Cognitive AI framework: advances in the simulation of

human thought

Dr. Rommel Salas-Guerra

Professor of Cyberpsychology, AU University, San Juan,

Puerto Rico and Doctoral Program in Information Systems
Management, AGM University, Orlando, Florida, USA

2025
 Abstract

The Human Cognitive Simulation Framework constitutes a paradigmatic advance in integrating

human cognitive capabilities in artificial intelligence systems, optimizing personalization and
adaptability in human-AI interaction. This model merges contexts of short-term memory
(conversation context), long-term memory (interaction context), advanced cognitive processing,
and efficient knowledge management, guaranteeing contextual coherence during sessions and the
storage of persistent information, which will enhance personalization and continuity in future AI
interactions with the user. This framework emphasizes the relevance of a unified database that
synchronizes these contexts, ensuring accessibility and operational efficiency, highlighting the
incorporation of logical, creative, and analog processing modules like those of the left and right
hemispheres of the human brain, which serve to perform structured tasks and complex contextual
inferences. Therefore, in Cognitive AI, the dynamic updating of knowledge allows the integration
of information in real-time, enriching the pre-trained knowledge of the model and ensuring its
adaptability, being useful in particularly relevant applications such as education, human behavior
and knowledge management, promoting the development of adaptive learning platforms,
promoting tools for personalized psychological interventions and generating Assertive synthetic
knowledge. Although Cognitive AI can process large volumes of data and improve the user
experience, it faces challenges associated with scalability, mitigation of cognitive biases and
compliance with ethical standards, therefore, this framework establishes a robust basis for future
research, proposing topics such as continuous learning algorithms, sustainability and multimodal
adaptability in information processing devices, so that it can consolidate itself as a transformative
model in emerging disciplines.

Keywords: Cognitive AI, Memory Contexts (Short and Long Term), Dynamic Knowledge
Update, Advanced Cognitive Processing
1. Introduction to Cognitive AI

1.1. Definition

The emulation of human cognitive processing is one of the challenges faced by Artificial

Intelligence (AI), this paradigm focuses on an approach that goes beyond the traditional

capabilities of AI systems, such as processing textual, graphic or audible information by

answering queries; but also in understanding, learning, and adapting based on previous

interactions (Sereati et al., 2020). Therefore, Cognitive AI seeks to replicate human cognitive

functions such as memory, learning, and reasoning to achieve a more natural and effective

interaction with users, facilitating the development of a new stage of Human-Computer

interaction (Malyala et al., 2023).

1.2. The purpose and scope

The purpose of this Framework is focused on bridging the gap between traditional AI systems,

which operate primarily on predefined rules and patterns (Pandey & Bhat, 2023), and a more

advanced level of intelligence that simulates human cognitive capabilities; Cognitive AI seeks to

provide a structured basis for the standardization of this new type of architecture that allows AI

systems not only to access and use previously trained knowledge, but also to accumulate, update

and manage the knowledge acquired over multiple interactions, improving personalization and

adaptability in their responses (Zheng et al., 2023).

1.3. Basic principles of knowledge management and integration

Within this approach, three essential key concepts are highlighted to understand how it works,

which will be discussed in depth in the following sections, however each of them will be

described in the form of an introduction, we start with “the context of interaction” which acts as
a long-term memory, which stores and manages persistent information about the user's

interactions, allowing for continuous personalization and a deeper understanding of user

preferences (Fuente & Pousada, 2019).

On the other hand, we have the “key conversational context” concerning its functions like that of

short-term memory, retaining relevant information only during the current session to maintain

the coherence and flow of the dialogue (Marri, 2024). Finally, “the integration of knowledge” is

a fundamental aspect as it allows us to combine the initial training data of the model with live

interactions, enabling the AI system to evolve and adapt based on past experiences and new data

acquired (Liu et al., 2023).

