© SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

Conversational AI Unleashed: A Comprehensive Review

of NLP-Powered Chatbot Platforms
SHUBHAM CHAURASIA1, SHUBHA JAIN2, HARI OM VISHWKARMA3, NISHANT SINGH4
1
Assitant Professor (Computer Science & Engineering), Axis Institute of Technology & Management,
Kanpur
2
Professor (Computer Science & Engineering), Axis Institute of Technology & Management, Kanpur
3-4
Undergraduate Student (Computer Science & Engineering with Data Science), Axis Institute of
Technology & Management, Kanpur

Abstract- Recent advancements in Natural The diverse applications of NLP span across machine
Language Processing (NLP) have drastically translation, sentiment analysis, information retrieval,
changed how people and computers communicate. and more. However, one particularly transformative
This beginner-oriented research article explores application that has garnered significant attention is
NLP-based chat platforms, investigating how they NLP-based chat platforms.
work, what they can do, and the challenges they face.
The study looks at modern techniques for creating A. Overview of NLP and its Applications:
chatbots, including rule-based systems, generative NLP, at its core, seeks to equip machines with the
models, and retrieval methods. It dives into the ability to comprehend and interact with human
technologies driving these platforms, like pre-trained language – a feat that has long been a pinnacle of
language models and context management. Ethical technological achievement. Through the integration of
concerns like bias and privacy are also covered. The linguistics, computer science, and cognitive
article examines how these chat platforms affect psychology, NLP has unlocked the potential for
fields like customer service, healthcare, education, computers to not only decipher the semantic nuances
and entertainment. By summarizing existing of text and speech but also respond in a manner that
research, the article highlights current trends and simulates human-like conversations. This
future possibilities in NLP-based chat systems, transformative technology has already found utility in
offering insights to newcomers and interested areas such as virtual assistants, language translation,
parties. and sentiment analysis, bringing unprecedented
efficiency and depth to human-computer interaction.
Indexed Terms- Natural Language Processing
(NLP), Chat platforms, Conversational systems, B. Emergence and Importance of NLP-Based Chat
Chatbots, Pre-trained language models, Human- Platforms:
computer interaction, Ethical considerations, User The emergence of NLP-based chat platforms has
engagement. marked a turning point in the way individuals engage
with computers. These platforms leverage the power
I. INTRODUCTION of NLP to create dynamic and interactive
conversational systems, commonly known as chatbots
In the era of rapid technological advancement, the or virtual agents. These systems are no longer limited
field of Natural Language Processing (NLP) has to providing static responses; instead, they possess the
emerged as a groundbreaking discipline bridging the ability to understand context, engage in multi-turn
gap between human language and computational conversations, and generate coherent and contextually
systems. NLP, a subfield of artificial intelligence, relevant replies. As a result, NLP-based chat platforms
focuses on enabling computers to understand, have become a cornerstone in redefining user
interpret, and generate human language in a manner experiences across various sectors, from customer
that resonates with human communication patterns. service and healthcare to education and entertainment.

