SuzannaWentzel / Sentiment-Alignment-Interplay

Code implementation of Time-based overlap, and notebook for studying interplay of sentiment & alignment.

Thesis: Alignment, Agreement, and Sentiment in online discussions

Contains the data processing of my Master Thesis and example notebooks for further reference

Running the project

Data should be added to a Data folder, add the path to the datapath variable in preprocessing_notebook.py. Data should be in .csv and have the following columns (see also dummy_data.csv):

• discussion_id
• topic
• post_id
• text
• username
• parent_post_id
• parent_missing

After adding the data, the following notebooks can be used step by step to investigate the interplay between sentiment and alignment:

1. Preprocessing => preprocessing.py
2. Alignment analysis => compute_lexical_word_alignment_v2_all.py
3. Sentiment analysis => sentiment_analysis.py
4. Interplay analysis => interplay_analysis.py

Cleaning up the code is still a work in progress, but will appear here:

1. Preprocessing => preprocessing_notebook.py
2. Alignment analysis => lexical_alignment_notebook.py
3. Sentiment analysis => sentiment_notebook.py
4. Interplay analysis => interplay_alignment_sentiment_notebook.py

The code for the time-based-overlap lives in linguistic_alignment_analysis/time_based_overlap.py

Dependencies

This repo needs the following packages (including their dependencies):

• matplotlib
• nltk
• numba
• numpy
• pandas
• stanza
• scipy
• transformers