fake-news-detection

Final Year Project

- /
        -- credibility.py : generates dictionary of authors / speakers with credibility scores and stores object on disk
        -- model.py : generates word embeddings using Word2Vec model after tokenization and data cleaning
        -- config.py : stores global tuning parameters
        -- tfidf.py : generates and stores traditional tfidf vectors
        -- classifiers.py : formats train and test data, trains model and stores the results obtained on disk

- data /
        -- vectors.tsv : output of word2vec model, i.e generated word vectors
        -- metadata.tsv : vocabulary corresponding to word2vec model
        -- labels.tsv : labels of data tuples after aggregation
        -- input_vectors_words.tsv : input tuples to classifiers with words
        -- input_vectors.tsv : input tuples to classifiers in the form of vector
        -- tfidf.tsv : tfidf vectors correspoding to statements in datasets
        -- credibility_lookup.pickle : serialized credibility lookup
        -- output_tfidf.json : results of classifier with tfidf
        -- output_word2vec.json : results of classifier with word embeddings