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 1

 © SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

C. Purpose and Scope of the Review Article: This involves classifying user input into predefined
This review article aims to comprehensively explore categories, enabling the chatbot to discern user
the landscape of NLP-based chat platforms, delving goals and respond accordingly.
into their architectural intricacies, underlying
technologies, ethical considerations, and impact across B. Dialogue Management:
diverse domains. By synthesizing current research, Effective conversation necessitates more than isolated
trends, and developments, this article seeks to provide responses; it requires coherent exchanges that consider
a holistic understanding of the advancements in NLP- the evolving context. The Dialogue Management
based chat systems. Furthermore, the review will shed component orchestrates these interactions:
light on the challenges that researchers and 1. Context Tracking and Preservation: Dialogue
practitioners encounter in designing, deploying, and Management maintains a contextual memory of
maintaining these systems, while also proposing ongoing conversations. It enables the chatbot to
potential future directions for innovation. recall past interactions, ensuring seamless
continuity across multiple user turns and
As we embark on this journey into the realm of NLP- preventing information loss.
based chat platforms, we unravel the intricate interplay 2. Multi-turn Conversation Handling: With the
between language, cognition, and technology. ability to retain historical context, chat platforms
Through an in-depth analysis of the state-of-the-art engage in multi-turn dialogues. This entails
techniques and their practical applications, we aim to understanding the progression of the conversation,
provide valuable insights for researchers, developers, anticipating user needs, and crafting relevant
and stakeholders who are keen to harness the replies that align with the ongoing discourse.
transformative potential of NLP to reshape the way we
interact with machines. C. Natural Language Generation (NLG):
Generating responses that not only make sense but
II. ARCHITECTURAL COMPONENTS OF also emulate human-like language is a hallmark of
NLP-BASED CHAT PLATFORMS proficient chat platforms. The Natural Language
Generation (NLG) component focuses on producing
NLP-based chat platforms are intricate systems that coherent and contextually appropriate replies:
seamlessly blend linguistic understanding, 1. Content Creation and Language Generation
computational reasoning, and dynamic response Techniques: NLG employs techniques ranging
generation. This section uncovers the core from template-based generation to more advanced
architectural components that underpin the approaches involving neural networks. This
functionality of these platforms. component tailors responses by combining
extracted information, context, and stylistic
A. Natural Language Understanding (NLU): elements.
In the heart of every NLP-based chat platform lies the 2. Response Coherence and Fluency: A prime
Natural Language Understanding (NLU) component, objective of NLG is to ensure that generated
responsible for deciphering the intricacies of user responses are not only accurate but also read
input and translating it into actionable insights. Key naturally. This involves maintaining consistent
elements of NLU include: tone, structure, and vocabulary throughout the
1. Semantic Analysis and Entity Recognition: NLU conversation to enhance the user's conversational
employs advanced techniques to parse user experience.
queries, identifying entities such as names, dates,
and locations. This semantic understanding allows III. LITERATURE REVIEW
chat platforms to grasp the context and relevance
of user requests accurately. The evolution of Natural Language Processing (NLP)
2. Intent Detection and Classification: Through technologies has led to remarkable advancements in
sophisticated machine learning algorithms, NLU the realm of human-computer interaction. NLP-based
determines the underlying intent of user messages. chat platforms, also known as chatbots, represent a

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 2

 © SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

pivotal application of these technologies, transforming chatbots automate responses, reducing response
the way individuals interact with technology. This time and improving user satisfaction (Hussain et
section reviews the existing literature on NLP-based al., 2020). Healthcare chatbots provide medical
chat platforms, encompassing their architecture, information and guidance (Denecke et al., 2019),
capabilities, challenges, and applications. while educational chatbots offer personalized
• NLP-based chat platforms comprise various tutoring and support (Rani et al., 2021).
architectural components that enable them to Additionally, chatbots serve as companions and
understand and generate human language in entertainment sources in the entertainment sector
conversations. Natural Language Understanding (Zhang et al., 2020).
(NLU) is a critical component that involves • Despite the progress made, challenges persist.
semantic analysis, intent detection, and entity Variability in user input, understanding nuanced
recognition (Joulin et al., 2016). Dialogue language, and ensuring user satisfaction remain
management, on the other hand, focuses on critical areas for improvement. Future research
tracking conversation context and ensuring should focus on creating emotionally intelligent
coherent multi-turn interactions (Serban et al., chatbots that can comprehend and respond
2017). appropriately to user emotions (Devault et al.,
• Chat platforms are built using different 2014).
approaches, including rule-based systems,
generative models, and retrieval-based methods. IV. TYPES OF NLP-BASED CHAT
Rule-based systems employ predefined rules to PLATFORMS
generate responses (Shawar & Atwell, 2007).
Generative models, exemplified by GPT-style NLP-based chat platforms encompass a spectrum of
language models (Radford et al., 2019), can methodologies that define their conversation strategies
generate contextually relevant responses. and capabilities. This section dissects three distinct
Retrieval-based methods leverage predefined types of chat platforms, each characterized by its
response databases to select appropriate replies underlying approach and unique characteristics.
(Wu et al., 2017).
• Pre-trained language models like BERT (Devlin et A. Rule-Based Systems:
al., 2018) and GPT (Brown et al., 2020) have Rule-based chat platforms rely on explicit sets of
significantly enhanced chatbots' capabilities. predefined rules and patterns to interpret user input
These models facilitate transfer learning and fine- and generate responses. These platforms operate on
tuning, enabling chatbots to generate coherent and structured logic to determine appropriate replies.
contextually accurate responses. Semantic 1. Syntax-Based Rule Engines: In this approach, the
understanding, enabled by techniques such as chatbot identifies specific keywords or patterns in
named entity recognition and semantic role user input to trigger corresponding responses.
labeling, further improves dialogue These responses are often predefined and follow a
comprehension (Hakkani-Tür et al., 2016). fixed script based on recognized syntax.
• As NLP-based chat platforms become more 2. Limitations and Challenges: While rule-based
prevalent, ethical considerations become systems offer a controlled and deterministic
paramount. Bias mitigation strategies are crucial to conversational experience, they struggle with
ensure fairness and avoid perpetuating biases nuances, context, and ambiguity present in natural
present in training data (Bolukbasi et al., 2016). language. Their responses are limited to predefined
Privacy concerns necessitate responsible handling scenarios and cannot adapt well to unexpected or
of user data, and transparency in AI decision- complex user queries. Additionally, maintaining
making is imperative to establish trust and updating the rule sets can be labor-intensive
(Diakopoulos et al., 2016). and hinder scalability.
• NLP-based chat platforms have a profound impact
across various domains. In customer service,

