Retrieval and Generative Approaches for a Pregnancy Chatbot in Nepali with Stemmed and Non-Stemmed Data : A Comparative Study

TL;DR

This study compares retrieval and generative NLP approaches for developing a Nepali pregnancy chatbot, analyzing the impact of stemmed and non-stemmed data on model performance, and identifies the most effective models for each approach.

Contribution

It provides a comparative analysis of retrieval and generative NLP models for a Nepali pregnancy chatbot, highlighting the effects of stemming on model performance.

Findings

DistilBERT achieved 0.9165 accuracy on non-stemmed data.

Transformer models performed better on stemmed data.

Generative model BLEU scores were 0.3570 (1-gram) and 0.1413 (2-gram).

Abstract

The field of Natural Language Processing which involves the use of artificial intelligence to support human languages has seen tremendous growth due to its high-quality features. Its applications such as language translation, chatbots, virtual assistants, search autocomplete, and autocorrect are widely used in various domains including healthcare, advertising, customer service, and target advertising. To provide pregnancy-related information a health domain chatbot has been proposed and this work explores two different NLP-based approaches for developing the chatbot. The first approach is a multiclass classification-based retrieval approach using BERTbased multilingual BERT and multilingual DistilBERT while the other approach employs a transformer-based generative chatbot for pregnancy-related information. The performance of both stemmed and non-stemmed datasets in Nepali language has been analyzed for each approach. The experimented results indicate that BERT-based pre-trained models perform well on non-stemmed data whereas scratch transformer models have better performance on stemmed data. Among the models tested the DistilBERT model achieved the highest training and validation accuracy and testing accuracy of 0.9165 on the retrieval-based model architecture implementation on the non-stemmed dataset. Similarly, in the generative approach architecture implementation with transformer 1 gram BLEU and 2 gram BLEU scores of 0.3570 and 0.1413 respectively were achieved.