2. Core Components of the Framework

2.1. User Interaction Module

The User Interaction Module aims to act as the main interface between the AI system and the

user, allowing data entry and conversation maintenance, among its main roles is to capture user

requests and feedback in real-time, providing a fluid and accessible experience (Meng et al.,

2024), therefore, this module will be responsible for initiating and managing interactions,

adapting to user behavior to improve communication and foster a friendly and efficient user

environment. Among the main features of this module, we propose two combined elements that

will provide a robust framework of security, continuity, and coherence of the user experience.

a) User authentication, which will guarantee the security and personalization of

interactions, the monitoring of the interaction that allows recording and analyzing usage

patterns.
 b) Session management, through which you can facilitate the management of multiple

conversations in an orderly manner.

2.2. Context of Conversation (Short-Term Memory)

Below, the conversation context will be explained in detail, which can be compared to some

fundamental characteristics of short-term memory, as it manages the temporal information that is

used during interaction sessions, this "context" focuses on retaining the details necessary for the

current conversation, ensuring that the system can respond in a coherent and contextualized way

(Marri, 2024), this type of memory is essential for maintaining fluidity in dialogue and

responding appropriately to user questions or comments in a limited time frame (Liu et al.,

2023).

A key feature of the “conversation context” is its re-establishment at the end of the session,

being a design aimed at guaranteeing the efficiency of the system, this because the retained

information is only useful during the current interaction and therefore the flow of data within this

context it includes the capture of immediate responses, real-time updates, and the user's

questions, allowing the system to maintain an up-to-date and contextual understanding of the

conversation (Finch & Choi, 2024).

2.3. Context of Interaction (Long-Term Memory)

The other type of memory is the "interaction context" which could be represented as the long-

term memory of the system, functioning as a repository of persistent information that stores user

preferences, historical data and previous interactions, this type of memory is crucial for the

personalization of the experience, as it allows the system to retain and use relevant information

throughout multiple sessions, this is because by maintaining details about the user's history, the
system can offer more accurate answers and adapt its behavior to better meet the user's

expectations and needs (Fuente & Pousada, 2019).

The storage of this memory is implemented in a structured database designed to ensure

persistence and efficient access, as well as a well-defined synchronization process which allows

important data from the "conversational context" to be transferred to long-term memory,

ensuring that relevant information remains available for future interactions and thus improving

the continuity of the user experience (Finch & Choi, 2024).

2.4. Knowledge Base Integration

"Pre-trained model knowledge" is based on training with large datasets containing general

information of the world and complex language patterns, this source of knowledge allows the AI

system to answer a variety of questions and perform natural language understanding tasks,

providing a solid foundation for interactions, therefore, the capabilities of the pre-trained model

include pattern recognition, logical reasoning, and the generation of contextually appropriate

responses (Durt & Fuchs, 2024).

On the other hand, "dynamic knowledge refreshing" refers to the ability of the system to

evaluate, store, and use new information acquired during user interactions, this process, known

as persistent memory updating, ensures that the system remains adaptable and can integrate new

data into its knowledge structure, improving the personalization and relevance of its responses

(Zheng et al., 2023).

2.5. Context Management Database

The "structure of a unified database" is a fundamental piece to manage both the conversational

context (short-term memory) and the interaction context (long-term memory), this database

allows for the efficient storage and retrial of key information, facilitating the updating and

synchronization of data as needed, unifying both contexts in a single database ensures that the

system can access all relevant information quickly and consistently (Finch & Choi, 2024).

For this reason, "state persistence mechanisms" include session persistence, which ensures that

data from a session can be maintained and potentially elevated to long-term memory if it is

determined to be relevant (Durt & Fuchs, 2024), and context synchronization which is the

process by which data from the current session is evaluated and transferred to long-term storage.

They optimize the retention of meaningful information.

2.6. Cognitive Processing Modules

Within this module we have "logical and analytical processing" which in the AI system is

similar to the functions of the left hemisphere of the human brain, which is responsible for

handling structured tasks that require sequential reasoning, data analysis, and language

processing (Sholihah, 2022). It is essential for performing logical operations and providing

precise and detailed answers that require a high level of technical and structured understanding.

On the other hand, "creative and pattern recognition processing" is comparable to the functions

of the right hemisphere, which manages tasks that require a broader understanding of the context,

allowing for the generation of creative responses and the interpretation of complex data (Aberg

et al., 2016). The system uses this type of processing to adapt to different contexts and make
inferences, offering answers that go beyond the literal and encompass richer and more nuanced

interpretations.

3. Data Flow and Process Logic

3.1. Interaction Flow

The interaction flow describes the structured process by which information is handled from the

moment the user initiates a query until the update of the system memories, it is important to

emphasize that this flow begins with the user's search input, where the request is initially

processed in the context of conversation (short-term memory), being here where the system

analyzes and responds based on the temporal data of the current session (Finch & Choi, 2024),

therefore, this management of the conversation context is essential to maintain the coherence and

relevance of the responses during the Human-AI interaction.

It is also important to consider that within this flow process, as the session progresses, an

evaluation of the relevance of the interaction in the long term is carried out; During this process,

the system determines if the information captured during the session should persist beyond the

current interaction, therefore, if under this premise the information is identified as relevant, it is

transferred to the context of interaction (long-term memory), thus ensuring that the system can

use this data in future interactions to improve personalization and user understanding. Below we

will explain these two concepts in detail (Fuente & Pousada, 2019; Liu et al., 2023).
3.1.2. Relevance Validation

As we explained earlier, once the system identifies information

potentially relevant, the system proceeds to validate it using methods based on algorithms and

predefined rules. For this, it applies a data weighting system, assigning a score to each element

of information according to its relative importance (Zheng et al., 2023).

If this score exceeds a set threshold, the information is classified as suitable to be stored in long-

term memory, another critical step is the evaluation of temporal persistence, which is to

determine if the data has a prolonged impact or is simply ephemeral. For example, a one-time

reminder of a meeting may not meet persistence criteria, while details related to recurring events

or personal preferences are often validated for long-term storage (Finch & Choi, 2024).

To better understand this process, it is important to understand that the validated information is

compared with the existing data in the context of interaction, so if it complements, updates, or

improves prior knowledge without redundancy, its relevance is confirmed, guaranteeing efficient

memory management (Meng et al., 2024).

3.1.3. Decision Flow for Persistence

In the case of persistence decisions, the system follows a structured flow to ensure that only

significant information is retained in long-term memory, this is because, during each session, the

captured information is initially stored in short-term memory, where it is subjected to an

exhaustive analysis using the aforementioned criteria (Torrentira, 2024).

In this analysis, the system determines whether the data meets the requirements to be retained, if

information is considered relevant and persistent, transferred to the context of interaction, and

integrated as part of long-term memory (Christoph & Fuchs, 2024). Conversely, if the data fails

to meet the established criteria, it is deleted after the session to prevent system overload and

preserve operational efficiency. This approach ensures that the system not only manages memory

effectively but also enhances the user experience by retaining only information that offers

meaningful long-term value.

3.2. Use of Knowledge

In a Cognitive AI system, knowledge is used through the combination of its accessibility and pre-

training, through its dynamic integration based on previous interactions, with pre-trained

knowledge being the fixed base of information that allows the AI system to have a solid starting

point which is key to providing immediate and reliable answers (Marri, 2024).

It is also important to emphasize that pre-trained knowledge has a limitation which is that it

cannot be updated automatically, therefore, it is important to consider that this knowledge is

derived from large volumes of data with which the model was initially trained, allowing it to

understand natural language and effectively answer a variety of general questions (Durt & Fuchs,

2024).

In addition to pre-trained knowledge we have dynamic knowledge which is the information that

is continuously adapted and updated with new interactions, which allows the AI system to

personalize and improve its responses over time, this type of knowledge is consulted and updated

in real-time as the system interacts with the user, ensuring that responses are increasingly

relevant and personalized (Zheng et al., 2023), with this updating process being a reinforcement
in the system's ability to learn from previous experiences and apply that learning in future

interactions, optimizing the personalization and accuracy of responses.

