This repo contains the experiment results and code for SemEval 2020 Task 11 - Proganda Detection.
The following data files should be helpful for getting up and running quickly in order to do modelling:
data/task1_data.csv: contains all the training lines and extracted sequences for Task 1 binary sequence labeling. The BIO annotations were generated with simple white space tokenization.data/task1_dev.csv: contains all the dev lines and article_ids that need to be classified for submissio to the dev leaderboard.task_2_data.csv: contains all the train and dev data for task multiclass classification challenge.
In the utils folder, I've created some common utilities that might be useful.
Generate Task 2 formatted submission
generate_t2_sub(preds: List[str]) -> List[str]
Get Original Article text
get_article(file: str) -> str:
Get file path given a train article id
get_task1_file(article_id: int) -> str:
Get file path given a dev article id
get_task2_file(article_id: int) -> str: