discursive_diversity module

features.discursive_diversity.get_DD(chat_data, conversation_id_col, speaker_id_col)

Computes degree of divergence amongst the meanings conveyed by speakers in a given conversation. This is a conversation level feature.

Parameters:

chat_data (pd.DataFrame) – DataFrame containing chat data with ‘conversation_num’, ‘speaker_nickname’, and ‘message_embedding’ columns.

Returns:

pd.DataFrame with ‘conversation_num’ and ‘discursive_diversity’ columns representing discursive diversity per conversation.

Return type:

pd.DataFrame

features.discursive_diversity.get_cosine_similarity(vecs)

Computes cosine similarity between a list of vectors.

Parameters:

vecs (list) – List of vectors (this must be a pair).

Returns:

Cosine similarity value.

Return type:

float

features.discursive_diversity.get_unique_pairwise_combos(lst)

Computes all unique pairwise combinations of the elements in a list.

Code sourced from: https://pubsonline.informs.org/doi/suppl/10.1287/mnsc.2021.4274

Parameters:

lst (list) – Array or list of elements.

Returns:

List of unique pairwise combinations of elements of the input list.

Return type:

list