How to use the simplification.term_replacement function in simplification

similarity_algorithm=2, filtering_algorithm=1,
                       number_of_terms=1000, simplify_terms=False, model=None,
                       data_dump_path=None):
    """returns a pair similarity dictionary for the map and set of terms in the map. Heatmap can
    be calculated seperately and then overlaid. Will need to convert dictionary representation
    to dot file format"""
    flattened = flatten(structured_nps)
    set_status('ranking terms', model=model)
    if start_words is not None:
        # start words should be a list like ["machine learning", "artificial intelligence"]
        start_words = [tuple(s.split()) for s in start_words]
        ranked_phrases, phrase_frequencies, scored_phrases = call_rank(ranking_algorithm, flattened, number_of_terms, start_words=start_words, model=model)
    else:
        ranked_phrases, phrase_frequencies, scored_phrases = call_rank(ranking_algorithm, flattened, number_of_terms, model=model)
    if simplify_terms:
        structured_nps = simplification.term_replacement(structured_nps, ranked_phrases)
    set_status('calculating similarity', model=model)
    sim_matrix, phrase_lookups = call_similarity(similarity_algorithm, structured_nps, ranked_phrases, model=model, status_callback=lambda s: set_status(s, model=model))
    if data_dump_path:
        import pickle
        from os.path import join
        def prefix_path(rel):
            return join(data_dump_path, rel)
        with open(prefix_path('sim_matrix.pickle'), 'w') as f:
            pickle.dump(sim_matrix, f)
        with open(prefix_path('phrase_lookups.pickle'), 'w') as f:
            pickle.dump(phrase_lookups, f)
        with open(prefix_path('phrase_frequencies.pickle'), 'w') as f:
            pickle.dump(phrase_frequencies, f)
    phrase_pairs = call_filter(filtering_algorithm,  sim_matrix, phrase_lookups, model=model)
    normed = similarity.similarity_dict_to_distance(phrase_pairs)
    # build set of terms in graph