Improving Data Market Implementation Using Gray Relational Analysis in Decentralized Environments

Authors

• Praveen Kumar Kanumarlapudi Induvidual

Abstract

 Introduction:

Data marketplaces have emerged as key platforms that Facilitating the sharing of information and services between businesses and people, and even within organizations. These marketplaces aim to make data more accessible, interoperable, and reusable, supporting the principles of FAIR and the concept of data democratization. As organizations increasingly rely on machine learning (ML) for real-time decision-making, data marketplaces offer a promising solution to efficiently access high-quality training data and drive innovation and business growth across multiple sectors.

Research signification: The research highlights the growing importance of data markets in enabling effective data sharing and monetization, while addressing critical challenges such as data security, trust, and control. By examining different market models – centralized, decentralized, federated – and assessing parameters such as real-time data support and smart contract integration, the study provides valuable insights for businesses looking to leverage data as a strategic asset. The findings have implications for industries aiming to improve supply chains, enhance AI capabilities, and foster sustainable digital ecosystems.

Research methodology:

Alternatives: Centralized Data Marketplace, Decentralized Block chain-based Marketplace, Federated Data Exchange Model, Subscription-based Data Platform, Pay-per-use Data Access Model.

Evaluation parameters : Real-time Data Support, Reputation Management, Smart Contract Integration, Monetization Flexibility.

Result: The results Subscription-based Data Platform achieved the highest rank, while Decentralized Blockchain-based Marketplace the lowest rank being attained.

Conclusion:  “Data Marketplace Implementation to the GRA, demonstrates that Subscription-based Data Platform achieves the highest ranking.”

Data markets are primarily used to exchange information and services between businesses or people. But they are also having significant potential for internal use within organizations. Research indicates that a large portion of corporate data remains untapped. In this context, concepts such as the FAIR principles, which emphasize making data discoverable, accessible, interoperable, and reusable, and the concept of data democratization are frequently explored in academic and professional literature. Data democratization aims to encourage and enable a broad range of employees within an organization to discover, understand, access, use, and share data, while ensuring compliance with data security and regulatory standards. [2] Machine learning (ML) algorithms need High-quality training data is needed to work effectively. However, obtaining such relevant data is very challenging organizations, especially those just starting to integrate ML into their operations. This challenge becomes even more important as businesses increasingly need to make real-time predictions, such as dynamic pricing in ride-sharing services or personalized coupon distribution in the retail and food sectors, where outdated data quickly loses value. To solve this problem, the goal is to create a data marketplace: a real-time trading platform designed specifically for the exchange of ML training data. [3] The rationale behind this approach is that given the evolving nature of data markets and the increasing interest they attract from non-academic audiences, there may be greater engagement and discussion on platforms such as Social media.

As a result, conventional citation-based metrics may not fully capture the influence of these articles. [4] The rapid growth of the Internet and mobile phone usage is driving significant changes in various aspects of life. According to a report by We Are Social and Hoot suite, more than 4 billion people worldwide are now internet users, while the number of unique mobile phone users has surpassed 5 billion. A major consequence of this trend is the shift from traditional markets to electronic markets (e-markets). More and more people prefer to make purchases through online platforms on their smart phones rather than visiting physical stores. [5] Data marketplaces create structured frameworks that facilitate the sharing of Making data easily discoverable and adaptable to the formats consumers need. These platforms bridge the gap by understanding consumer data needs and communicating them to data owners. Data owners are encouraged to share their information because companies can make money when customers are willing to pay for access. Customers are also urged to communicate their data needs when using the platform is addressing challenges of discovery and integration on their behalf in exchange for some form of compensation, such as monetary rewards. By enabling a transparent flow of information and providing incentives, data marketplaces create value for all participants involved. [6] One of the key challenges facing private data marketplaces is that users have limited control over how their data is used by buyers. Users also need to place a significant amount of trust in the marketplace to enforce the usage rules they set for their data. This includes trusting that the platform is not secretly conducting unauthorized calculations or operations on their data. [7] This study aims to address a gap in existing research by examining the post-implementation phase A B2B e-marketplace system hosted in the cloud. This study looks at a system that a large UK contracting company has implemented using a long-term case study methodology with three key supply chain partners to improve its procurement processes. The solution features a cloud-based front-end and back-end interface with the contractor and company ERP systems and its suppliers.

