CONTEXTUAL AI ASSISTANT USING TRANSFORMERS AND NLP

Computer Science and Technology

Building proper answering of the questions that can communicate naturally with humans is a difficult and interesting problem in real world. Rapid advancement in this field is typically constrained by the persistent issue of lack of data. As a result of the expectation that these systems will learn grammar, reasoning, decision-making, and syntax from insufficiently large task-specific datasets. Pre-trained language models, recently presented have the potential to fill the data gap and generate contextualized word embeddings with great benefits. Considered the NLP version of ImageNet, these models have been shown to capture different aspects of language, including hierarchical relationships, long-term dependencies, and emotions. Recent developments in the field of pretrained language models are discussed. We also consider how to use the strengths of these language models to design more engaging and eloquent conversational agents. As a result, the purpose of this work is to explore whether these pre-trained models can address the issues that responding systems face as well as how their architecture can do so.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.