To conclude with this section we must mention that both types of knowledge interact

synergistically to optimize the responsiveness of AI systems, combining the robustness of a pre-

trained knowledge base, which offers a comprehensive and grounded understanding of language

and pre-established patterns, with the adaptability of dynamic knowledge, which allows

continuous updating and contextual personalization (Meng et al., 2024).

3.3. Decision Framework

Within the data flow the decision framework is a critical structure that guides the process of

identifying, evaluating, and storing the data generated during the interaction with the user, this

framework determines whether the information captured during the current session is significant

enough to be retained in long-term memory (Finch & Choi, 2024).

This process begins with the identification of relevant data, followed by an evaluation in which it

is decided whether that data meets the persistence criteria, for this reason if the decision is

affirmative, the information is transferred to the context of interaction, ensuring that it is

available for future consultation and use (Fuente & Pousada, 2019).

To represent this process clearly and visually, we have the following flowchart which describes

the complete cycle from user input, management in the conversation context, persistence

assessment, to long-term memory update, this decision framework also provides a transparent

approach to how data is handled, ensuring that the transition between short-term and long-term
memory is carried out efficiently and according to predefined criteria that improve the user

experience and the functionality of the system (Christoph & Fuchs, 2024; Torrentina, 2024).

Figure 1
Cognitive AI Data Flow. Own Creation

4. Cognitive Comparative Analysis

4.1. Human Cognitive Processes vs. AI Model Capabilities

This analysis seeks to understand how human cognitive processes and the capabilities of an AI

model provide a deep understanding of the similarities and differences in an information

management environment, as human beings and their cognitive processes such as perception,

reasoning, memory, and decision-making take place in the brain. which can learn from

experience, adapt to new situations, and apply accumulated knowledge to solve problems

effectively (Durt & Fuchs, 2024).

For this reason, this Framework aims to seek a deeper understanding of how Cognitive AI

technology can replicate biological functions through algorithms and models that mimic the

structure of a neural network, capturing complex patterns of data to respond intelligently and

adaptively (Zheng et al., 2023), below, a matrix that explains this comparative analysis.

Table 1
Benchmarking Matrix: Traditional AI and Cognitive AI. Own Creation

Process Traditional AI Cognitive AI

User Input The request is received, but The request is received and
it does not retain context. stored in short-term
memory.

Conversation Management Responds according to pre- Personalize the response

programmed patterns. based on previous
interactions in the same
session.

Data Storage Information is deleted at The information is

the end of the session. evaluated and, if relevant,
stored in long-term
memory.

Adaptability No personalization in AI enhances and adapts its

future sessions. responses in future
interactions.
Unlike what is explained with the human brain, in an AI model, the pre-trained model acts as the

brain nucleus, containing a vast repository of general knowledge and language patterns acquired

during its initial training with large amounts of data (Torrentira, 2024), this model allows the AI

to perform tasks such as language understanding, text generation and response to questions

consistently and quickly (Chen et al., 2024).

However, although this nucleus mimics some capabilities of the human brain, such as the

Logical reasoning and comprehension (Ye et al., 2023), are limited in their ability to improvise

outside of the patterns and data on which it was trained, unlike the natural flexibility and

adaptability of the human mind (Finch & Choi, 2024).

4.2. Analogies of Short-Term and Long-Term Memory

A useful comparison between AI models and the human brain is the functioning of memory,

because, in humans, short-term memory, or working memory, temporarily retains information to

process immediate tasks, such as remembering an address or following a conversation, this

storage capacity being crucial to maintain consistency in interactions and handle multiple pieces

of information at the same time (Marri, 2024), below, a matrix that explains this comparative

analysis.
Table 2
Benchmarking Matrix: Human Memory and AI Short-Term Memory. own creation

Aspect Human Memory AI Memory

(Short Term) (Short Term)

Information retention Temporarily retains Stores information relevant

information to the current session

Memory Duration Second to minutes During the session, it is

deleted at the end of the
session.

