1 Introduction

E-commerce has fundamentally changed the decision-making of customers in the purchasing process. Few researchers, meanwhile, have looked into how social commerce elements affect consumers’ decision-making processes when making purchases. We have developed a proposed QNN model that originated from the principles of the social learning theory. Furthermore, it examines how that attitude influences their desire to purchase the goods and services. The primary determinants of purchase intention appear to be cognitive and emotive assessments, with cognitive appraisal having a stronger predictive ability than affective appraisal, according to the findings of a survey conducted among 243 real users of social commerce websites [1]. In e-commerce, fake online reviews have a big impact on online shoppers, sellers, and market efficiency. There has been research on fake reviews, but as of now, no comprehensive survey that can systematically assess and summarize its causes and effects has been conducted. An antecedent-consequence-intervention conceptual framework is presented in this study to establish a preliminary research agenda for the investigation of fake reviews [2]. This research endeavors to introduce the emerging field of quantum machine learning, which combines aspects of both machine learning and quantum computing. It delves into significant scientific literature pertaining to quantum machine learning, commencing with foundational concepts of quantum logic. Furthermore, it explores various elements and algorithms within the realm of quantum computing [3]. These advanced skills show the potential to tackle the problems presented by fraudulent reviews, providing stronger and more precise methods for protecting the credibility of online feedback systems [4]. The increasing prevalence of evidence-based training across various industries has led to a proliferation of different types of reviews. However, the diverse vocabulary used in these reviews may result in a lack of clarity and misuse of terms, potentially obscuring their overall impact. This study aims to provide a descriptive overview of the most common review formats, supplemented by illustrative case studies from commercial domains. Through scoping searches, the terminology associated with literature review and synthesis was scrutinized. The main review types were then analyzed using a straightforward analytical framework known as Search, AppraisaL, Synthesis, and Analysis (SALSA) [5].

QML offers an interesting solution for detecting fake reviews by combining scalability, robustness, and powerful data processing techniques. The accomplishment of QML models is greatly impacted by their capacity to extract vital insights from a wide range of data sources [6]. The QML system rigorously analyses user profiles along with review material, taking into the details such as browsing patterns, writing styles, and behavioral characters. By integrating these diverse domains, it successfully differentiates between genuine and deceptive reviews. QML has several significant benefits in addressing the propagation of fraudulent reviews. First and foremost, big data allows for the handling of diverse datasets by using its scalability. Utilizing its ability to handle the vast amount of e-commerce reviews is necessary for detecting fake ones. The counterfeit reviews frequently integrate seamlessly with authentic reviews. Furthermore, the QML’s scalability enables the detection and analysis of review data, allowing the prompt identification and mitigation of the negative impacts caused by fraudulent reviews. A QML model is crucial for protecting both businesses and customers by promptly dealing with fraudulent testimonials. QML’s powerful data processing capabilities exceed conventional models by extensively analyzing user behavior and account information of the reviewer. This widespread methodology allows for a sophisticated comprehension, enabling the system to precisely distinguish between authentic and misleading reviews [7].Quantum Machine Learning (QML) enables businesses and customers to read online reviews more positively. An essential feature of the QML model is its ability to evaluate sentiment in reviews. By precisely interpreting the sentiments carried in user comments regarding products and services. QML allows us to accurately measure customer satisfaction levels and identify potential cases of fraudulent reviews more efficiently in e-commerce domains [8]. Identifying inconsistencies in the belief patterns of customers allows the model to improve its ability to recognize flawed information and identify possibly misleading reviews. The QML model goes beyond content analysis by including several analytical principles, such as sentiment evaluation and language patterns, to enhance its competencies. Gaining a more profound comprehension allows the model to distinguish more precisely between authentic and deceitful evaluations, thus enhancing the overall reliability of user data on online platforms. Moreover, the QML model recognizes that user behavior patterns assist as valuable markers of review reliability. The model can distinguish genuine users from those involved in suspect activities by examining factors such as public relationships between user accounts, frequency of contribution, and reliability [9].