The study focuses in particular on the fact that in the first year of implementation of the system, its use has not yet been fully enforced within the organization. [8] Data sharing between companies offers significant opportunities for the industrial sector, supporting process optimization, the development of innovative products and services, and promoting both economic growth and sustainability. In this context, data is emerging as a valuable asset that can be traded on data markets. Despite this potential, data markets in the industrial sector have seen limited success. A major contributing factor is the reliance on centralized market structures, which limit the control that data providers have over their data and are heavily dependent on the trust and impartiality of the market operator. [9]Users can create new IoT/M2M applications and build IoT/M2M infrastructures with OpenMTC, an implementation of the oneM2M standard. Each component is represented as a resource in the OpenMTC framework, and the platform makes it easy to create, modify, update, and delete these resources. In contrast, FIWARE is an open source project that seeks to provide international guidelines for context-aware data management to facilitate the development of intelligent applications. The FIWARE platform connects IoT devices to cloud-based big data services and context-aware information management by integrating multiple components and providing a set of APIs. [10] As a result, cities are increasingly considering the concept of a community-driven IoT data marketplace. This model enables local IoT device owners to sell their data through a marketplace, which provides valuable insights to application developers, including those within city administrations. A growing body of literature highlights The value and benefits of these systems.

The architecture and essential components of IoT data markets have not received much attention in research. We previously demonstrated the first implementation of the I3 data market. [11] This study analyzes e-commerce data from a case study of one of the largest online marketplaces to calculate the consumer price index (CPI) at the local level Indonesia. To collect the required data, a web scraper will be developed. Using the collected data, this study will compare consumption values ​​obtained from market sales data with those reported by BPS-Statistics Indonesia.In addition, the research aims to analyze whether the CPI patterns from both market data and BPS are consistent, which provides insights to stakeholders to inform better decision-making policies. [12] The purpose of the handshake is limited to the transmission of SDA data, which is one of many possible uses for an orbital market. Unlike data exchange, the trading of services involves fundamentally different mechanisms and is difficult to verify from an external perspective, such as the view of a consensus agent. Physically based methods for recording service exchanges - especially those required for servicing in space - offer a promising way to support these types of transactions in the market. [13] Technological advancements have helped some marketplaces grow their businesses by adopting various marketing strategies that play a key Digital marketing plays a Plays a role in attracting additional customers and increasing revenue. There are numerous benefits to digital marketing, and Indonesian market applications should make extensive use of these strategies. How do users interact and explore the interesting features in these applications influences their opinions and judgments about various marketplaces in the country?[14] The current implementation uses C functions to perform the encryption and decryption tasks required by the customer, supplier, and marketplace. These procedures read a file containing the data to be processed, execute the necessary procedures, and then save the result in another file. Other agents can then retrieve these files. via various protocols such as HTTP, SMTP, or SOAP. Since the full set of MAGNET components is not yet available, this approach allows for seamless integration into the complete system in the future. Below is an overview of the functions implemented within the proposed protocol and their roles.

Alternatives:

Centralized Data Marketplace: A centralized platform where data providers and consumers interact through a single authoritative entity that manages data storage, access control, transactions, and monetization mechanisms. This typically ensures efficiency, but can raise concerns about control, security, and scalability.

Decentralized Block Chain-Based Marketplace: A distributed data trading platform built on blockchain technology that enables peer-to-peer data exchange without centralized control. It uses smart contracts to automate transactions and ensure transparency, data integrity, and trust among participants.

Federated Data Exchange Model: Data remains within its original location (data sovereignty is maintained), but insights are shared or aggregated using secure query mechanisms. This is ideal for privacy-sensitive environments such as healthcare and finance.

Subscription-Based Data Marketplace: A data marketplace model where consumers pay a recurring fee (monthly/yearly) to access datasets or services. This provides providers with predictable revenue and seamless access to consumers, often used by data-as-a-service (DaaS) vendors.

Pay-per-use data access model: A usage-based monetization model in which users pay only for the amount or frequency of data consumed. This provides flexibility for consumers and allows providers to efficiently monetize access to accurate data.

Evaluation Parameters:

Real-time Data Support: The ability of the marketplace to facilitate seamless exchange and access constantly updated data streams with low latency is critical for time-sensitive applications such as financial trading, health monitoring, or IoT systems.

Reputation Management: Methods for assessing and reflecting the reliability and trustworthiness of data providers and consumers based on historical performance, data quality, and user feedback. This builds trust and credibility within the ecosystem.

Smart Contract Integration: The ability of the system to implement programmable contracts that automatically enforce agreed-upon rules, terms, and conditions during data transactions. This reduces reliance on intermediaries and improves transaction security.

Monetization Flexibility: The extent to which the marketplace supports a variety of pricing and revenue models, such as subscriptions, pay-per-use, auctions, or licensing - allowing providers to tailor their offerings to meet diverse customer needs and market dynamics.

Method:The coefficient of variation (ξ) is a key parameter in Deng's GRA model, although its importance is often overlooked by researchers. Many studies in the literature assume the value of ξ to be 0.5 without providing a justification for this choice. It is generally assumed that ξ is equal to 0.5, although the reason behind this assumption is not universally established. The definition of Deng's GRA model itself implies that ξ lies within the range (0, 1), which implies that ξ is a dynamic parameter rather than a fixed one. Furthermore, some scholars argue that variations in ξ do not affect the final ranking of factors or alternatives determined by the GRA model, which further complicates the matter. [17] The better the user's needs (often subjective) are met, the more favorable the outcome will be, and vice versa. Due to the significant difference between the criteria used to evaluate user opinions and the objective metrics derived from the products, TOPSIS and its extension techniques cannot directly generate subjective user-defined product rankings.

