Welcome to the forefront of conversational AI as we explore the fascinating world of AI chatbots in our dedicated blog series. Discover the latest advancements, applications, and strategies that propel the evolution of chatbot technology. From enhancing customer interactions to streamlining business processes, these articles delve into the innovative ways artificial intelligence is shaping the landscape of automated conversational agents. Whether you’re a business owner, developer, or simply intrigued by the future of interactive technology, join us on this journey to unravel the transformative power and endless possibilities of AI chatbots.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Scientific Reports volume 15, Article number: 33074 (2025)
1196
Metrics details
Due to the increasing demand for intelligent customer service in the financial sector, the use of AI-powered chatbots is becoming unavoidable. Existing chatbot systems lack sufficient knowledge of complex language patterns, which limits their ability to handle intricate customer queries effectively. This study introduces an advanced AI-based customer service framework called iDigiChat, which integrates a knowledge graph (KG) and artificial neural network (ANN) techniques for more precise and responsive interactions. The system converts customers’ queries into a structured graph format using Text-to-Graphql. It then applies Knowledge Graph Completion models to extract relational insights, which are classified and refined through a multi-layered ANN framework (KG-ANN). The simulation was conducted using a large-scale dataset consisting of 34,089 chatbot logs and 317,438 customer service call logs from a leading Korean bank, collected over a three-year period (2018–2020). The results show that the system achieves 90% R2 and significantly reduces both mean absolute error and mean squared error by 25%, enabling faster and more reliable customer interactions.
The development of AI has profoundly transformed financial institutions. Thanks to these technological advances, AI chatbots now play a key role in financial customer service, offering efficient and personalised banking solutions. This study examines the evolution of AI in finance and highlights the rising demand for intelligent chatbots. With a growing need for accuracy and transparency, especially during digital transitions post-COVID-19, this research introduces iDigiChat, a chatbot model that combines the strengths of Knowledge Graph (KG) and Artificial Neural Network (ANN).
The growing interest in modern technologies significantly transforms the financial sector. In recent years, the use of artificial intelligence (AI) and its contributions to various applications have been remarkable. According to a recent review from 1989 to 2024, AI is now widely utilised in different financial platforms such as credit scoring, fraud detection, robo-advisory services, and financial inclusion. At the same time, concerns about fairness, transparency, and accouncategorises each channel-product ility—particularly the need for explainable AI (XAI)—are increasingly important in both research and practical applications1. Similarly, a comprehensive analysis of SN Business & Economics found that the rising interest in AI, based on neural networks and expert systems, can significantly transform areas such as trading, risk assessment, fraud detection, customer service, and personalised financial planning2. This enhances predictability and efficiency. From a business and risk perspective, social and economic factors significantly influence the growth of AI in finance, especially in cyber insurance and financial governance. The study highlights that most companies are investing more in AI tools for fraud detection, chatbot-based customer support, cyber-security, and risk management. This shift gained further attention during the COVID-19 pandemic, leading many financial institutions to adopt smarter, more secure digital solutions3. In practical banking, the Financial Times reports that generative AI is increasingly speeding up mortgage approvals. These advanced models can respond to complex customer queries and help manage risk situations4. However, the report emphasises that strong rules and oversight remain essential to ensure these tools are utilised responsibly. Similarly, IEEE Spectrum examines the careful integration of AI into finance5.
Nowadays, chatbots are considered an essential part of modern banking to enhance both service and efficiency. Several researches highlight their advantages and contribution to customer service support. Smart chatbots, designed by advanced language models, can understand complex customer queries and also help to avoid confusion and build trust6,7. Banks use chatbots for simple, everyday tasks like checking balances and paying bills to maintain proper records, as they can handle many requests quickly without manual support6. Study8 focused on fintech in Indonesia and observed that chatbots continuously assist people with services and support through popular chat apps. According to9, users in Nigeria found that, even with limited access, chatbots can support customers and provide proper financial assistance. Finally, in Jordan, chatbots are designed to feel personal; they increase customer satisfaction and play a significant role in the bank’s financial operations10. Considering these studies, we see that chatbots can strengthen customer relationships, make services more accessible, and give banks a clear competitive edge.