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 3

 © SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

B. Generative Models: creativity, and retrieval-based methods provide

Generative models take a more dynamic approach by accuracy. The dynamic interplay of these types
using machine learning techniques to generate underscores the complexity and diversity of NLP-
responses. These models learn patterns from large driven conversational technologies.
datasets and produce contextually relevant and diverse
replies. V. TECHNOLOGIES EMPOWERING NLP-
1. Seq2Seq Architectures: Sequence-to-sequence BASED CHAT PLATFORMS
(Seq2Seq) models, often implemented using
Recurrent Neural Networks (RNNs) or NLP-based chat platforms derive their sophistication
Transformers, can map input sequences (user from a rich amalgamation of cutting-edge
messages) to output sequences (chatbot technologies. This section unveils the foundational
responses). They excel in capturing complex technologies that empower these platforms to
relationships in language and are capable of comprehend, contextualize, and generate human-like
generating coherent and contextually aligned conversations.
replies.
2. GPT-Style Language Models: Inspired by the A. Pre-trained Language Models:
"Generative Pre-trained Transformer" (GPT) Pre-trained language models have emerged as a
series, GPT-style models employ a single large transformative force, enabling chat platforms to
neural network to predict and generate text based leverage a wealth of linguistic knowledge and
on context. They are particularly skilled in open- patterns. These models have fundamentally reshaped
domain conversations and can produce creative the landscape of natural language understanding and
and contextually relevant responses. generation.
1. Transfer Learning and Fine-Tuning: Pre-trained
C. Retrieval-Based Methods: language models are initially trained on massive
Retrieval-based chat platforms store a repository of text corpora to learn general linguistic patterns.
pre-defined responses and match incoming user Chat platforms then fine-tune these models on
queries to the most appropriate stored response. These domain-specific or task-specific data, optimizing
platforms excel at providing accurate and contextually their performance for more targeted applications.
relevant information. 2. Role in Generating Contextually Relevant
1. Response Ranking and Selection: In this method, Responses: These models have an inherent
the system ranks and selects responses from a understanding of syntax, semantics, and context
database based on their relevance to the user's due to their training on diverse text sources. This
query. This relevance is often determined by knowledge equips them to generate responses that
measuring similarity between user input and stored align with the conversational context, contributing
responses. to coherent and meaningful interactions.
2. Hybrid Approaches for Improved Performance:
Recognizing the strengths and weaknesses of B. Semantic Understanding:
individual methods, hybrid approaches combine Semantic understanding technologies enable chat
generative and retrieval-based techniques. These platforms to delve beyond surface-level meanings,
systems leverage generative models to handle interpreting the deeper nuances and intentions
open-ended queries and rely on retrieval-based embedded in user input.
methods for factual and accurate information, 1. Named Entity Recognition and Disambiguation:
aiming to provide a balanced and versatile Chat platforms employ Named Entity Recognition
conversation experience. (NER) to identify entities like names, dates, and
locations in user queries. By disambiguating these
The choice of NLP-based chat platform depends on entities, platforms ensure accurate comprehension
the desired trade-offs between control, flexibility, and more precise responses.
context sensitivity, and scalability. While rule-based 2. Semantic Role Labeling for Deeper
systems offer structure, generative models unleash Comprehension: Semantic Role Labeling (SRL)

