def get_test_lexicon()

in src/weakly_sup.py [0:0]

• 35 lines of code
• 10 McCabe index (conditional complexity)

def get_test_lexicon(
                test_set, test_lexicon, criss, lexicon_inducer, info, configs, best_threshold, best_n_cand
            ):
    induced_lexicon = list()
    pred_test_lexicon = collections.defaultdict(collections.Counter)
    probs = extract_probs(
        test_set, criss, lexicon_inducer, info, configs
    )
    for i, (x, y) in enumerate(test_set):
        pred_test_lexicon[x][y] = max(pred_test_lexicon[x][y], probs[i].item())
    possible_predictions = list()
    for tsw in set([x[0] for x in test_lexicon]):
        ssw = to_simplified(tsw)
        for stw in pred_test_lexicon[ssw]:
            ttw = to_traditional(stw)
            pos = 1 if (tsw, ttw) in test_lexicon else 0
            possible_predictions.append([tsw, ttw, pred_test_lexicon[ssw][stw], pos])
    possible_predictions = sorted(possible_predictions, key=lambda x:-x[-2])
    word_cnt = collections.Counter()
    correct_predictions = 0
    for i, item in enumerate(possible_predictions):
        if item[-2] < best_threshold:
            prec = correct_predictions / (sum(word_cnt.values()) + 1) * 100.0
            rec = correct_predictions / len(test_lexicon) * 100.0
            f1 = 2 * prec * rec / (rec + prec)
            print(f'Test F1: {f1:.2f}')
            break
        if word_cnt[item[0]] == best_n_cand:
            continue
        word_cnt[item[0]] += 1
        if item[-1] == 1:
            correct_predictions += 1
        induced_lexicon.append(item[:2])
    eval_result = evaluate(induced_lexicon, test_lexicon)
    return induced_lexicon, eval_result