def create_ground_truth()

in evaluation_pipeline/retrieval.py [0:0]

• 13 lines of code
• 5 McCabe index (conditional complexity)

def create_ground_truth(history, query_ids):
    queries = []
    relevant_ids = []
    for query in history['search_query'].unique():
        query_id = query_ids[query]
        queries.append(query_id)
        ground_truths = []
        for i, row in history.iterrows():
            relevant = 1 if query == row['search_query'] else 0
            if relevant == 1:
                ground_truths.append(int(i))
        relevant_ids.append(ground_truths)

    return dict(zip(queries, relevant_ids))