In recent days, AI chatbots in finance have shown notable development and provide 24/7 customer support, fraud detection, and personalised advice. Several research studies show BERT-based bots that can be used to accurately understand user content and intelligently accept complex queries from humans11. According to12, it shows that the chatbots improve bank profits by efficiently handling routine services like bill payments. Scientists in Jordan reported voice chatbots that help to develop customer satisfaction and a competitive advantage13. A U.S. government report noted that nearly 37% of Americans use bank chatbots to solve complex finance-related issues14. A South Korean study highlighted that service quality, usability, security, and convenience are the main factors in achieving sustained chatbot usage15. Further details about chatbots and their improvements are illustrated in Table 1.
Based on the strong background and the growing impact of chatbots in financial services, this study introduces iDigiChat, an intelligent customer service chatbot explicitly designed for the banking sector. The model combines a Knowledge Graph (KG) and an Artificial Neural Network (ANN) to understand user queries and respond more accurately. The customer queries are translated into a graph-based format using Text-to-GraphQL. The Knowledge Graph processes this structure to identify relationships and complete missing information. Then, the ANN classifies the request and generates the most relevant response. The main contributions of the study are as follows:
The study presents iDigiChat, an AI-powered chatbot model combining KG and ANN to improve understanding of customer queries and provide faster and accurate responses in financial services.
The model uses Text-to-GraphQL for logical query translation and ANN for intelligent service classification. This ensures better intent recognition, reduced human efforts and improved user experience.
The simulation was performed under 34,089 chatbot logs and 317,438 customer service records from a leading Korean bank.
The model achieved over 25% reduction in error rates and shows expected service efficacy.
The remaining part of the study is organised as follows: Section “Related works” discusses the existing implementations of chatbots and their contributions to the respective domains. Section “Proposed methodology” clearly illustrates the proposed methods and their advantages. Section “Result analysis” presents the visual experiments to demonstrate the proposed efficacy. Finally, Section “Conclusion” concludes the study.
The words “chatting” and “robot,” which are frequently employed in messaging apps or transmitters, are combined to form the phrase “chatbot.” A chatbot is a piece of messaging software that can handle variations, deliver correct replies, generate models that continually enhance the correct responses through chats with clients, and save pertinent answers to inquiries on a server9. Chatbots build a self-learning model through computer algorithms and mathematical computations and deliver consumer responses and other relevant data as closely as feasible to user requests10. A chatbot is an interface business use to connect with financial consumers and deliver the information consumers and advertisers need11. Chatbots also improve service satisfaction; when compared to traditional self-service tools, they provide more interactive and personalised support, which contributes to sustainable service delivery in financial sectors12.
Erica, a virtual assistant from Bank of America, debuted in the financial industry for the first time in May 2017. Erica’s initial appearance reminded me of Siri from Apple. Erica responded with simple texts and voice replies, including details about the transaction, maximum sums, and account balances13. Furthermore, it offered cutting-edge services like applying for a credit score improvement, introducing fund products, applying for a financial institution’s loan, receiving interest rate advice, paying utility bills, and receiving investment management consulting assistance14. The chatbot learned about users’ profiles, past banking purchases, locations, and daily routines using machine learning and deep computing technology to offer precise and personalized services15. Based on this, we observe that the consumers could benefit from the ease of quickly and easily obtaining customized financial services through a chatbot.
In contrast to how they are implemented in online or mobile apps, chatbots can be divided into two categories: retrieving and generating models. Initially, a chatbot built on the search framework employs a rule-based approach that offers predefined responses per the circumstances of a particular subject or query. Most early chatbot variants banks and other financial institutions used were created using rules16. Still, as business information continues to amass, advanced machine learning is now achievable with the commercialization of chatbots17. Secondly, the generative framework for the chatbot is a machine learning technique that enhances the precision of new answers via self-evolution as client and interaction information amass18. Generative chatbot systems depend on cross-language-based pretrained models that allow better language generalization and contextual accuracy, particularly in handling diverse customer queries throughout all the regions19.
Also, learning-based approaches allow chatbots to better model and classify human–machine interactions and improve their ability to understand user content and respond in real time20. The platform recognises the query from the consumer and the sentence’s intent, thanks to the most recent advancements in deep learning technologies21, and provides a suitable response. As a result, it is possible to suggest customised goods to consumers. Experiments are currently being undertaken for commercialisation that capture the current feelings of consumers through unique patterns and fundamental characteristics22. Relation-aware semantic parsing techniques help systems to translate user queries into structured responses, which is particularly useful in financial domains that require precision23.
