Natural Language Processing and Text Mining 2024, group project 16 Eetu Hyypiö, Sami Karhumaa & Rainer Laaksonen
The program is provided as a jupyter notebook. The last task is also provided as python source file, since the notebook kernel may crash when executing the task. Should this happen, run task 8 with python task8.py
Steps to using the notebook:
- Clone the repository
- Install dependencies from requirements.txt
- Install desired spacy language models. Default is the English medium model, "en_core_web_md". This can be installed with: python -m spacy download en_core_web_md Other options are the small "en_core_web_sm" and large "en_core_web_lg" models. All three are used in the report.
- Download wordnet-1.6 from https://wordnet.princeton.edu/download/old-versions
- Download Wordnet Domains 3.2 from https://wndomains.fbk.eu/download.html
Make sure that you have included the "wnaffect.py" and "emotion.py" files and wn-domains-3.2 and wordnet-1.6 directories. These are used in the emotion and sentiment calculations. The dataset is saved in the "ijcnlp_dailydialog" directory. All other files are data files that the notebook blocks create from the Dailydialog dataset.