Storage capacity Limits around seven Depends on the capacity of

elements the system

Flexibility and Upgrade Flexible but susceptible to Updatable in real-time, but

interference no post-session retention

Using Interactions Process immediate tasks Maintain consistency in the

such as remembering an conversation
address

Multitasking Can handle multiple items Processes multiple requests

simultaneously with effort based on model capacity

However, on the other hand, long-term memory will allow human beings to store information in

a more lasting way while being involved in learning and accumulating knowledge over time,

which allows people to apply past experiences to solve future problems (Fuente & Pousada,

2019).

In an AI system, these concepts are reflected in the methods called "Contexts", being the context

of conversation, the analogy of short-term memory, and the context of interaction of long-term
memory. To better understand these concepts, they will be explained below. In the case of the

conversational context, it retains data for the duration of a session, like how working memory

handles information temporarily. Once the session ends, this memory is restarted unless relevant

information is identified that must be stored in long-term memory (Liu et al., 2023).

When stored in long-term memory, the interaction context comes into play, which functions like

long-term memory, storing accumulated data that can be consulted and used in future

interactions, which allows AI to improve its ability to personalize and retain important

information persistently (Meng et al., 2024).

Finally, we must emphasize that, although this analogy is of utmost relevance in Cognitive AI as

it will help us understand how AI models are designed to mimic the structure of human

cognition, there are some limitations in terms of depth and flexibility in adaptive learning that

should be explored in the future (Christoph & Fuchs, 2024).

5. Technical Implementation and Infrastructure

5.1 Tech Stack

The technological stack of a Cognitive AI framework is made up of various tools and

technologies designed to handle and process large volumes of data efficiently, among them are

structured databases which are essential to store and manage both the interaction context and the

conversation context, ensuring the persistence of data and its rapid recovery (Marri, 2024).

These databases include certain technologies such as PostgreSQL, NoSQL, and MongoDB,

which offer flexibility in managing unstructured and structured data

(Aguilar Vera et al., 2023), It is also important to emphasize the role of cloud infrastructure, such

as that provided by AWS, Microsoft Azure, or Google Cloud, which facilitate the scalable

storage and processing capacity necessary to handle complex AI operations, allowing fast and

secure access to computing resources (Torrentira, 2024).

On the other hand, we have data management which is supported by container orchestration

technologies, such as Kubernetes, which ensure efficiency in the management of applications

and their deployment, in addition, to these technologies we have real-time data processing tools

and batches, such as Apache Kafka and Spark, which guarantee the updating and synchronization

of the long-term and short-term memory of AI (Aruna, & Gurunathan, 2024).

As we have been able to observe in the discussions, this technological stack allows us to

understand in an easier and summarized way how a robust and scalable environment works,

which can allow a Cognitive AI system to adapt to the workload and requirements in human-AI

interaction (Meg et al., 2024).

5.2. Scalability and Performance Considerations

Scalability plays a key role in the implementation of a Cognitive AI framework, as it allows the

system to handle an increase in the number of users or the complexity of queries without losing

performance, to achieve this, distributed architectures are implemented that use the capacity for

horizontal scaling, i.e., the addition of more nodes to the system to distribute the workload

(Zheng et al., 2023).

This information management strategy is complemented by database optimization through

partitioning techniques and the use of caching to accelerate data retrieval through the use of
monitoring tools, such as Prometheus and Grafana, which are used to monitor system

performance and detect bottlenecks in real-time (Chieu & Zeng, 2008).

Performance is addressed by optimizing processing algorithms and implementing AI models that

minimize latency in the generation of responses, which must be able to perform fast inferences,

even when accessing large volumes of long-term memory data, another complement is

parallelization techniques, and the implementation of GPUs or TPUs (processors designed for

machine learning operations) which help maintaining efficiency in processing large amounts of

data and simultaneous tasks, ensuring a smooth, and seamless user experience (Pandey & Bhat,

2023).

