Analytical and Sentiment based text generative chatbot
Sahil Sawant, Ankit Vishwakarma, Prerana Sawant, Prasenjit Bhavathankar · 2021
Chatbots have existed for a decade and they have been used for many objective specific tasks in the industrial area extensively. The rise of deep neural nets has enhanced further growth of these chatbots. The chatbots are used to understand the sentences and decipher the meaning and continue the conversations depending on the need but they don't capture the emotion of the users. While conversing with a user, especially for the industries which implement these chatbots, maintaining a good interpersonal relationship is everything. With a chatbot which not only provides answers to the user of it but also understands it on a sentimental level, this can be achieved. Through our paper we focus on building a chatbot which generates responses on the basis of the emotions of the users in order to recreate a more sympathetic and human-like aspect towards the user. We have used the feed forward neural network architecture that fits in perfectly for a fast and accurate response. We have used the knowledge on Natural Language Processing to pre-process the data in a suitable format. The model is trained to understand the emotion based on the sentences and understand the meaning with a dataset specifically curated by the authors of the paper. The accuracy achieved on the model was 93.45% with a response generation time of 50 milliseconds.