The QML enables it to effectively discern reviews, hence enhancing the reliability of online review systems. Quantum Neural Networks (QNNs) are a notable advancement in quantum computing, apart from conventional Quantum Machine Learning (QML) techniques that rely on quantum principles. QNNs, which integrate quantum computing and neural network architecture, can efficiently process complex patterns and high-dimensional data [10]. Quantum Neural Networks (QNNs) show potential in overcoming the limitations of classic machine learning methods in areas like fraud detection, where advanced analysis of intricate data is crucial. QNNs provide enhanced scalability, noise resilience, and efficiency, which make them an attractive option for addressing real-world problems in several fields [11]. The combination of Quantum Machine Learning (QML) and Quantum Neural Networks (QNNs) provides a robust approach to tackle the ongoing problem of fraudulent reviews. This complete strategy, which includes evaluating sentiment, analyzing language, and assessing user behavior, effectively prevents the propagation of false material on internet platforms. Consequently, it cultivates confidence among consumers in the online marketplace and improves the overall caliber of information accessible to them. Quantum neural networks (QNNs) are an exceptional and pioneering advancement that can completely transform the artificial intelligence field. They can unlock new opportunities in data analysis and decision-making processes [12].

2 Literature Review

As consumers increasingly rely on online reviews for informed purchasing decisions of their goods and services. However, the widespread of fake testimonials poses a substantial threat to consumer trust and decision-making processes. Quantum Machine Learning (QML) emerges as a promising solution, offering distinct advantages in handling complex data with robustness and scalability crucial for real-time applications [13]. Compared to traditional approaches, research indicates that QML techniques have the potential to enhance the accuracy and efficiency of identifying fake reviews. Kariya et al. proposed a prominent encoding approach that extends the QSVM algorithm, enabling higher-dimensional vectors to be fed into the training data oracle [14]. Mishra et al. conducted another key work on the implementation of Quantum Decision Trees (QDT) to detect fake reviews. Unlike standard decision trees, QDTs classify data points using quantum entropy, providing a revolutionary approach that outperforms traditional algorithms in discriminating between legitimate and false reviews. These findings highlight the potential of quantum algorithms optimized for decision-making tasks to surpass existing methods in detecting false reviews [15]. DeepFake has recently become very popular; any audiovisual output developed using deep learning technology that appears realistic to viewers is referred to as “DeepFake.” Separating DeepFake materials from genuine ones with the human eye has always been a difficult task, but recent studies have shown that numerous technologies can deliver positive results, but with limitations [16]. QNN has a larger storage capacity and computing efficiency than conventional alternatives. This paper will discuss the evolution of QNN over the last six years in three parts: implementation approaches, quantum circuit models, and challenges encountered. The first portion, the implementation technique, focuses on several basic algorithms and theoretical frameworks for developing QNN models, such as VQA. The second section covers several quantum circuit models of QNN, such as QBM, QCVNN, and so on. The final section discusses some of the most difficult difficulties now being faced. In short, this field is still in its early stages, but it holds both magical and practical relevance [17]. Quantum mechanics is the study of nature and its behavior at the atomic and subatomic levels. Many issues can be solved more easily using quantum mechanics because of its unique quantum features, such as superposition and entanglement. The generative adversarial network (GAN), a significant machine learning technology that excels at generative tasks, has been enhanced with quantum versions. A QGAN can have a wholly quantum or hybrid quantum-classical architecture, which may necessitate additional data processing in the quantum-classical interface. Similarly to classical GANs, QGANs are trained with a loss function such as max likelihood, Wasserstein distance, or total variation. The gradients of the loss function can be generated using the parameter-shift method or a linear combination of unitaries to update network parameters [18]. Quantum Machine Learning provides a platform for various mining techniques that take advantage of recent advances in quantum computing. Quantum Machine Learning (QML) focuses on rapid problem-solving synthesis utilizing a quantum framework and various methods [19]. Quantum computing methods to various classic target identification and localization techniques, particularly for radar images, are studied and discovered to pose a variety of quantum statistics and quantum measurement concerns with far-reaching implications. Such methods are computationally demanding, requiring the coherent processing of commercial data sets to extract a small number of low-profile objects from a congested environment [20]. Quantum computing is a rapidly emerging topic that has garnered significant funding from industry and academics over the last decade. Several real quantum computers are now freely available to users via cloud services, with some implementations handling hundreds of qubits [21]. Non-linear and unexpected faults can produce noise that bypasses the linear error correction techniques that optical receivers typically use. Because quantum machine learning methods offer advantages over classical algorithms, we anticipate that optical signal processing will benefit from these advantages [22]. Because of the growing amount of Internet purchases, the identification of fake consumer reviews has received a lot of attention in recent years. However, as evidenced by recent research, the semantic meaning of reviews may be very essential for text classification. Furthermore, the emotions disguised in the reviews may be another evidence of fraudulent content. To increase the efficacy of false review identification, we present two neural network models that use classic bag-of-words, word context, and consumer emotions [23]. In e-commerce, customer reviews can have a substantial impact on an organization’s revenue. Before making a purchase decision for any goods or service, online users consult reviews. As a result, the legitimacy of online evaluations is critical to organizations, since it can have a direct impact on their reputation and profitability. To address the issue of fake reviews, this survey study discusses the task of detecting bogus reviews, summarizing the existing datasets, and gathering methods. It evaluates the current feature extraction strategies [24]. The proposed model combines multiple quantum algorithms specifically designed to tackle the problems associated with fraudulent online reviews. Quantum Neural Networks (QNNs) are algorithms that use quantum principles to process intricate patterns and high-dimensional data. Quantum neural networks (QNNs) are highly effective in identifying fraudulent patterns that may go undetected by conventional methods, hence significantly improving the accuracy of fraud detection systems. Furthermore, Quantum Generative Adversarial Networks (QGANs) enhance the training dataset and support the model’s ability to withstand adversarial attacks. Quantum-assisted Semi-supervised Learning (QSSL) utilizes unlabeled data to improve the performance of the model in circumstances where labeled data are scarce. Quantum algorithms provide sophisticated abilities to interpret and analyze data, allowing the hybrid model to successfully counteract fraudulent activity in online reviews.