AI chatbots were “developed to develop a connection with users while possessing specific abilities for customer support” as their virtual companions. AI chatbots have both awareness and emotional intelligence (EQ) capabilities. AI chatbots are designed for more effective, readily available, and current data retrieval to enhance IQ capabilities24. The IQ abilities are driven by image recognition, real-time data retrieval, and machine learning, which allows the chatbot to provide quick services25. These features enable users to interact more effectively with chatbot systems, thereby enhancing trust and usability in digital banking platforms26. Additionally, AI chatbots have been proven to deliver quick customer service. They achieve this by learning from users’ past queries, behavioural patterns, and purchase histories to provide targeted advice and more responsive support environments27.
Chatbots can also support decision-making by analysing product features, customer reviews, and personalised data to provide practical recommendations. These systems contribute more to data-driven marketing and financial decision models28. Further studies have shown that chatbots also encourage customer empowerment. By providing interactive and human-like support, they enable users to engage with services more consistently29. Trust and service quality are crucial to the success of chatbots in banking. Research has shown that the reliability, security, and perceived value of chatbot systems directly impact user adoption in mobile and AI-based banking platforms30.
By thoroughly analysing the above discussions, the existing research highlights the capabilities of chatbots in enhancing financial services through personalised responses, emotional intelligence, and real-time decision-making. However, we observe some critical gaps in integrating these advanced features into a unified framework. Most studies focus on rule-based generative models in isolation, lacking a detailed approach to combining deep learning, semantic parsing, and real-time adaptation. Additionally, only a limited number of works investigate chatbot efficacy using large-scale, real-world banking datasets that encompass diverse customer queries. There is also a shortage of structured methods to balance the accuracy of automation with the emotional and contextual understanding of users. To address these gaps, we proposed our suggested KG-ANN model.
The proposed methodology combines KG technology with an ANN to improve the chatbot’s knowledge, classification, and response generation. The process begins with data collection from a central Korean bank, spanning a three-year period from January 2018 to December 2020. This dataset comprises 34,089 chatbot log records and 317,438 customer service call logs, focusing on four primary financial products and services. Once the data collection is done, the data undergoes preprocessing using a Text-to-GraphQL technique, which converts customer queries into a graph-based format. This enables the system to identify meaningful relationships and represent them as entities and connections within the knowledge graph. The Knowledge Graph Completion (KGC) model is then applied to predict missing links and strengthen the semantic structure of the data. These outputs are transformed into the ANN framework, which is structured to analyse and classify service types based on emotional, informative, and trust-related features. The ANN consists of four sub-models, each handling features such as social attraction, emotional credibility, emotional connection, and buying intention. This enables the model to accurately classify the precise outcomes of user intent and service needs. The complete architecture and data flow of the model are illustrated in Fig. 1 of the study.
Architecture of proposed method.
To investigate the impact of customer goods and services expenditures on bank profits (return rate rise) through two channels utilising ARS, this study analysed product information from a major Korean banking firm. The simulation dataset is adapted from6. Based on client information and the statistically significant nature of each channel’s contribution to bank profitability, we examined banking services and goods made available through chatbots or customer care phone calls. We believe that banks will be able to develop stable financial measures with the aid of our analyses. Additionally, we hope to objectively determine how much the customer care departments of all financial institutions, especially financial institutions, can be replaced by AI-based chatbots. From January 2018 to December 2020, when the chatbot was initially implemented at Bank A, we acquired 34,089 personal records of four essential items and services sold through this channel from this financial institution. Additionally, we gathered 317,438 unstructured voice recordings based on similar items from customer service. We could organise the unorganised information using a text conversion method, a Knowledge Graph, and a conversion methodology. All conditions were identical, except that each of the four goods was handled by either a chatbot or a customer support representative; it is safe to infer that the statistical effect is predictable.
Tables 2,3 and 4, which categorises each channel-product group specifically, shows roughly 9.5 times as many customers who bought products via customer support as those using the chatbot among every customer who bought financial services using the Automatic Response System (ARS). As a result, 92.3% of the information about the parameters made purchases via intermediaries, while just 9.8% used chatbots. Regarding age categories, senior citizens (55.2%) are more likely than younger people (45.8%) to purchase products and services through excellent customer service. All four of the goods supplied through customer service follow the same pattern. Notably, the disparity increases by 14.8%, or almost 5% more than the median of 9.57%. 55.9% of those who purchase goods use support services to pay their utility bills.