5.3. Data Security and Privacy Protocols

As for data security and privacy, this is an issue of relevance in a system of Cognitive AI. Due to

the handling of personal and sensitive data, the protocols that must be used guarantee the

protection of information in all stages of data storage and transfer (Torrentira, 2024). In addition,

it is important to emphasize that encryption technologies, such as TLS (Transport Layer

Security), ensure the integrity of data during transmission, while techniques such as data-at-rest

techniques, such as AES (Advanced Encryption Standard), protect the information stored in

databases (Zhang et al., 2023; Hazra et al., 2024), for this reason, the authentication and

authorization of users through access control systems, such as OAuth and two-factor

authentication (2FA) protocols, strengthen the security of interactions and access to the system

(Vilasini et al., 2024).

Data privacy is ensured by implementing policies in compliance with data protection regulations,

such as the GDPR (General Data Protection Regulation) and CCPA (California Consumer
Privacy Act), with these policies being clear rules on how personal data should be collected,

processed, and stored (Bakare et al., 2024).

Furthermore, data anonymization and pseudonymization techniques are also integrated to protect

the user's identity in long-term memory storage and during persistent memory updates. These

approaches ensure that AI can operate safely and ethically, protecting user information while

optimizing the personalization and relevance of responses (Christoph & Fuchs, 2024).

6. Applications and Use Cases

6.1. Practical Implementations

The practical applications of a Cognitive AI framework are wide, highlighting its impact on

sectors such as education and human behavior. In the educational field, Cognitive AI can make it

possible to create adaptive learning platforms that respond to the individual needs of students.

These platforms could adjust the content and level of difficulty based on progress and learning

style, acting as personalized tutors that provide real-time feedback and specific

recommendations, thus improving the effectiveness of the educational process (Marri, 2024).

On the other hand, in cyberpsychology, Cognitive AI can facilitate personalization in digital

psychological and therapeutic interventions, since these systems can adapt to the emotions,

behaviors, and needs of the user in real-time, offering emotional support, early detection of

psychological problems, and personalized guides for well-being (Liu et al., 2023).

Therefore, these two application examples demonstrate how Cognitive AI has the potential to

transform key sectors through personalization and adaptability, significantly improving the user

experience, as is the case with personalized assistants based on Cognitive AI, improving
traditional assistants, through a deeper understanding in the management of user preferences

over time to offer a more natural interaction and relevant (Meng et al., 2024).

Therefore, today and in a holistic way, Cognitive AI can also be a tool applicable to business

decision-making, by processing large volumes of data, identifying patterns and generating

recommendations to optimize business strategies, improving customer service and automating

complex processes, to obtain efficiency, by reducing costs and enabling organizations to adapt

quickly to changes in the work environment.

6.2. Advantages and Limitations

The strengths in this research work were able to identify some advantages of Cognitive AI

especially in its ability to offer a high level of personalization and adaptability, due to the

integration of long-term and short-term memory, since these systems can learn from past

interactions and apply that knowledge to improve future responses and decisions (Zheng et al.,

2023).

Bringing this customization to a more satisfactory user experience, since the system adapts and

responds more precisely to individual needs, with the capabilities that AI has to process and

analyze large volumes of data in real-time time allowing faster and more informed decision-

making.

However, some inherent limitations and challenges have been identified in this study that must

be considered, such as the handling of large amounts of data, which requires advanced

infrastructure to repel scalability and efficiency problems, in addition to one of the most

important challenges that is the risk of cognitive biases, which can arise if the training data or
previous interactions contain implicit biases which could influence the decision-making of AI

and affect the fairness and objectivity of its responses (Fuente & Pousada, 2019).

Therefore, to mitigate these risks, it is critical to implement training and evaluation strategies that

promote transparency and reduce bias, as well as ensure that AI operates ethically and complies

with privacy and data protection regulations, such as GDPR and CCPA (Bakare et al., 2024).

7. Conclusion and Future Prospects

7.1. Summary of the potential of Cognitive AI

The Cognitive AI framework represents a significant advance in the integration of technologies

that mimic and optimize human cognitive abilities. by combining short- and long-term memory

with a structured storage infrastructure and advanced cognitive processing, this framework

allows a more natural and personalized interaction between the user and the AI.

