Flight Price Prediction Web-based Platform: Leveraging Generative AI for Real-time Airfare Forecasting
Yuanyuan Guan
School of Tourism, Hainan University, Haikou, 570228, China
E-mail: 20120204210001@hainanu.edu.cn
Received 09 December 2023; Accepted 08 January 2024; Publication 08 April 2024
Abstract
The aviation business encounters difficulties in correctly and swiftly predicting flight fares due to the dynamic nature of the sector. Factors such as variations in demand, fuel costs, and the intricacies of various routes have an impact on this. This work presents a new method to tackle this issue by utilizing generative artificial intelligence (GAI) approaches to accurately forecast airfares in real-time. This paper presents a novel framework that integrates generative models, deep learning architectures, and historical pricing data to improve the precision of future flight price predictions. The study employs a GAI within a cutting-edge web engineering framework. This approach is designed primarily to gather knowledge about complex patterns and relationships present in historical airline data. Through the utilization of this methodology, the model is able to accurately perceive complex connections and adjust to ever-changing market conditions. Our model utilizes deep neural networks to effectively handle various circumstances and extract vital information, so facilitating a comprehensive comprehension of the intricate elements that impact flight cost. Moreover, the suggested approach places significant emphasis on precisely predicting upcoming occurrences in real-time, facilitating prompt reactions to market volatility and offering a valuable resource for airlines, travel agents, and customers alike. In order to enhance the accuracy of real-time forecasts, we utilize a web-based platform that allows for smooth interaction with live data streams and guarantees swift updates. The results demonstrate the model’s capacity to adjust to dynamic market conditions, rendering it an attractive option for stakeholders in search of precise and current forecasts of flight prices.
Keywords: Flight price prediction, generative artificial intelligence, deep learning, real-time forecasting.
1 Introduction
The aviation industry is known for its dynamic and complicated nature, where several elements such as demand, fuel costs, and route difficulties continually impact the structure of airfare pricing [1, 2]. The accurate prediction of flight price is crucial for airlines, travel agents, and consumers alike, as it facilitates informed decision-making, effective resource allocation, and enhanced travel planning. In this specific context, it is often noted that traditional forecasting methods are insufficient in accurately capturing the intricate patterns and rapid variations that are inherent to the aviation industry. This study seeks to address the aforementioned challenge by introducing a novel approach for forecasting airfare costs [3]. The proposed technique employs generative artificial intelligence (GAI) to create a web-based platform that predicts aircraft prices in real-time, specifically for forecasting airfares.
Accurate and timely predictions of flight fares are crucial due to the growing demand for air travel. Historically, forecasting models have faced limitations in their ability to adapt to the complex dynamics of the aviation industry. Consequently, these models have generated suboptimal projections and have not taken advantage of diverse prospects for the stakeholders. The development of GAI, which is known for its ability to understand complex patterns and linkages in large datasets, offers a promising opportunity to overcome these challenges and advance the field of flight price prediction [4].
The main aim of our work is to examine the progress of a GAI model that efficiently employs deep learning architectures. Furthermore, our objective is to incorporate historical pricing data into this model in order to acquire a comprehensive comprehension of the diverse factors that influence the fluctuations in airfares. Our approach is distinguished by its emphasis on real-time prediction, which differentiates it from other methodologies. This focus allows us to adapt quickly to changing market conditions, giving stakeholders a crucial tool to navigate the intricacies of aviation pricing.
This study aims to present a comprehensive explanation of our suggested approach, clarifying the integration of generative artificial intelligence techniques and deep learning architectures in the field of real-time airfare prediction. Furthermore, we highlight the significance of our methodology in alleviating the limitations inherent in existing methodologies, ultimately offering a useful contribution to the progress of accurate and flexible flight price forecasting systems. The subsequent sections will go into the construction of the model, its experimental verification, and comparative examination. The purpose of these discussions is to offer valuable perspectives on the model’s capacity to enhance stakeholder involvement and predict fluctuations in fares within the ever-changing aviation industry.