The above statistical data mentioned in Tables 2,3 and 4 is used for evaluation. The proposed method classifies the products based on customer service and chatbot users.
The knowledge graph method utilises Text-to-Graph Query Language (QL), which enables the translation of user or customer queries into a graphical structure. It transforms customer language queries into logical representations and makes it suitable for further processing. The primary task of the model is to address a multi-class classification problem, where customer service queries are categorised into predefined service types such as ‘loan inquiry’, ‘bill payment’, ‘investment support’, and ‘credit score assistance. The knowledge graph is often developed on top of the current modules in the proposed chatbot to link all retrieved data, combine structured and unstructured information, and display it to the user as a legitimate knowledge panel in a structured and understandable manner.
A knowledge graph (K=left({e}_{i},{r}_{j},{e}_{k}right)) is a collection of triplets where ({e}_{i}) and ({e}_{k}) denote the entities, rj represents the relation between them; accordingly, provided a set of entities (E) and a set of relationships (R). We typically refer to ({e}_{i}) and ({e}_{k}) as the head of the item ({h}_{i}) and tail ({t}_{k}) of the triplet. A third-order binary tensor, known as the adjacency tensor of (K), termed (Xin 0, 1 | E) || R || E |, can identify a knowledge graph’s identity. If (({h}_{i},{r}_{j},{t}_{k})) is true, then ({X }_{ijk}= 1); otherwise, ({X }_{ijk}= 0).
Using known triplets in (K), KG attempts to foresee legitimate but unseen triplets. To correlate a score (s({h}_{i},r,{t}_{k})) with each possible triplet (({h}_{i},r,{t}_{k})), KG systems construct a score function s: (Etimes Rtimes Eto R). The scores reflect the likelihood of triplets. KGC models first fill in the blanks for every entity in the knowledge graphs for queries like (({h}_{i},{r}_{j},?)) or ((?,{r}_{j},{t}_{k})), and then they grade the resulting triplets. Valid triplets should perform better than invalid triplets, according to expectations.
Modern KG models include knowledge graph embedding (KGE) designs that connect every item (whether it’s a head or tail organization, ({e}_{i}in E)) and relation (({r}_{j}in R)) to an embedded ({e}_{i}) and ({r}_{j}) on carefully chosen embedding spaces. Then, they immediately create an evaluation mechanism to simulate how relationships and entities behave. Following is a review of three typical KGE models.
TransE
TransE uses Minkowski lengths f+or establishing function scores in the actual world. The scoring function is particularly given its embeddings (h, r, and t.)
where 1/2 denotes that the distance might be either L1 or L2.
DistMult
DistMult provides functions for scoring on the actual space using the inner product. Notably, the function that scores is given the embeddings (h, r), and (t).
here (r) is a diagonal matrix formed from the relation vector (r) and (h) and (t) are entity embeddings.
ConvE
ConvE uses convolutional neural networks to specify scoring algorithms. The evaluation mechanism is a
An activating operation, a 2D convolution, a filter used in convolutional layers, a 2D shape for real vectors, (sigma) is the activation function, ω is the convolutional kernel, and W is a trainable matrix of weights, respectively.
Figure 2 shows the structure of Knowledge Graph Completion.
Structure of KGC.
We utilized the ANN to identify potential non-linear relationships and rank the significance of each component after examining the causal connection through KG. According to previous studies, the input neurons in ANN models can only be significant independent variables. The four endogenous components of the KG-ANN model (such as social interest, psychological trustworthiness, emotional connection, and buying intention) led us to partition the framework into four neural network models. Three inputs of emotional, informative, and reputational support and one output of social attraction are considered as Model A. Three inputs (emotional, informational, and esteem support) and one outcome (emotional credibility) are present in Model B. Social attractiveness and emotional credibility is two inputs in Model C, and emotional connection is one outcome. Lastly, model D has a single output (buying intention) and two inputs (social attraction and affective believability). The four ANN models used in this investigation are depicted in their architecture in Fig. 3.
Structure of proposed KG-ANN method.