It is also important to emphasize that the robustness of the system in terms of personalization and

contextual learning positions it as an essential tool in the new era of artificial intelligence, its

potential lies in its ability to adapt and evolve with each interaction, offering more accurate and

relevant responses over time, therefore, this adaptability improves the user experience in

applications ranging from cyberpsychology, educational environments and different business

solutions (Zheng et al., 2023).

7.2. Next steps in research and development

Although the current capabilities of the Cognitive AI framework are extensive, the next steps in

research and development should focus on further improving its learning capacity and

adaptability, a key direction identified in this study is the future work of continuous learning
algorithms that allow Cognitive AI to continue learning and adjusting without the need to be

trained from scratch. which would increase their ability to respond to new and changing

situations. We also identified the need to implement mechanisms to minimize cognitive biases

and improve equity in decision-making, ensuring that the system maintains a high ethical and

fair standard in its responses and actions.

Another relevant issue is the use of lighter and more optimized models that can contribute to the

generation and implementation of more systems based on Cognitive AI with applications to

mobile phones and IoT devices (Rashid et al., 2022), in addition to the integration of advanced

perception and multimodality capabilities (e.g., text blending, voice and image), which would

expand the applicability of the framework to new scenarios and use cases, ensuring that the

growing expectations of AI users are met.

8. References

• Aberg, K. C., Doell, K. C., & Schwartz, S. (2016). The "Creative Right Brain" Revisited:

Individual Creativity and Associative Priming in the Right Hemisphere Relate to

Hemispheric Asymmetries in Reward Brain Function. Cerebral Cortex, 27(10), 4946–

4959. https://doi.org/10.1093/CERCOR/BHW288

• Aruna, K., & Gurunathan, P. (2024). Enhancing Edge Environment Scalability:

Leveraging Kubernetes for Container Orchestration and Optimization. Concurrency and

Computation: Practice and Experience. https://doi.org/10.1002/cpe.8303

• Aguilar Vera, R. A., Jácome, A. N., Díaz Mendoza, J., & Gómez Gómez, O. (2023).

NoSQL Database Modeling and Management: A Systematic Literature Review. Revista

Facultad de Ingeniería, 32(65), e16519.

https://doi.org/10.19053/01211129.v32.n65.2023.16519

• Bakare, S. S., Adeniyi, A. O., Akpuokwe, C. U., & Eneh, N. E. (2024). Data privacy laws

and compliance: a comparative review of the eu gdpr and usa regulations.

https://doi.org/10.51594/csitrj.v5i3.859

• Chen, J., Wang, S., Qi, Z., Zhang, Z., Wang, C., & Zheng, H. (2024). A Combined

Encoder and Transformer Approach for Coherent and High-Quality Text Generation.

https://doi.org/10.48550/arxiv.2411.12157

• Chieu, T. C., & Zeng, L. (2008). Real-time Performance Monitoring for an Enterprise

Information Management System. International Conference on E-Business Engineering,

429–434. https://doi.org/10.1109/ICEBE.2008.93

• Christoph, Durt., & Thomas, Fuchs. (2024). Large Language Models and the Patterns of

Human Language Use. 106-121. http://doi.org/10.4324/9781003312284-7

• Finch, S. E., & Choi, J. D. (2024). ConvoSense: Overcoming Monotonous Commonsense

Inferences for Conversational AI. In Transactions of the Association for Computational

Linguistics (Vol. 12, pp. 467–483). MIT Press. https://doi.org/10.1162/tacl_a_00659

• Fuente, J., & Pousada, M. (2019, February 20). The systems of memory. Psychology of

Attention and Memory. Retrieved January 23, 2025, from

https://campus.uoc.edu/annotation/9aec20091b7a3348226c564e34e7997f/833597/PID_0

0260031/PID_00260031.html

• Hazra, R., Chatterjee, P., Singh, Y., Podder, G., & Das, T. K. (2024). Data Encryption and

Secure Communication Protocols. Advances in Web Technologies and Engineering Book

Series, 546–570. https://doi.org/10.4018/979-8-3693-6557-1.ch022

• Liu, J., Mei, Z., Peng, K., & Vatsavai, R. R. (2023). Context Retrieval via Normalized

Contextual Latent Interaction for Conversational Agent. In 2023 IEEE International

Conference on Data Mining Workshops (ICDMW) (pp. 1543–1550). IEEE.

https://doi.org/10.1109/icdmw60847.2023.00196

• Malyala, S. S., Guntaka, J. R., Chintala, S. V., Vattikuti, L., & Tummalapalli, S. (2023).