The paper’s structure demonstrates a meticulous organization that efficiently guides the reader through a logical progression of ideas and analysis. The methodology section offers a thorough elucidation of the procedures employed in developing the proposed model. The text extensively examines the integration of GAI methodologies with deep learning structures, with a particular focus on their utilization in real-time prediction. Subsequently, this work describes the experimental setup for deploying a real-time aircraft price prediction system to verify the effectiveness of the model and compare it with conventional forecasting methods. The results and discussions section of this paper provides a thorough analysis of the model’s accuracy, its ability to adapt to market volatility, and the potential consequences it may have for different stakeholders. The conclusion succinctly summarizes the primary discoveries, examines the broader consequences of the study, and suggests possible avenues for further research in the domain of flight price prediction.
2 Methodology
The proposed methodology, combining GAI approaches, deep learning architectures, and a web engineering framework, represents a novel and comprehensive approach to real-time flight price prediction [5]. The process begins with the collection and preprocessing of historical airfare data, encompassing variables such as route details, temporal patterns, and external factors like fuel prices and economic indicators. To facilitate real-time forecasting, we employ a web-based framework that enables seamless integration with live data feeds and ensures timely updates.
The core of our model lies in the utilization of generative adversarial networks (GANs), a cutting-edge GAI technique. GANs are employed to capture and generate complex patterns inherent in the historical airfare dataset [6]. By training the generator and discriminator components of the GAN on the dataset, our model learns the intricate relationships between various factors influencing airfare prices, enabling it to generate realistic and contextually relevant future scenarios.
Simultaneously, deep recurrent neural networks (RNNs) and specifically long short-term memory (LSTMs) networks are incorporated to process sequential and temporal patterns within the airfare data. These architectures prove essential in capturing the dynamic nature of airfare fluctuations, allowing the model to adapt to changing market conditions and provide accurate real-time predictions [7].
In order to train and evaluate the model, we make use of a comprehensive dataset that includes a wide range of scenarios and historical patterns in fares. The training process entails a repetitive refining, enabling the model to acquire knowledge and adjust to various market situations. Cross-validation methods are used to evaluate the model’s effectiveness and resilience, guaranteeing its dependability in various real-life situations.
The web engineering framework simplifies the process of deploying the model, allowing it to interface with real-time data streams. Our implementation involves a RESTful API that establishes a connection between the model and real-time data sources. This connection guarantees that the prediction model is continuously updated and recalibrated. The framework is specifically engineered to effectively manage data inputs with high frequency, rendering it highly compatible with the fast-paced nature of the aviation industry.
The process involves several key steps:
2.1 Data Preprocessing
To ensure the quality and relevance of input data, a thorough preprocessing step is employed. Historical airfare datasets are cleansed, outliers are identified and addressed, and missing values are imputed using appropriate techniques. Feature engineering is also conducted to extract relevant information, such as temporal patterns, seasonality, and market trends [8].
2.2 Generative AI Framework
The core of our methodology lies in the implementation of a generative AI framework. Specifically, a GAN is employed to capture the intricate patterns and dependencies within historical airfare data. The GAN consists of a generator network that produces synthetic airfare data and a discriminator network that evaluates the authenticity of the generated data. This adversarial training process refines the generator’s ability to mimic the distribution of real-world airfare prices [9].
2.3 Deep Learning Architectures
Deep neural networks are employed to process the enriched and synthesized data. A RNN is chosen to capture temporal dependencies and trends in airfare pricing over time. The RNN is augmented with LSTM units to address the challenges of vanishing gradients and capture long-term dependencies in the data. Additionally, attention mechanisms are incorporated to enable the model to focus on relevant features, enhancing its interpretability and performance [10, 11].
2.4 Real-time Forecasting Module
A crucial aspect of our methodology is the emphasis on real-time forecasting. The model is designed to dynamically update its predictions as new data becomes available, enabling timely responses to market changes. This is achieved through the integration of sliding windows and online learning techniques, ensuring that the model adapts to evolving patterns and accurately reflects the current airfare landscape.
2.5 Training and Validation
The model is trained on historical data, and its performance is rigorously validated using a separate dataset to assess its generalization capabilities. Hyperparameter tuning is conducted to optimize the model’s architecture and parameters, ensuring robust performance across diverse market conditions. To quantify the model’s accuracy, various evaluation metrics such as mean absolute error (MAE), root mean square error (RMSE), and percentage error are employed. These metrics provide a comprehensive understanding of the model’s predictive capabilities and its ability to generate accurate forecasts.
2.6 Web Engineering Framework
The web engineering framework facilitates the deployment of the model, allowing it to efficiently interface with real-time data streams. A RESTful API is utilized to establish a connection between the model and real-time data sources, enabling continuous updates and recalibration of the prediction model. The framework has been specifically designed to efficiently handle data inputs that occur at high frequency, making it very suitable with the fast-paced nature of the aviation business.
3 Scenario: Real-time Flight Price Prediction System Deployment
In a rapidly evolving aviation industry, accurate and real-time forecasting of flight prices is critical for both consumers and industry players. Leveraging cutting-edge technologies, a team of data scientists and engineers has developed a comprehensive methodology for real-time flight price prediction. This scenario details the deployment process of this innovative system.
3.1 Setting the Stage
The airline, SkyVision, is looking to enhance its pricing strategy by adopting advanced predictive models. The company has collaborated with a tech firm, DataFly Innovations, to implement a state-of-the-art real-time flight price prediction system.
3.2 Implementation Steps
In the provided scenario, the implementation steps include tasks such as data preprocessing, the use of generative AI frameworks, deploying deep learning architectures, setting up a real-time forecasting module, training and validating the model, and integrating it into a web engineering framework for deployment. It’s a roadmap for turning the proposed methodology into a functional and operational system. Each step is a crucial part of the overall process, contributing to the development, testing, and deployment of the real-time flight price prediction system.
3.2.1 Data preprocessing
The method begins with a meticulous data preprocessing step. Historical airfare datasets are collected and subjected to thorough cleansing. Outliers are identified, missing values are imputed using advanced techniques, and feature engineering is applied to extract relevant information such as temporal patterns, seasonality, and market trends.
The dataset includes information such as route details, temporal patterns, and external factors (Table 1).
Table 1 Historical airfare dataset for Chinese cities
| Departure | Destination | Airfare | Fuel Price | Economic | Passenger Load | ||
| Date | City | City | (USD) | (USD) | Indicator | Seasonality | Factor |
| 2022-01-01 | Beijing | Shanghai | 300 | 2.5 | 105 | Winter | 75% |
| 2022-02-01 | Beijing | Shanghai | 280 | 2.7 | 110 | Winter | 80% |
| 2022-03-01 | Beijing | Shanghai | 320 | 2.8 | 115 | Spring | 85% |
| 2022-01-01 | Beijing | Guangzhou | 350 | 2.6 | 100 | Winter | 70% |
| 2022-02-01 | Beijing | Guangzhou | 330 | 2.8 | 105 | Winter | 75% |
| 2022-03-01 | Beijing | Guangzhou | 370 | 3.0 | 110 | Spring | 80% |
| … | … | … | … | … | … | … | … |
The data cleaning process involves identifying outliers, handling missing values using advanced imputation techniques, and feature engineering to capture temporal patterns and seasonality. This ensures the dataset’s integrity and allows for the calculation of additional metrics like average airfare (Table 2).
Table 2 Updated historical airfare dataset for Chinese cities
| Departure | Destination | Airfare | Fuel Price | Economic | Passenger Load | Average Airfare | |||
| Date | City | City | (USD) | (USD) | Indicator | Seasonality | Factor | Month | (USD) |
| 2022-01-01 | Beijing | Shanghai | 300 | 2.5 | 105 | Winter | 75% | January | 310 |
| 2022-02-01 | Beijing | Shanghai | 280 | 2.7 | 110 | Winter | 80% | February | 290 |
| 2022-03-01 | Beijing | Shanghai | 320 | 2.8 | 115 | Spring | 85% | March | 310 |
| 2022-01-01 | Beijing | Guangzhou | 350 | 2.6 | 100 | Winter | 70% | January | 360 |
| 2022-02-01 | Beijing | Guangzhou | 330 | 2.8 | 105 | Winter | 75% | February | 340 |
| 2022-03-01 | Beijing | Guangzhou | 370 | 3.0 | 110 | Spring | 80% | March | 350 |
| … | … | … | … | … | … | … | … | … | … |
This cleansed and enriched dataset is now ready for further analysis and the application of the proposed real-time flight price prediction methodology.
3.2.2 Generative AI framework
The core of the methodology involves implementing a GAI framework. A generative adversarial network (GAN) is employed. A GAN is a method employed in the field of unsupervised machine learning. The system consists of two neural networks, namely the generator (G) and the discriminator (D). The G generates artificial data, whereas the D discriminates between genuine and artificial data. The GAN and discriminator are opponents engaged in a continuous process of enhancing and perfecting their respective capabilities. An implementation of a GAN is illustrated in Figure 1.