It is shown in Eqs. (4) and (5) ({W}_{ji}) is the weight between (ith) input neuron and (jth) hidden neuron and ({V}_{kj}) denotes weight between the (jth) hidden neuron and (kth) output neuron, respectively, ({x}_{i}) is the input to the network and ({y}_{j}) is the output of hidden neuron (j) and finally ({O}_{k}) is the output of the neuron (k). Equation (6) shows that a typical sigmoid function is increasing monotonically and capable of differentiation, with values ranging from 0 to 1;
The SSE (sum square of error) formula is given in Eq. (7), where ({d}_{pk}) denotes the desired outcome of neuron (k) and ({o}_{pk}) is the actual output of the neuron (k) with respect to input pattern (p); (P) is the total number of input patterns and (K) is the total number of output neurons.
The simulation of the model aims to classify the customer service queries across chatbot and traditional support channels. The dataset, consists of 34,089 chatbot logs and 317,438 customer support call logs, was split into training (80%) and testing (20%) sets to ensure balanced evaluation. Each query was categorised into specific financial service types, such as fund inquiries, bill payments, loan-related queries, and new product subscriptions. The model was evaluated using the evaluation metrics of MAE, MSE, and R2. Statistical tests, such as ANCOVA and t-tests, were also conducted to validate the model’s ability to distinguish channel effectiveness based on age groups and service types. Table 5 shows the hyperparameters of the model.
Table 6 shows the ANCOVA test results. The ANCOVA test shows how the customer interactions through a chatbot or a human agent, impact the customer financial product purchase. It also considers differences like age and service type. The model had a degrees of freedom (DF) of 2 and explained a variation of 146.4783 between groups. The average variation (mean square) within the model was 3.6878, where the error variation was 3.9735, based on over 357,000 records. The F-value of 4.0033 and the low p value (l < 0.0001) confirm the model effectiveness in analysing the difference in their behavior. In Fig. 4 shows the result of covariance analysis using ANCOVA test. It uses Table 5 for evaluation.
Experimental result of ANCOVA test.
Based on consumer services and chatbot channels, the data demonstrate that the simultaneous sales of fresh goods and current offerings have a substantial impact on the rise or fall in bank earnings. Therefore, rather than a chatbot, new product-oriented finances and residential subscriptions are better suited for customer support. Contrarily, offerings related to already-existing goods, such as the payment of loan interest or utility bills, are better suited to being handled using chatbots, which benefits bank net revenue.
We concluded that, depending on the two customer channels—customer operations and chatbots—both Junior and Senior consumers have a substantial impact on the growth or decline of bank revenues. The percentage of Seniors was higher when purchasing products through customer service, whereas the percentage of Juniors was higher when using the chatbot. In summary, the age group that makes up a disproportionately significant part has a favourable impact on bank profitability. In Fig. 5 shows the purchasing services based on age group with chatbots. By using chatbot the junior services are increased. Similarly, in Fig. 6 shows the purchase growth based on age groups without chatbots.
Purchasing growth based on age group with chatbots.
Purchasing group based on age group without chatbot.
Figure 7 illustrates the KG-ANN model’s performance improvement as the number of training samples increases. From the figure we observe that, as training samples increase from 0 to 175, MAE consistently decreases, which indicates low prediction errors, and R2 increases, showing a stronger correlation between predicted and actual outcomes. At 100 samples, R2 reaches 60%, and MAE drops below 40%, which shows significant model improvement. With 175 samples, R2 reaches nearly 90%, and MAE drops nearly to 25%, which highlights the model’s effectiveness on the test data.
Effect of training data size on model performance.
The F test (F = 8.12) demonstrates that the presumption that there is equal variance is met in the scenario of Hypothesis 3a (Table 7). As a result, we use the pooling t-test, and the test result (t = 1.4352) supports the theory. Therefore, there is no disparity in the bank’s net revenue (New items-Junior group) when comparing customers who purchase items via the customer service department with consumers who buy goods through chatbots.
The Table 8 shows the comparison analysis of the chatbots. According to31, they built a chatbot using BERT and Bayesian methods for investment banking and reached about 92% accuracy in understanding user intent. Then,6 created a rule-based chatbot for basic banking tasks like bill payments and achieved 75% accuracy. And32 achieved about 80% customer satisfaction efficacy, but it was more focused on user experience than advanced task performance. When compared with all these chatbots, our KG-ANN-based bot performs better in understanding and classifying the customer queries and achieved an efficacy of over 90% R2 score.