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 4

 © SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

aids in understanding the relationships between mitigate biases related to gender, race, and other
words in a sentence. This enables platforms to sensitive attributes.
grasp the roles different entities play in an action 2. Fairness in Response Generation across Diverse
or event, enhancing the depth of understanding and User Groups: Ensuring fairness involves crafting
generating more contextually relevant replies. responses that are culturally sensitive, neutral, and
unbiased across diverse user groups. Chatbot
C. Dialogue Context Management: outputs should be scrutinized to prevent
Efficient conversation necessitates the management of reinforcing harmful stereotypes or inadvertently
dialogue context – the memory of past interactions that marginalizing certain users.
guides current and future responses.
1. Memory Networks and Context Embedding: B. Privacy and Data Security:
Memory networks serve as a form of external 1. Handling Sensitive User Information: NLP-based
memory, allowing chat platforms to retain relevant chat platforms often process personal and sensitive
information from previous turns. This enables information. Developers must implement robust
platforms to retrieve historical context and weave encryption and data anonymization techniques to
it into ongoing conversations, enhancing protect user data from unauthorized access or
continuity and coherence. breaches.
2. Long-Term and Short-Term Context Fusion: 2. Consent and Data Protection Measures: User
Dialogue context management involves fusing consent is paramount. Platforms should be
both short-term context (current conversation) and transparent about data collection and usage,
long-term context (persistent knowledge) to obtaining explicit consent from users to ensure
produce informed responses. The integration of ethical data handling. Adhering to data protection
these contexts facilitates meaningful interactions regulations such as GDPR is essential to safeguard
and prevents repetition or information loss. user privacy.

The intricate synergy of pre-trained language models, C. Responsible AI Deployment:

semantic understanding technologies, and context 1. Transparency and Explainability of Chatbot
management mechanisms forms the bedrock of NLP- Decisions: Users have the right to understand how
based chat platforms. This technological convergence chatbots arrive at their responses. Employing
empowers chatbots to navigate the intricacies of explainable AI techniques allows users to
language, decode user intentions, and create engaging comprehend the rationale behind chatbot
conversations that mirror human interaction. decisions, promoting trust and accountability.
2. Mitigating Potential Negative Societal Impacts:
VI. ETHICAL CONSIDERATIONS IN NLP- NLP-based chat platforms can inadvertently
BASED CHAT PLATFORMS propagate misinformation, amplify hate speech, or
escalate harmful content. Developers must
As NLP-based chat platforms become integral to our proactively implement content moderation
digital lives, addressing ethical concerns is paramount. mechanisms and engage in ongoing monitoring to
This section delves into the critical ethical curtail negative societal consequences.
considerations that guide the development,
deployment, and utilization of these platforms. Navigating these ethical considerations is imperative
for the responsible development and deployment of
A. Bias Mitigation: NLP-based chat platforms. As these platforms play an
1. Identifying and Addressing Biases in Training increasing role in shaping human-computer
Data: Biases present in training data can lead to interactions, it is essential to prioritize fairness,
discriminatory or prejudiced responses from chat privacy, and societal well-being to foster a positive
platforms. To counter this, developers must and inclusive digital environment.
rigorously audit and preprocess training data to

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 5

 © SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

VII. IMPACT AND APPLICATIONS OF NLP- and fosters a deeper understanding of the subject
BASED CHAT PLATFORMS matter.

NLP-based chat platforms have ushered in a new era D. Entertainment and Social Interaction:
of human-computer interaction, with far-reaching 1. Chatbots as Companions and Entertainment
implications across various domains. This section Sources: NLP-based chat platforms have found a
explores the transformative impact and diverse niche in providing companionship and
applications of these platforms. entertainment. They can engage users in casual
conversations, tell jokes, share stories, and offer a
A. Customer Service and Support: sense of companionship.
1. Automating Customer Interactions: NLP-based 2. Challenges in Creating Emotionally Intelligent
chat platforms are revolutionizing customer Chatbots: Developing emotionally intelligent
service by automating routine queries and tasks. chatbots that understand user emotions and
This not only reduces the workload on human respond appropriately is a challenge. While
agents but also allows them to focus on more progress has been made, creating chatbots with
complex and specialized issues. genuine emotional understanding remains an
2. Improving Response Time and User Satisfaction: ongoing pursuit.
Chatbots provide instant responses around the
clock, significantly enhancing response times. This The versatile applications of NLP-based chat
results in increased user satisfaction, as customers platforms have redefined how individuals interact with
receive timely assistance, driving positive technology. From revolutionizing customer service to
customer experiences. enhancing healthcare accessibility, education, and
entertainment, these platforms are shaping a future
B. Healthcare Assistance: where seamless human-computer interactions are the
1. Providing Medical Information and Guidance: norm. However, with their transformative potential
NLP-based chat platforms offer a user-friendly comes the responsibility to ensure accuracy, ethics,
channel for individuals seeking medical and a positive impact across all spheres of application.
information. These platforms can provide accurate
answers to medical queries, reducing the need for VIII. CHALLENGES AND FUTURE
users to rely solely on search engines. DIRECTIONS
2. Ensuring Accuracy and Ethical Considerations:
While these platforms enhance access to healthcare While NLP-based chat platforms have made
information, accuracy is critical. Developers must remarkable strides, they face a set of challenges that
ensure that responses are based on reliable sources offer valuable insights for future advancements. This
and follow ethical guidelines to prevent section delves into these challenges and outlines
misinformation. potential directions for continued innovation.

C. Education and Learning: A. Handling User Input Variations:

1. Personalized Tutoring and Assistance: NLP-based NLP-based chat platforms encounter a wide array of
chatbots are transforming education by offering user input variations, from slang and colloquialisms to
personalized learning experiences. These typos and grammatical errors.
platforms can adapt to individual learning styles, • Challenges: The platforms must be equipped to
provide explanations, and assist students in their understand and process diverse language styles
studies. accurately, even in scenarios where inputs deviate
2. Gamification and Engagement in Education: from standard grammatical rules.
Chatbots can infuse gamification elements into • Future Directions: Developing robust models that
education, making learning more interactive and can handle linguistic variations, integrate context
engaging. This approach boosts student motivation effectively, and correct errors will be crucial for

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 6

 © SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

improving the accuracy and responsiveness of chat powered chat systems will guide their evolution in
platforms. a socially responsible manner.

B. Understanding Nuanced Language and Context: As NLP technologies advance, addressing these
Human communication often involves subtle nuances, challenges will be pivotal in shaping the future of chat
sarcasm, and context-dependent meanings that are platforms. By aligning research, development, and
challenging for machines to interpret accurately. ethical considerations, these platforms can evolve into
• Challenges: Chat platforms struggle with grasping tools that not only provide efficient and meaningful
the intricate layers of meaning, cultural references, interactions but also positively impact various aspects
and implied sentiments embedded in human of human society.
language.
• Future Directions: Advancements in sentiment CONCLUSION
analysis, contextual understanding, and cross-
cultural communication models are essential for In the dynamic landscape of technological innovation,
enabling chatbots to engage in more nuanced and NLP-based chat platforms stand as a testament to the
contextually accurate conversations. remarkable strides made in bridging the gap between
human language and artificial intelligence.
C. Enhancing User Satisfaction and Engagement: Throughout this exploration, several key findings have
Creating a seamless and engaging user experience is emerged, underscoring the profound impact and
paramount for the success of chat platforms. intricate nature of these conversational systems.
Maintaining user interest throughout a conversation
and avoiding repetitive responses is a persistent A. Summary of Key Findings:
challenge. The review has unveiled the intricate architecture that
• Challenges: Sustaining user engagement over powers NLP-based chat platforms, from the core
extended interactions requires continuously components of natural language understanding,
generating diverse and meaningful responses that dialogue management, and natural language
cater to user preferences. generation to the diverse types of platforms that
• Future Directions: Innovations in dialogue encompass rule-based systems, generative models,
generation, reinforcement learning, and adaptive and retrieval-based methods. We have witnessed the
learning mechanisms will drive the development of pivotal role of technologies such as pre-trained
chatbots that can dynamically adapt their tone, language models, semantic understanding, and context
style, and content to keep users engaged. management in shaping the capabilities of these
platforms.
D. Ethical and Societal Implications in a Broader
Context: B. Emphasis on the Transformative Potential of NLP-
As NLP-based chat platforms become more integrated Based Chat Platforms:
into society, ethical considerations around bias, The transformative potential of NLP-based chat
privacy, and societal impact gain prominence. platforms is undeniable. They have revolutionized
• Challenges: Striking a balance between free customer service, empowered healthcare assistance,
expression and responsible content moderation, personalized education, and added an interactive layer
ensuring unbiased responses, and addressing the to entertainment. These platforms not only offer
potential loss of jobs in customer service due to efficient automation but also foster meaningful
automation are key ethical challenges. human-computer interactions, enhancing user
• Future Directions: Developing advanced experiences across various domains.
mechanisms for bias detection and mitigation,
implementing robust privacy controls, and creating C. Call for Continued Research and Responsible
policies for the responsible deployment of AI- Development:
As we peer into the future, the journey of NLP-based
chat platforms is far from over. Challenges in handling

