ssp.snorkel

---

class ssp.snorkel.labelling_function.SSPLabelEvaluator(text_column='text', label_column='label', raw_tweet_table_name_prefix='raw_tweet_dataset', postgresql_host='localhost', postgresql_port='5432', postgresql_database='sparkstreamingdb', postgresql_user='sparkstreaming', postgresql_password='sparkstreaming')

Bases: ssp.posgress.dataset_base.PostgresqlDatasetBase

run_labeler(version=0)

class ssp.snorkel.labelling_function.SSPTweetLabeller(input_col='text', output_col='slabel')

Bases: sklearn.base.BaseEstimator, sklearn.base.TransformerMixin

Snorkel Transformer uses LFs to train a Label Model, that can annotate AI text and non AI text :param input_col: Name of the input text column if Dataframe is used :param output_col: Name of the ouput label column if Dataframe is used

ABSTAIN = -1

NEGATIVE = 0

POSITIVE = 1

static bigram_check(x, word1, word2)

evaluate(X, y)

fit(X, y=None)

Parameters

• X – (Dataframe) / (List) Input text

• y – None

Returns

Numpy Array [num of samples, num of LF functions]

is_ai_tweet = LabelingFunction is_ai_tweet, Preprocessors: []

is_not_ai_tweet = LabelingFunction is_not_ai_tweet, Preprocessors: []

static negative_search(data, positive_keywords, false_positive_keywords)

normalize_prob(res)

not_ai = LabelingFunction not_ai, Preprocessors: []

not_big_data = LabelingFunction not_big_data, Preprocessors: []

not_cv = LabelingFunction not_cv, Preprocessors: []

not_data_science = LabelingFunction not_data_science, Preprocessors: []

not_neural_network = LabelingFunction not_neural_network, Preprocessors: []

not_nlp = LabelingFunction not_nlp, Preprocessors: []

static positive_search(data, key_words)

predict(X)

transform(X, y=None)