3 Proposed Algorithm

Although machine learning algorithms have proven to be successful in detecting fraud, they frequently encounter difficulties in keeping pace with the advancing complexity of fraudulent activities. Quantum computing offers a novel and distinctive method of computation, as depicted in Fig. 1, due to its exceptional computational skills.

Fig. 1
figure 1

Quantum neural networks for fake review detection

The primary objective of this strategy is to enhance the security and safety of the digital environment by utilizing both quantum and computing technologies, as depicted in Fig. 2.

Data Preprocessing

Data preparation is an early step in the data analysis pipeline that impacts the efficiency of machine learning models. This process involves the collection and organization of unrefined data, which includes responsibilities such as removing errors, handling missing records, and formatting the data for efficient analysis. Preprocessing confirms the cleanliness and organization of data, for the development of accurate and reliable fraud detection systems.

Quantum Feature Extraction

Quantum algorithms are capable of processing complex patterns and vast amounts of data, making them fit for extracting features in cases involving fraud detection. Methods such as amplitude encoding and quantum principal component analysis (PCA) facilitate the effective use and representation of data. This enhanced understanding of the data enables the creation of more precise and efficient fraud detection systems. Quantum computing shows potential for strengthening the ability to extract features, leading to enhanced performance of fraud detection systems.

Quantum Neural Networks

QNNs, in contrast to conventional neural networks, utilize ideas derived from quantum mechanics to process data. They employ qubits, which are the quantum equivalent of classical bits, along with quantum gates to modify data. QNNs have a distinct architectural design that gives them a significant edge in fraud detection when it comes to dealing with intricate classification works that are essential for detecting fraudulent actions. Here’s how QNNs achieve this.

  1. 1.

    Analyzing Complex Boundaries: Visualise a decision boundary as a separate line that splits valid transactions from fraudulent ones. Quantum neural networks (QNNs) skillfully direct through these complexities, hence boosting their ability to precisely detect fraudulent reviews.

  2. 2.

    Uncovering Subtle Patterns: These patterns frequently appear as subtle signals hidden within the data, functioning as markers of future fraudulent procedures. Quantum neural networks’ sensitivity to these subtle and intricate patterns allows them to accurately detect fraudulent activity with improved precision.

By combining these qualities, QNNs become a powerful tool for detecting fraud reviews. A key technique in a Quantum Neural Network (QNN) entails the use of quantum gates and qubits to implement a function (f) on the input data.

The Hybrid Quantum-Classical Model

The Hybrid Quantum-Classical Model integrates the advantages of classical and quantum methods to tackle fraud detection. Traditional neural networks are proficient in fraud detection, but they do not possess the optimization abilities seen in quantum algorithms. On the other hand, quantum algorithms innate parallelism of quantum physics to achieve faster processing. The Hybrid Quantum-Classical Model leverages these complementary strengths as follows: Quantum Optimization and Classical Processing Power.

The model capitalizes on the strengths of each approach, resulting in enhanced Efficiency and improved Efficacy.

The Hybrid Quantum-Classical Model serves as a hands-on solution that connects the existing limitations with the future possibilities of quantum computing in the field of fraud detection. The purpose of these considerations was to guarantee that the hybrid model could efficiently utilize the advantages of both classical and quantum techniques, while also tackling the difficulties related to scalability and practical implementation in fraud detection systems.

Model Training

Classical optimization algorithms are crucial in this phase since they assist the neural network acquire successfully from labeled instances, revealing hidden patterns in the dataset. Quantum Optimisation is employed for targeted optimization tasks during training, such as feature selection or fine-tuning model parameters. Quantum algorithms are utilized for activities in which they provide processing benefits, leading to increased efficiency in crucial optimization domains. The integration of many methods finally results in the creation of a more precise and resilient model for detecting false reviews.

Enhancing Optimisation Using Quantum Speedup

Quantum computing offers an advantage in resolving complex optimisation issues, related to Quadratic Unconstrained Binary Optimisation (QUBO). Quantum algorithms are highly capable of efficiently overcoming difficulties, offering solutions that are specifically designed for jobs that resemble QUBO. This acceleration is of great importance, especially in fraud detection systems, where making decisions immediately is crucial. To measure the acceleration obtained by quantum computing, Quantum Speedup = (Time taken by conventional methods) / (Time expended by quantum algorithms). This formula highlights the significant improvements in performance that quantum computers have.

Fig. 2
figure 2

A proposed framework for quantum neural networks to detect false reviews

The high-speed processing capabilities of quantum computing have significant potential to revolutionize difficulties such as fraud detection. An important difficulty in quantum computing is the coherence and error rates of qubits, which play a vital role in influencing the correctness and reliability of quantum computations. Short coherence periods and frequent errors can delay the competence of Quantum Machine Learning (QML) systems in fraud detection. To fully leverage the abilities of quantum computing in avoiding fraudulent operations, it is crucial to address these obstacles.

Data Characteristics and Algorithmic Selection

Choosing the best suitable methodology is critical for optimizing the effectiveness and precision of the system. The relationship between these parameters can be represented by a function, f(coherence time, error rate, data attributes, fraud detection requirements). This optimization confirms that the selected technique aligns efficiently with the distinct characteristics of the fraud data and the operational requirements of the detection task. Fraud detection systems can effectively counter fraudulent operations by carefully evaluating these parameters and optimizing the function, thus leveraging the full potential of quantum computing.

Anticipating the Future

The enduring progress in quantum computing technology is generating enthusiasm for the creation of advanced fraud detection techniques. Quantum algorithms can significantly accelerate fraud detection processes, allowing for quicker identification of fraudulent activities in comparison to conventional approaches. Furthermore, Quantum Machine Learning (QML) models result in more reliable detection outcomes and a decrease in false positives. Quantum-powered systems can enhance proficiency by optimizing resource allocation and processing power for fraud detection tasks. This will ultimately improve security and confidence in digital transactions and platforms.

4 Recommended Formula for Determining Phony Evaluations

To access a quantum computing platform, either through cloud services or local quantum simulators, are essential like IBM Quantum Experience, Google Quantum Computing, Rigetti Forest, or Microsoft Quantum Development Kit.

The quantum programming framework such as Qiskit (for IBM Quantum Experience), Cirq (for Google Quantum Computing), or pyQuil (for Rigetti Forest) is necessary for developing and executing quantum algorithms.

For Machine Learning Libraries such as TensorFlow, PyTorch, or scikit-learn may be required for classical preprocessing of data, integration with quantum components, or post-processing of results.

# Import necessary libraries.

import neural_networks_classical as cnn # Library used for Classical neural networks.

import qfe as quantum_function_removal # Library used for Quantum function removal.

import qnn # Library used for Quantum neural networks.

import qopt # Library used for Quantum optimization.

# Preprocess the raw data to prepare it for evaluation.

preprocessed_data = preprocess_data(raw_data).

The code imports the essential libraries for both traditional and quantum computing. Subsequently, the unprocessed data goes through preprocessing, encompassing essential procedures such as data scrubbing, managing null values, and structuring the data for consequent analysis.

# Quantum Feature Extraction.

# Extract quantum features from preprocessed data.

quantum_features = quantum_function_removal.extract_features(preprocessed_data).

Quantum algorithms are used to extract quantum features from the pre-processed data. Techniques like amplitude encoding and quantum amplitude amplification are used to record best complex patterns that are vital for fraud detection.

# Initialization of Quantum Neural Network (QNN).

# Initialize the proposed Quantum Neural Network (QNN).

quantum_nn = qnn.initialize_quantum_nn().

During this stage, a Quantum Neural Network (QNN) is a specialised type of neural network specifically planned to function on quantum computers. The system employs quantum gates and qubits to manipulate data and is precisely designed for main classification tasks.

# Training Hybrid Models.

# Train classical neural network on preprocessed data.

classical_nn = train_classical_nn(preprocessed_data).

# Train quantum neural network on quantum features.

quantum_nn = train_quantum_nn(quantum_nn, quantum_features).

Both traditional and quantum neural networks are trained simultaneously. The conventional neural network learns from the processed data, while the quantum neural network is trained using the extracted quantum parameters. This approach ensures that a diverse range of data representations are obtained, enhancing the overall learning process.

# Merging Models for Hybrid Model.

# Merge classical and quantum models to create a hybrid model.

hybrid_model = merge_models(classical_nn, quantum_nn).

A hybrid model is formed through the integration of classical and quantum models. This model synergistically integrates the advantages of both paradigms, hence improving the efficiency and efficacy of the fraud review detection system.

# Optimization using Quantum Speedup.

# Optimize the hybrid model using quantum speedup techniques.

optimized_model = qopt.optimize_model(hybrid_model).

Quantum speedup techniques are utilized to optimize the hybrid model, particularly beneficial for functions such as quadratic unconstrained binary optimization (QUBO). Quantum algorithms expedite optimization processes, transforming the approach to complex analyses, such as fraud detection.

# Inductive and Prediction.

# Make predictions on new data using the optimized model.

predictions = make_predictions(optimized_model, new_data).

The optimized hybrid model is employed to make predictions on novel data. Improved precision in identifying fraudulent reviews enhances the effectiveness of the fraud detection system.

# Interpretation of Post-Processing Results.

# Analyze and interpret the post-processing results.

interpreted_results = interpret_results(predictions).

Ultimately, the predicted results generated by the model undergo comprehensive assessment and examination. This step involves analyzing the accuracy of predictions, evaluating the performance of the model, and refining detection tactics based on the interpreted results.