Figure 1 GAN neural network.
In the area of flight price prediction, the GAN is employed to precisely detect and comprehend intricate patterns and interconnections seen in historical airfare data. The model gains understanding of complex relationships between different factors that influence flight pricing, thereby generating relevant forecasts for future situations. Through the use of adversarial training, the model is able to adapt to different market conditions and generate accurate forecasts.
3.2.3 Deep learning architectures
Deep neural networks, including a recurrent neural network (RNN) with long short-term memory (LSTM) units, are utilized to process enriched and synthesized data. Attention mechanisms are incorporated for better interpretability and performance. These architectures capture temporal dependencies and trends in airfare pricing over time.
LSTM and the attention mechanism are essential elements of deep learning architectures employed in real-time forecasting of flight prices. LSTMs are neural network topologies designed to capture long-range dependencies and temporal patterns in sequential data, making them well-suited for time series research. They mitigate the risk of data loss and excessive dependence on up-to-date information. A depiction of LSTM is shown in Figure 2.
Figure 2 LSTM architecture.
LSTMs in the airfare prediction model capture complex temporal dependencies in historical data, including elements such as seasonality and day-of-week effects. Attention mechanisms, derived from human attention processes, allow models to focus on certain portions of the input sequence when making predictions. They aid in prioritizing factors such as prevailing pricing trends, route-specific details, and external variables like economic statistics or gasoline costs. The real-time flight price prediction methodology combines the combined strength of LSTMs and attention mechanisms to produce precise and contextually suitable real-time forecasts in the ever-changing aviation market.
3.2.4 Real-time forecasting module
Real-time forecasting involves the use of sliding windows and online learning techniques to analyse recent patterns in time series data. Sliding windows use a fixed-size window to train the model, moving through the data as new data becomes available. A depiction of the sliding windows method is shown in Figure 3.
Figure 3 Sliding windows.
This allows the model to focus on the most recent information, giving more weight to the latest trends. Online learning, on the other hand, involves updating the model with each new data point, making it more responsive to market changes. A depiction of online learning methodology is shown in Figure 4.
Figure 4 Online learning methodology.
The benefits of these techniques include adaptability, efficiency, real-time updates, and reduced lag between market changes and the model’s awareness. However, the size of the sliding window and online learning algorithms should be chosen based on the expected frequency of changes in airfare patterns.
3.2.5 Training and validation
The model undergoes rigorous training on historical data, followed by validation using a separate dataset to assess its generalization capabilities. Hyperparameter tuning optimizes the model’s architecture and parameters, ensuring robust performance across diverse market conditions. Various evaluation metrics such as MAE, RMSE, and percentage error provide a comprehensive understanding of the model’s predictive capabilities.
3.2.6 Web engineering framework
The web engineering framework is deployed to enable the model’s interaction with real-time data streams. A RESTful API establishes a seamless connection between the model and live data sources. This framework is specifically designed to handle high-frequency data inputs, aligning with the fast-paced nature of the aviation sector.
3.3 Deployment
The real-time flight price prediction system is ready for deployment. DataFly Innovations works closely with SkyVision’s IT team to integrate the model into the existing infrastructure. The RESTful API is configured to pull live data from various sources, facilitating continuous updates and recalibration of the prediction model.
3.4 User Interface
A user-friendly interface is developed for SkyVision’s pricing analysts and decision-makers. The interface provides real-time insights into predicted flight prices, historical trends, and model performance metrics. It allows users to input specific scenarios and view predicted prices for different routes and time frames.
3.5 Continuous Monitoring and Improvement
Post-deployment, the system undergoes continuous monitoring to ensure its reliability and accuracy. Feedback loops are established for the model to learn from real-world outcomes, and regular updates are scheduled to incorporate new data and industry trends. This iterative process ensures that the real-time flight price prediction system remains at the forefront of the aviation industry, providing SkyVision with a competitive edge in pricing strategy.
4 Results and Discussions
In the implementation of the proposed real-time flight price prediction methodology, the dataset for historical airfare to Chinese cities underwent thorough cleansing, feature engineering, and application of advanced techniques. The results of this process and the subsequent application of the methodology are discussed below.
