NLP Assignment Datasets

The associated files are embeddings of risk sections of 10K filings on paragraph level. Here, there are three embedding spaces. Once, Word2Vec embeddings with average-pooling for paragraph level, once Word2Vec embeddings with hierarchical pooling on paragraph level, and BERT embeddings. 

Furthermore, an LDA model trained on these risk sections is uploaded.Lastly, the 10K filings from 2024 and 2025 are uploaded as a .zip file, as well as the python function to load them and convert them to a dataframe, that can later be merged with the embeddings. The function is called `load_json` in the file `tenk_functions.py`.