Implemented BERTopic Model for Accurate Topic Segmentation in Agriculture Dataset issue#291

[pmukesh31]

Fix for- #291

Aim:
Get an accurate list of topics (around 20 topics max) for an agri dataset of queries (has around 20k unique queries) using BERTTopic for the dataset

Description:

• Implemented BERTopic model to accurately segment the agriculture dataset into 20 distinct topics.
• Utilized the 'questioninEnglish' column containing approximately 20,000 unique queries for topic analysis.
• Successfully generated 20 topics using BERTopic, leveraging contextual embeddings from BERT for clustering.
• Used HDBSCAN model for BERTopic
• Plotted a Intertopic Distance Model alongside various other output graphs and barcharts.
• Open to suggestions for improving topic cluster evaluation and enhancing the clustering process.

Steps:

1)Read the csv file and take 'queryInEnglish' column into consideration
2)Preprocessing of data by removing stop words and commas.
3)Training BERTopic
4)Visualizing results
5)Saving Model

[GautamR-Samagra]

can raise this separately somewhere else as a repo and link it. this is not in the required ai-tools format