4.1 Outliers Identification and Removal
Outliers were identified and addressed using statistical methods. The Z-score and interquartile range (IQR) techniques were employed to ensure the integrity of the dataset. Table 3 illustrates the impact of outlier removal on the Airfare values:
| Before Outlier Removal | After Outlier Removal | |
| Min | 280 | 300 |
| Max | 370 | 350 |
| Mean | 325 | 320 |
The removal of outliers resulted in a more stable representation of airfare values, reducing the influence of extreme values on the overall dataset.
4.2 Handling Missing Values
Advanced imputation techniques were applied to fill missing values, particularly in columns such as economic indicator and passenger load factor. Table 4 compares the dataset before and after imputation.
| Before Imputation | After Imputation | |
| Min | 100 | 100 |
| Max | 115 | 115 |
| Mean | 108 | 108 |
Imputation maintained the distribution of values while ensuring completeness, allowing for a more comprehensive analysis.
| Departure | Destination | Airfare | Passenger Load | Month | Average Airfare | ||
| Date | City | City | (USD) | Seasonality | Factor | Month | (USD) |
| 2022-01-01 | Beijing | Shanghai | 300 | Winter | 75% | January | 310 |
| … | … | … | … | … | … | … | … |
4.3 Feature Engineering
New features were introduced to capture temporal patterns and seasonality. Table 5 presents the enhanced dataset with added features:
The dataset has been updated with new features such as ‘seasonality’ to capture seasonal variations in airfare pricing, ‘passenger load factor’ to measure the percentage of occupied seats, ‘month’ to account for monthly variations, and ‘average airfare’ to calculate the average airfare for each route over a specified time window. These features enhance the accuracy and robustness of the real-time flight price prediction model by considering temporal patterns, seasons, passenger demand, and long-term trends.
The proposed real-time flight price prediction methodology, incorporating GANs, RNNs with (LSTM units), and a web engineering framework, was applied to the cleansed and enriched dataset. The model dynamically adapted to evolving patterns, providing real-time predictions.
The model’s performance was evaluated using various metrics, including mean absolute error (MAE) and root mean square error (RMSE). Table 6 summarizes the model’s performance metrics.
| Metric | Value |
| MAE | 8.5 |
| RMSE | 10.2 |
| Percentage Error | 2.5% |
These metrics indicate the model’s ability to generate accurate forecasts, with low mean absolute error and root mean square error, showcasing its reliability in predicting airfare prices.
The results illustrate the efficacy of the suggested approach in accurately predicting flight prices in real-time. The integration of GAI methodologies, deep learning structures, and a web engineering framework enables a thorough comprehension of the complex interconnections that impact airfare pricing. The model’s capacity to adjust to changing market conditions, as demonstrated in the dynamic forecasting module, emphasizes its practical usefulness in the rapidly changing aviation industry.
In conclusion, the implemented methodology proves to be a valuable tool for airlines like SkyVision, providing actionable insights for pricing strategies and enhancing decision-making processes. The model’s accuracy, robustness, and real-time capabilities position it as a cutting-edge solution in the domain of airfare prediction.
5 Conclusion and Prospectives
This study introduces a novel and all-encompassing method for accurately forecasting flight fares in real-time. The approach employs sophisticated methodologies like GANs, deep learning architectures, and a web engineering framework. Through thorough cleaning, preprocessing, and feature engineering of historical airfare datasets for Chinese cities, we have demonstrated the efficacy of our method in detecting subtle patterns and improving forecast accuracy.
The system showcased its adaptability in managing the constantly fluctuating airfare fluctuations, ensuring swift responses to changing market conditions. By leveraging GANs, the model effectively generated genuine and contextually suitable future scenarios, accurately capturing intricate relationships within the airfare dataset.
Deep learning architectures, such as RNNs with LSTM units, played a vital role in handling sequential and temporal patterns, solving the problem of vanishing gradients, and enabling the model to respond to changing market conditions. The incorporation of factors such as seasonality, passenger load factor, month, and average cost augmented the dataset, hence enabling a more exhaustive analysis of airfare patterns.
Our system incorporates a real-time forecasting module that has the ability to adjust predictions in real-time when new data is acquired. The model evaluation results, utilizing metrics such as MAE and RMSE, underscored the accuracy and dependability of the model. This feature establishes it as a valuable instrument for airlines seeking to enhance pricing strategies and decision-making procedures.
The online engineering platform, with a RESTful API, facilitated seamless connection with live data streams, ensuring continual updates and recalibration of the prediction model. The framework’s proficiency in handling high-frequency data inputs renders it exceptionally well-suited for the dynamic and rapid nature of the aviation industry.
