI code
"type": "med",
"source": 1, # 0 for title, 1 for body
"span": [60, 68] # Mention's token span in text
},
{
"mention_id": "mention_1_196502"
"string": "lamictal",
"entity_id": "196502",
"type": "med",
"source": 1,
"span": [193, 201]
}
],
"352393": [
{
"mention_id": "mention_0_352393"
"string": "Abilify",
"entity_id": "352393",
"type": "med",
"source": 1,
"span": [106, 113]
}
]
},
"dates": # list of date-time 'entities' grouped by equal value
"20170622000000": [
{
"mention_id": "mention_0_20170622000000"
"string": " 2 years ago",
"value": "2017-06-22 00:00:00"
"entity_id": "20170622000000",
"type": "date",
"source": 1, # 0 for title, 1 for body
"span": [29, 40] # Mention's token span in text
}
],
"20190401000000": [
{
"mention_id": "mention_0_20190401000000"
"string": "Last April",
"value": "2019-04-01 00:00:00"
"entity_id": "20190401000000",
"type": "med",
"source": 1,
"span": [70, 80]
}
]
},
"labels": { # temporal relations
"196502": {
"20170622000000": "start",
"20190401000000": "stop,
"DCT": "after"
} "352393": {
"20170622000000": "before",
"20190401000000": "start,
"DCT": "on"
}
},
"user_id": "user01",
"subreddit": "Bipolar", # Name of subreddit the post was scraped from
python src/relation_extraction.py --config [config_name]
python3.8.16torch.2.0.1- CUDA 11.7