Open Access

Table 4

Comprehensive review of recent works involving AI for chatbot implementation.

Sr no. Research paper Year Algorithm used Research findings
1. M. Nuruzzaman and O. K. Hussain, ‘A Survey on Chatbot Implementation in Customer Service Industry through Deep Neural Networks,’ [33] 2018 ANN Artificial Neural Network (ANN) owing to its capability to handle the complicated combination of features provides the most appropriate base to work upon for a problem statement such as Conversational Agents or Chatbots.
2. Lee, M. C., Chiang, S. Y., Yeh, S. C., & Wen, T. F. ‘Study on emotion recognition and companion Chatbot using deep neural network’ [21] 2020 RNN RNN provides a better response to problem statements about Seq2Seq framework of RNN built over Domain-Specific Knowledgebase.
3. Bali, M., Mohanty, S., Chatterjee, S., Sarma, M., & Puravankara, R. Diabot: ‘A Predictive Medical Chatbot using Ensemble Learning’ [24] 2019 Ensemble Learning Ensemble Learning as a meta-algorithm has the potential to provide better generalization. The increased performance can be mapped with strong correlations with a humane sense of conversation
4. Pathak, K., & Arya, A. ‘A Metaphorical Study Of Variants Of Recurrent Neural Network Models For A Context Learning Chatbot’ [34] 2019 LSTM LSTM is the most appropriate choice when the states of dialogues & responses in a conversation need to be tracked and predicted.
5. G. Dzakwan and A. Purwarianti, ‘Comparative Study of Topology and Feature Variants for Non-Task-Oriented Chatbot using Sequence to Sequence Learning,’ [35] 2018 GRU The Encoder-Decoder framework over GRU comes in as an alternative to LSTM for the chatbot. GRU offers similar if not better performance over LSTM when compared over the parameters like the size of the dataset, the resources being consumed.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.