The method presented outlines a systematic approach for integrating classical and quantum computing methodologies to detect fraud in online reviews. This structured framework comprises several key stages, beginning with data preparation, followed by quantum feature extraction, training hybrid models, optimization utilizing quantum speedup, prediction, and interpretation of outcomes. To further enhance performance, additional procedures such as model review, refinement based on feedback, and continuous learning from fresh data can be incorporated. This comprehensive strategy capitalizes on the strengths of both classical and quantum computers to improve precision and effectiveness, particularly in the domain of fraud detection and prediction. Continuous improvement is prioritized through ongoing model evaluation, adjustments based on feedback, and adaptation to evolving data landscapes, ensuring consistent and accurate identification of fraudulent reviews over time.

5 Experimental Results

Applying diverse machine learning algorithms is paramount in detecting and avoiding fraudulent activity, and safeguarding businesses from financial harm and illegal transactions based on fake reviews. Within this landscape, the Quantum Neural Network (QNN) emerges as a highly promising innovation in fraud detection algorithms. Its integration into fraud detection systems proposes a sophisticated approach that surpasses conventional machine learning and artificial intelligence methods. The higher performance of QNN can be attributed to the advanced processing capabilities of quantum computers, empowering more complete investigations and keen accuracy in identifying deceptive actions in the online review dataset. The adoption of QNN represents a significant milestone in addressing the increasing complexity of fraudulent activities, providing administrations with an effective means to fortify the resilience of financial systems against emerging risks. The provided paragraph outlines a comparison of various quantum machine learning algorithms, including Quantum Support Vector Machine (QSVM), Quantum Decision Tree (QDT), Proposed Quantum Neural Network (QNN), Quantum Generative Adversarial Network (QGAN), and quantum-assisted semi-supervised learning (QSSL). Performance metrics such as precision, recall, F-1 score, and accuracy are used to evaluate these algorithms’ proficiency in leveraging quantum computing principles as shown in Table 1.

Table 1 Performance metrics of QNN compared with traditional Algorithms

Research investigations indicate that the Proposed QNN demonstrates significant advantages in both speed and accuracy, particularly in targeted optimization tasks. This suggests that QNN is highly effective in detecting fake reviews compared to conventional algorithms. The equations mentioned leverage quantum gates and qubits, allowing for information manipulation in ways unattainable by classical formulas. Quantum Generative Adversarial Networks (QGANs) have garnered attention in research due to their rapid convergence rate and the quality of samples they produce. Furthermore, harnessing quantum properties for data representation has been shown to improve performance, particularly in contexts with sparse labeled data, such as transparent quantum-assisted semi-supervised learning methods. However, despite their inherent potential, these algorithms encounter obstacles like decoherence, vulnerability to quantum noise, and the restricted accessibility of quantum hardware. Among all the algorithms assessed, Quantum Neural Networks (QNNs) emerge as the most precise, boasting an impressive accuracy rate of 0.86 as shown in Fig. 3.

Fig. 3
figure 3

Tests of the accuracy of the Amazon Fake Review Dataset to identify fraudulent reviews

Additionally, Quantum Neural Networks (QNNs) exhibit a short training period and demonstrate the lowest time complexity among the evaluated algorithms. Quantum-assisted Semi-supervised Learning (QSSL) achieves an accuracy of 0.83. However, traditional algorithms such as Quantum Support Vector Machines (QSVM), Quantum Decision Trees (QDT), and Quantum Generative Adversarial Networks (QGANs) show notably lower precision values, ranging from 0.79 to 0.82, as depicted in Fig. 3.