Exploring How AI Answering Models Understand and Respond in Context. International

Journal for Research in Applied Science and Engineering Technology (Vol. 11, Issue 9,

pp. 224–228). https://doi.org/10.22214/ijraset.2023.55597

• Marri, S. (2024). Conceptual Frameworks for Conversational Human-AI Interaction

(CHAI) in Professional Contexts. International Journal of Current Science Research and

Review (Vol. 07, Issue 10). Everant Journals. https://doi.org/10.47191/ijcsrr/v7-i10-42

• Meng, S., Dollahite, R., Kaur, A., Schell, P., Sharma, A., Ventre, G., Kranz, I., Nudelman,

G., & Gerling, G. J. (2024). Developing Design Features to Facilitate AI-Assisted User
 Interactions. In 2024 Systems and Information Engineering Design Symposium (SIEDS)

(pp. 437–442). IEEE. https://doi.org/10.1109/sieds61124.2024.10534665

• Pandey, A., & Bhat, A. (2023). A Review on Textual Question Answering with

Information Retrieval and Deep Learning Aspect. In 2023 7th International Conference

on Intelligent Computing and Control Systems (ICICCS) (pp. 224–229). IEEE.

https://doi.org/10.1109/iciccs56967.2023.10142729

• Rashid, H.-A., Ovi, P. R., Busart, C., Gangopadhyay, A., & Mohsenin, T. (2022).

TinyM$^2$Net: A Flexible System Algorithm Co-designed Multimodal Learning

Framework for Tiny Devices (Version 3). arXiv.

https://doi.org/10.48550/ARXIV.2202.04303

• Sereati, C. O., Sumari, A. D. W., Adiono, T., & Ahmad, A. S. (2020). Towards cognitive

artificial intelligence device: an intelligent processor based on human thinking emulation.

TELKOMNIKA (Telecommunication Computing Electronics and Control) (Vol. 18, Issue

3, p. 1475). Universitas Ahmad Dahlan. https://doi.org/10.12928/telkomnika.v18i3.14835

• Sholihah, R. A. (2022). Language and Brain: Neurological Aspects in Language

Acquisition. Muharrik, 5(1), 220–230. https://doi.org/10.37680/muharrik.v5i1.1069

• Torrentira, M. (2024). Capabilities and Application of Artificial Intelligence (AI) Models

in Qualitative and Quantitative Data Mining, Data Processing and Data Analysis.

Archives des Sciences (Vol. 74, Issue 3, pp. 198–200). ScienceStride Publishing.

https://doi.org/10.62227/as/74331

• Vilasini, V., Hariharan, K., Chezhian, N., & Kaviyarasu, D. (2024). Secured Data Storage

Using OAuth and Encryption of Data Using AES with Patterns and Hidden Layer.

https://doi.org/10.1109/icsseecc61126.2024.10649427
• Ye, Z., Ai, Q., Liu, Y., Zhang, M., Lioma, C., & Ruotsalo, T. (2023). Language

Generation from Human Brain Activities. arXiv.Org, ABS/2311.09889.

https://doi.org/10.48550/arxiv.2311.09889

• Zhang, L., Yuan, F., Xu, J., & Shen, Y. (2023). Computer Network Security Strategy

Based on Data Encryption Technology. https://doi.org/10.1109/cicn59264.2023.10402235

• Zheng, W., Milic-Frayling, N., & Zhou, K. (2023). Contextual Knowledge Learning For

Dialogue Generation (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2305.18200