This study introduced iDigiChat, a smart customer service chatbot specifically designed for the financial sector. By integrating KG and ANN, the model effectively addresses the limitations of traditional rule-based chatbots. The system converts user queries into structured formats using Text-to-GraphQL, enhances semantic understanding, and classifies them using an ANN for personalised responses. The model was trained and tested on a large dataset from a Korean bank, which included over 34,000 chatbot logs and 317,000 customer service call logs. The simulation results demonstrated high effectiveness, achieving an R2 of up to 90% and reducing MAE and MSE by 25%, which confirms the model’s strong classification and predictive capabilities. Apart from the advantages, the model has some limitations. The model depends on predefined service categories and focuses only on text-based interactions. Future work should consider integrating voice-based inputs, expanding multilingual capabilities, and adopting reinforcement learning to make the system more adaptable across diverse financial platforms, thereby enhancing the user experience.
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.
Vuković, D. B., Dekpo-Adza, S. & Matović, S. AI integration in financial services: A systematic review of trends and regulatory challenges. Humanit. Soc. Sci. Commun. 12(1), 1–29 (2025).
Article Google Scholar
Yang, K. How to prevent deception: A study of digital deception in visual poverty livestream. New Media Soc. (2024).
Aleksandrova, A., Ninova, V. & Zhelev, Z. A survey on Ai implementation in finance,(cyber) insurance and financial controlling. Risks 11(5), 91 (2023).
Article Google Scholar
Wu, L., Long, Y., Gao, C., Wang, Z. & Zhang, Y. MFIR: Multimodal fusion and inconsistency reasoning for explainable fake news detection. Inf. Fus. 100, 101944 (2023).
Article Google Scholar
Li, D., Ortegas, K. D. & White, M. Exploring the computational effects of advanced deep neural networks on logical and activity learning for enhanced thinking Skills. Systems 11(7), 319 (2023).
Article Google Scholar
Hao, R. & Yang, X. Multiple-output quantile regression neural network. Stat. Comput. 34(2), 89 (2024).
Article MathSciNet Google Scholar
Li, L., Xia, Y., Ren, S. & Yang, X. Homogeneity Pursuit in the Functional-Coefficient Quantile Regression Model for Panel Data with Censored Data (2024).
Zhao, H. et al. Cross-lingual font style transfer with full-domain convolutional attention. Pattern Recognit. 155, 110709 (2024).
Article Google Scholar
Shi, J., Liu, C. & Liu, J. Hypergraph-based model for modeling multi-agent Q-learning dynamics in public goods games. IEEE Trans. Netw. Sci. Eng. 11(6), 6169–6179 (2024).
Article MathSciNet Google Scholar
Yang, H. et al. The global industrial robot trade network: Evolution and China’s rising international competitiveness. Systems 13(5), 361 (2025).
Article Google Scholar
Li, D., Zhang, M., Zhou, S. & Yu, X. Structural equation modeling and validation of virtual learning community constructs based on the Chinese evidence. Interact. Learn. Environ. 1–16 (2025).
Yin, L. et al. DPAL-BERT: A faster and lighter question answering model. CMES Comput. Model. Eng. Sci. 141(1), 771–786 (2024).
Google Scholar
Chen, S. et al. Enhancing Chinese comprehension and reasoning for large language models: an efficient LoRA fine-tuning and tree of thoughts framework. J. Supercomput. 81(1), 50 (2024).
Article Google Scholar
Arugula, B. Prompt engineering for LLMs: Real-world applications in banking and ecommerce. Int. J. Artif. Intell. Data Sci. Mach. Learn. 6(1), 115–123 (2025).
Google Scholar
Chen, X. et al. Joint scene flow estimation and moving object segmentation on rotational LiDAR data. IEEE Trans. Intell. Transp. Syst. 25(11), 17733–17743 (2024).
Article Google Scholar
Zhu, C. Research on emotion recognition-based smart assistant system: Emotional intelligence and personalized services. J. Syst. Manag. Sci. 13(5), 227–242 (2023).
Google Scholar
Pan, Y. & Xu, J. Human-machine plan conflict and conflict resolution in a visual search task. Int. J. Human Comput. Stud. 193, 103377 (2025).
Article Google Scholar
Zhang, J. et al. GrabPhisher: Phishing scams detection in ethereum via temporally evolving GNNs. IEEE Trans. Serv. Comput. 17(6), 3727–3741 (2024).
