evaluate.py

from problog.ddnnf_formula import DDNNF
from problog.formula import LogicFormula
from problog.program import PrologString

from utility import read_mapping
from sklearn import metrics

def predict_edge(path, user):
    with open(f"{path}/learn.pl", "r") as file:
        program = file.read()

    targets = read_mapping(f"{path}/test.txt")
    with open(f"{path}/predict.txt", "w") as file:
        for movie in targets.keys():
            prediction = list(DDNNF.create_from(LogicFormula.create_from(
                PrologString(program + f"\nquery(watched({user},{movie}))."))).evaluate().values())[0]
            file.write(f"{movie}:{prediction}\n")

def evaluate(path):
    targets = read_mapping(f"{path}/test.txt")
    predictions = read_mapping(f"{path}/predict.txt")
    points = list(targets.keys())
    targets = [float(targets[point][0]) for point in points]
    predictions = [float(predictions[point][0]) for point in points]
    return metrics.roc_auc_score(targets, predictions)