The proposed methodology represents a significant advancement in the field of real-time flight price prediction. The combination of cutting-edge technology, meticulous data preparation, and feature engineering has resulted in the creation of a model that not only adapts to market dynamics but also provides valuable insights for airlines in a highly competitive and constantly evolving industry. In the ever-changing aviation business, our methodology offers a robust and efficient solution, enabling you to make more informed decisions and implement optimized pricing strategies.
References
[1] N. Alapati et al., “Prediction of Flight-fare using machine learning,” in 2022 International Conference on Fourth Industrial Revolution Based Technology and Practices (ICFIRTP), Aug. 2022, pp. 134–138. doi: 10.1109/ICFIRTP56122.2022.10059429.
[2] C. K. Gupta, O. Nath Kushwaha, A. S. Yadav, and V. Kumar, “Dynamic Flight Price Prediction Using Machine Learning Algorithms,” in 2022 4th International Conference on Advances in Computing, Communication Control and Networking (ICAC3N), Sep. 2022, pp. 201–205. doi: 10.1109/ICAC3N56670.2022.10074448.
[3] R. Kashef, “Forecasting The Price of the Flight Tickets using A Novel Hybrid Learning model,” in 2023 IEEE World AI IoT Congress (AIIoT), Jun. 2023, pp. 305–310. doi: 10.1109/AIIoT58121.2023.10174373.
[4] D. De Silva, N. Mills, M. El-Ayoubi, M. Manic, and D. Alahakoon, “ChatGPT and Generative AI Guidelines for Addressing Academic Integrity and Augmenting Pre-Existing Chatbots,” in 2023 IEEE International Conference on Industrial Technology (ICIT), Apr. 2023, pp. 1–6. doi: 10.1109/ICIT58465.2023.10143123.
[5] J. Wang et al., “Generative Adversarial Networks for Electroencephalogram Signal Analysis: A Mini Review,” in 2023 11th International Winter Conference on Brain-Computer Interface (BCI), Oct. 2023, pp. 1–6. doi: 10.1109/BCI57258.2023.10078666.
[6] Mohana, D. M. Shariff, A. H, and A. D, “Artificial (or) Fake Human Face Generator using Generative Adversarial Network (GAN) Machine Learning Model,” in 2021 Fourth International Conference on Electrical, Computer and Communication Technologies (ICECCT), Sep. 2021, pp. 1–5. doi: 10.1109/ICECCT52121.2021.9616779.
[7] J. Kwon, C. Cha, and H. Park, “Multilayered LSTM with Parameter Transfer for Vehicle Speed Data Imputation,” in 2021 IEEE International Symposium on Circuits and Systems (ISCAS), Feb. 2021, pp. 1–5. doi: 10.1109/ISCAS51556.2021.9401352.
[8] F.-Z. Benjelloun, A. A. Lahcen, and S. Belfkih, “An overview of big data opportunities, applications and tools,” in 2015 Intelligent Systems and Computer Vision (ISCV), Mar. 2015, pp. 1–6. doi: 10.1109/ISACV. 2015.7105553.
[9] Y. Ali and M. A. Wani, “Improving Training of Generative Adversarial Networks,” in 2021 8th International Conference on Computing for Sustainable Global Development (INDIACom), Mar. 2021, pp. 81–86.
[10] Prabhat, Nishant, and D. Kumar Vishwakarma, “Comparative Analysis of Deep Convolutional Generative Adversarial Network and Conditional Generative Adversarial Network using Hand Written Digits,” in 2020 4th International Conference on Intelligent Computing and Control Systems (ICICCS), Feb. 2020, pp. 1072–1075. doi: 10.1109/ICICCS48265.2020.9121178.
[11] H. Sasaki, Y. Hidaka, and H. Igarashi, “Explainable Deep Neural Network for Design of Electric Motors,” IEEE Transactions on Magnetics, vol. 57, no. 6, pp. 1–4, Jun. 2021, doi: 10.1109/TMAG.2021.3063141.
Biography
Yuanyuan Guan, School of Tourism, Hainan University, Haikou, 570228, China. Research interest AI, ML, tourism.
Journal of Web Engineering, Vol. 23_2, 299–314.
doi: 10.13052/jwe1540-9589.2325
© 2024 River Publishers