Article ADS Google Scholar
Li, T., Li, Y., Zhang, M., Tarkoma, S. & Hui, P. You are how you use apps: User profiling based on spatiotemporal app usage behavior. ACM Trans. Intell. Syst. Technol. 14(4) (2023).
Cui, F., Cui, Q. & Song, Y. A survey on learning-based approaches for modeling and classification of human-machine dialog systems. IEEE Trans. Neural Netw. Learn. Syst. (2020).
Li, T., Li, Y., Xia, T. & Hui, P. Finding spatiotemporal patterns of mobile application usage. IEEE Trans. Netw. Sci. Eng. (2021).
Li, T. et al. The impact of Covid-19 on smartphone usage. IEEE Internet of Things J. 8(23), 16723–16733 (2021).
Article Google Scholar
Hwang, W., Yim, J., Park, S. & Seo, M. A comprehensive exploration on wikisql with table-aware word contextualization (2019). ArXiv Preprint ArXiv:1902.01069.
Gong, J., Yu, Q., Li, T., Liu, H., Zhang, J., Fan, H. & Li, Y. Demo: Scalable digital twin system for mobile networks with generative AI. Paper presented at the MobiSys ’23, New York, NY, USA (2023).
Huang, M. H. & Rust, R. T. A strategic framework for artificial intelligence in marketing. J. Acad. Mark. Sci. 49(1), 30–50 (2021).
Article Google Scholar
Sands, S., Ferraro, C., Campbell, C. & Tsao, H. Y. Managing the human-chatbot divide: How service scripts influence service experience. J. Serv. Manag. 32(2), 246–264 (2021).
Article Google Scholar
Shahzad, F., Xiu, G., Khan, M. A. S. & Shahbaz, M. Predicting the adoption of a mobile government security response system from the user’s perspective: An application of the artificial neural network approach. Technol. Soc. 62, 101278 (2020).
Article Google Scholar
Sharifi, M., Pool, J. K., Jalilvand, M. R., Tabaeeian, R. A. & Jooybari, M. G. Forecasting of advertising effectiveness for renewable energy technologies: A neural network analysis. Technol. Forecast. Soc. Chang. 143, 154–161 (2019).
Article Google Scholar
Sharma, S. & Khadka, A. Role of empowerment and sense of community on online social health support group. Inf. Technol. People 32(6), 1564–1590 (2019).
Article Google Scholar
Sharma, S. K. & Sharma, M. Examining the role of trust and quality dimensions in the actual usage of mobile banking services: An empirical investigation. Int. J. Inf. Manag. 44, 65–75 (2019).
Google Scholar
Yu, S., Chen, Y. & Zaidi, H. AVA: A financial service chatbot based on deep bidirectional transformers. Front. Appl. Math. Stat. 7, 604842 (2021).
Article Google Scholar
Ali, M. S., Swiety, I. A. & Mansour, M. H. Evaluating the role of artificial intelligence in the automation of the banking services industry: Evidence from Jordan. Human. Soc. Sci. Lett. 10(3), 383–393 (2022).
Google Scholar
Download references
School of Business Administration, Guizhou University of Finance and Economics, Guiyang, 550025, China
Ji He
Business School, Guizhou University, Guiyang, 550025, China
Yi Luo
International Academy, Guizhou City Vocational College, Guiyang, 550025, China
Ting Wang
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
Ji He: Supervision, Conceptualization, Methodology, Formal analysis, Writing—original draft, Investigation, Data Curation, Validation, Resources, Software, Visualization; Yi Luo and Ting Wang: Conceptualization, Methodology, Formal analysis, Writing—original draft, Writing—review & editing, Investigation, Data Curation, Validation, Resources, Software, Visualization.
Correspondence to Ji He.
The authors declare no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
He, J., Luo, Y. & Wang, T. iDigiChat: intelligent digital marketing service chatbot for providing efficient customer services using artificial intelligence. Sci Rep 15, 33074 (2025). https://doi.org/10.1038/s41598-025-14722-5
Download citation
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s41598-025-14722-5
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Advertisement
Scientific Reports (Sci Rep)
ISSN 2045-2322 (online)
© 2025 Springer Nature Limited
Sign up for the Nature Briefing: AI and Robotics newsletter — what matters in AI and robotics research, free to your inbox weekly.