Parameter settings for pre-processing models¶
(Default values are in ~/search/Pre-processing.yaml)
The benchmark provides several arguments for describing:
Basic setting of the parameters
See below for the details:
Pre-processing required parameters¶
preprocessing_model (str): The name of the pre-processing model, which automatically loads the configs from search/properties/models/<model_name>.yaml.
Ranklib parameters¶
name (str) : The name of the ranker class.ranker_path (str): The path to the Ranklib library.rel_max (int): The maximum relevance to be assigned to items.ranker (str): The name of the ranking model.ranker_id (int): The ID of the ranking model as per the Ranklib library.metric (str): The evaluation metric.top_k (int): The evaluation at top-k.lr (float): The learning rate for training.epochs (int): The number of epochs for training.train_data (list): The data types for training [“original”, “fair”].test_data (list): The data types for testing [“original”, “fair”].
Evaluation parameters¶
metrics (list): The list of fairness evaluation metrics having the following values [“diversity”, “select_rate”, “exposure”, “individual”, “igf”].rankings (list): The list of rankings to perform the evaluation on, having the following values: “score_col” - original data, “<score_col>__<score_col>” - trained and tested on original data, “<score_col>_fair” - transformed data, “<score_col>fair_<score_col>_fair” - trained and tested on transformed data.k_list (list): The top-k values for evaluation.