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 7

 © SEP 2023 | IRE Journals | Volume 7 Issue 3 | ISSN: 2456-8880

user input variations, understanding nuanced understanding. arXiv preprint

language, and ensuring ethical considerations persist. arXiv:1810.04805.
This calls for ongoing research to refine models, [7] Brown, T. B., Mann, B., Ryder, N., Subbiah, M.,
enhance accuracy, and tackle biases. Furthermore, a Kaplan, J., Dhariwal, P., ... & Amodei, D. (2020).
strong emphasis on responsible development is vital to Language models are few-shot learners.
ensure these technologies contribute positively to Advances in Neural Information Processing
society. Systems, 33.
[8] Hakkani-Tür, D., Tur, G., Celikyilmaz, A., Chen,
In closing, the trajectory of NLP-based chat platforms Y. N., Gao, J., Deng, L., & Wang, Y. Y. (2016).
exemplifies the fusion of technology and human Multi-domain joint semantic frame parsing using
communication, reshaping the way we interact with bi-directional RNN-LSTM. In Proceedings of the
machines. The potential to create more empathetic, 2016 Conference on Empirical Methods in
efficient, and versatile chatbots is boundless, but it Natural Language Processing (EMNLP) (pp.
must be accompanied by an unwavering commitment 1162-1171).
to ethical principles and responsible innovation. As [9] Bolukbasi, T., Chang, K. W., Zou, J. Y.,
researchers, developers, and stakeholders, the journey Saligrama, V., & Kalai, A. T. (2016). Man is to
ahead is marked by both challenges and opportunities, computer programmer as woman is to
inviting us to harness the transformative power of homemaker? Debiasing word embeddings. In
NLP-based chat platforms for a brighter technological Advances in neural information processing
future. systems (pp. 4349-4357).
[10] Diakopoulos, N., Friedler, S., Arenas, M.,
REFERENCES Barocas, S., Hay, M., Howe, B., ... &
Zevenbergen, B. (2016). Principles for
[1] Joulin, A., Grave, E., Bojanowski, P., Mikolov, accountable algorithms and a social impact
T., Bagherinezhad, M., & Others. (2016). Bag of statement for algorithms. Proceedings of the
tricks for efficient text classification. arXiv 2016 CHI Conference on Human Factors in
preprint arXiv:1607.01759. Computing Systems, 813-824.
[2] Serban, I. V., Sordoni, A., Lowe, R., Charlin, L., [11] Hussain, M. S., Al-Olimat, H. S., Al-Rifai, S., &
Pineau, J., Courville, A., & Bengio, Y. (2017). A Ghawanmeh, M. (2020). A survey of chatbot
hierarchical latent variable encoder-decoder systems: Current progress, future prospects, and
model for generating dialogues. In Thirty-First open issues. IEEE Access, 8, 18620-18638.
AAAI Conference on Artificial Intelligence. [12] Denecke, K., Deng, Y., & Huang, L. (2019).
[3] Shawar, B. A., & Atwell, E. (2007). Chatbots: Biomedical chatbots: Analyzing the potential
Are they really useful? LiC - Lingvisticae and privacy concerns. Stud Health Technol
Investigationes, 30(2), 173-186. Inform, 264, 1184-1185.
[4] Radford, A., Wu, J., Child, R., Luan, D., [13] Rani, P., Sharma, A., & Kumar, A. (2021).
Amodei, D., & Sutskever, I. (2019). Language Design and implementation of personalized
models are unsupervised multitask learners. educational chatbot. In 2021 12th International
OpenAI Blog, 1(8), 9. Conference on Computing, Communication and
[5] Wu, Y., Wu, W., & Zhu, X. (2017). A sequential Networking Technologies (ICCCNT) (pp. 1-6).
matching framework for multi-turn response [14] Zhang, X., Zhao, J., LeCun, Y., & Ranzato, M.
selection in retrieval-based chatbots. In (2020). Text-to-text neural generative models. In
Proceedings of the 55th Annual Meeting of the Proceedings of the 28th International Conference
Association for Computational Linguistics (Vol. on Computational Linguistics (pp. 3452-3462).
1, pp. 496-505).
[6] Devlin, J., Chang, M. W., Lee, K., & Toutanova,
K. (2018). BERT: Pre-training of deep
bidirectional transformers for language

IRE 1705011 ICONIC RESEARCH AND ENGINEERING JOURNALS 8