This project compares audit log summaries generated by a lightweight LLM (HuggingFace Transformers) vs traditional NLP (spaCy). Cosine similarity is computed and visualized using heatmaps.
Audit_Log_Insights.ipynb: Main notebook for preprocessing, summarization, and visualizationaudit_similarity_enriched.csv: Enriched dataset with similarity scoresheatmap_audit_similarity.png: Visual heatmap comparing LLM vs spaCy summaries
- TF-IDF + Cosine Similarity
- HuggingFace
pipeline("summarization") - spaCy
en_core_web_smfor keyword extraction
nlp, audit, huggingface, spacy, heatmap, similarity