Table 2 Assessment of algorithms considering accuracy, time complexity, loss function, and training time

QNN has the lowest temporal complexity, followed by QSSL, QDT, and QSVM. QGAN is the furthest time-complex algorithm. The proposed Quantum Neural Network (QNN) uses a categorical cross-entropy loss function. In comparison, Quantum Semi-Supervised Learning (QSSL), Quantum Support Vector Machine (QSVM), Quantum Decision Tree (QDT), and Quantum Generative Adversarial Network (QGAN) utilize contrastive, cross-entropy, and Wasserstein distance losses, respectively. QNN has the shortest training span, followed by QSSL, QDT, and QSVM, whilst QGAN necessitates the longest training period. To summarise, the examination of the outcomes presented in Table 2 demonstrates that the suggested Quantum Neural Network (QNN) emerges as the most effective method for identifying misleading reviews.

The experimental results demonstrate that the proposed Quantum Neural Network (QNN) algorithm outperforms challenging algorithms across a variety of metrics, including Accuracy, Precision, False Positive Rate, False Negative Rate, F-1 Score, and Area Under the ROC Curve (AUC). These results highlight the extraordinary efficacy of the QNN algorithm in identifying fraudulent reviews, which could eventually reduce the misclassification of genuine transactions. Additionally, Table 3 offers supplementary critical criteria for assessing the effectiveness of fraud detection algorithms, supplementing the initial four metrics (Precision, Recall, Accuracy, and F-1 Score).

Table 3 Quantum Machine Learning Algorithm Comparison
figure a

Systems of measurement like as False Positive Rate (FPR), False Negative Rate (FNR), Area Under the ROC Curve (AUC), and Cost-benefit analysis (CBA) are crucial in evaluating the performance and efficacy of fraud review detection systems in e-commerce. A lower False Positive Rate shows a greater accuracy in distinguishing between genuine transactions and fraudulent ones, whereas a decreased False Negative Rate indicates enhanced identification of fraudulent activities. The Area Under the ROC Curve provides a quantitative assessment of the system’s capacity to distinguish between fraudulent and non-fraudulent transactions at diverse thresholds. A larger AUC value indicates better performance. Cost-benefit analysis is assessing the costs and benefits of connecting a fraud detection system. The costs include the setup, maintenance, and inspection of false positives, while the benefits include fraud prevention and improved customer satisfaction. The Quantum Neural Network (QNN) algorithm is suggested because of its improved performance compared to other approaches in terms of False Positive Rate (FPR) and Area Under the Curve (AUC). It effectively reduces the misclassification of legal transactions and improves accuracy in recognizing fraudulent ones. Although Quantum-assisted Semi-supervised Learning (QSSL) holds potential, Quantum Support Vector Machines (QSVM), Quantum Decision Trees (QDT), and Quantum Generative Adversarial Networks (QGAN) have greater False Positive Rates (FPRs) in comparison to their Area Under the Curve (AUC) values. Considering performance and cost-benefit analysis, QNN is the most efficient choice for fraud review detection. However, the usefulness of any approach depends on the specific requirements of the fraud detection program.

6 Conclusion

The integration of quantum algorithms such as Quantum Neural Networks (QNNs), Quantum Generative Adversarial Networks (QGANs), and Quantum-assisted Semi-supervised Learning (QSSL) indicates a significant advancement in combating fraudulent actions, particularly in the realm of identifying counterfeit reviews in e-commerce. Quantum algorithms surpass conventional machine learning techniques in terms of accuracy and effectiveness, despite facing challenges like decoherence and quantum noise. Out of the algorithms considered, QNN stands out as the topmost performer, with an accuracy level of 0.86 and fast training epochs. The extraordinary effectiveness of quantum-assisted methods such as QSSL in accurately and rapidly detecting fraudulent activities. Nevertheless, it is crucial to recognize the need for ongoing investigation and improvement of quantum technology to overcome the existing limitations. Joining the capabilities of quantum computers has great potential to strengthen the efficiency of fraud review detection systems, ultimately leading to improved financial stability and increased consumer protection against illegal actions. To effectively utilize the potential of quantum computing, future research should prioritize overcoming current obstacles and investigating the abundant possibilities offered by quantum algorithms in a practical state of affairs. By using the capabilities of quantum computers, we may enhance the flexibility of fraud detection systems, thus promoting trust and assurance in digital transactions and online platforms by identifying genuine reviews.