However, KE provides a practical method for processing user-based criteria evaluations, allowing TOPSIS and its extensions to replace processed criteria for subjective product evaluation. In addition, since user feedback involves uncertainty, methods such as GRA-TOPSIS can also play a role. [18] An effective approach for classifying and monitoring bearing errors has been introduced. The proposed model for bearing signal classification It consists of two main stages. Features are initially extracted and then classified using a Gray Relationship Analysis (GRA) model in the second stage. Statistical features are identified during feature extraction by examining the neighbors of each value in their one-dimensional resonance signals. After acquiring the 1D-LBP signals, the statistical features are extracted from them and then classified using a GRA model. [19] This study uses GRA, TOPSIS, and MOORA methods to determine the best stocks listed on the Vietnamese stock market between 2016 and 2019 and evaluate the financial performance of 13 agricultural companies. Financial ratios are weighted using the AHP approach, and the options in this case study are ranked using GRA, TOPSIS, and MOORA. The proposed paradigm streamlines the assessment of financial performance process, and Spearman’s rank correlation test confirms that there are no significant differences between the rankings generated by the three methods. This distinguishes the study from previous research in this field. [20] Traditional GRA methods face difficulties in addressing intuitively ambiguous MADM problems when weight information is incomplete. There is a major challenge in deriving attribute weights from both intuitively ambiguous data and partially known weight information, which is fundamental to traditional GRA methods and warrants further study. This paper aims to improve the GRA framework by introducing an algorithm to deal with MADM problems with intuitively fuzzy data, if attribute values ​​are represented as naturally fuzzy integers and attribute weights are not fully specified.

The format of this paper is as follows: The main ideas surrounding intuitively fuzzy sets are briefly reviewed in the next section. [21] We present an extended fuzzy GRA approach to MCDM problems, where the criterion values ​​are represented as linguistic variables using a space-valued triangular fuzzy numbers, and criterion weights are not specified. To validate these weights, Optimization models are developed using the basic concepts of traditional GRA. After that, a detailed explanation of the computational steps of the improved GRA technique for MCDM with interval-valued triangular fuzzy estimators is given, which enables more ranking and selection optimal option. [22] This paper focuses on deriving attribute weights from spherical linguistic fuzzy information provided by the traditional GRA method and from completely unknown attribute weight data - an important issue that requires attention. The goal of this study is to improve with sufficient attribute weight information and attribute values ​​expressed as spherical linguistically ambiguous numbers, the GRA technique is used to solve MADM problems in a spherical linguistically ambiguous environment. [23] This study uses an integrated methodology that combines GRA and CRITIC techniques to assess the innovation performance of BRICS countries. In fact, the innovation performance of five BRICS countries is evaluated based on eight criteria. The first stage involves determining the importance of each criterion using the CRITIC method, while the second stage ranks the countries based on their innovation performance using the GRA method. [24] The primary objective of this study is to investigate GRG and GRA optimization techniques for improving SMMD performance in washing machines. The proposed optimization technique effectively predicts the optimal design parameters that require low MR fluid and low power consumption, leading to improved performance of the SMMD damper. [25] The aim of this study is to use artificial neural networks to predict the USM process for Ti alloy while simultaneously optimizing the MRR and TWR response factors to achieve optimal machine performance.

This was accomplished by using the GRA method with weight generation by entropy measurement, considering selected process variables related to slurry, tool, and machine. To the best of the authors' knowledge, this method for titanium USM has not been previously documented in the literature.

In addition, cryogenic treatment of both the tool and the work piece is included in the process parameters. [26] As a result, an EW-GRA-TOPSIS model was developed to assess the environmental quality of the port sea area by integrating the EW, GRA, and TOPSIS methods. Initially, the EW method is used to objectively determine the weight of each evaluation indicator. Then, GRA and TOPSIS methods are combined to form a new evaluation approach that calculates relative proximity. This enables the environmental quality status of each station in the port sea area to be ranked based on their relative proximity. Finally, the environmental quality status of each station is classified according to the evaluation quality score. [27] The GRA-based PCA approach, which reduces A lot of variables linked to a finite number unrelated principal components, highlights the need for further research to achieve accuracy that meets the highest industry standards. Although this method provides an effective to address the challenges of uncertainty, multiple inputs, and unique data, there is a gap in comprehensive reviews regarding the mechanical properties of graphene-based composites. [28] This approach focuses on analyzing the relationships between behavior, posture, and boundaries in systems with incomplete operational mechanisms, sufficient behavioral data, and treatment experience, and identifying underlying problems. In contrast to traditional mathematical analysis, GRA provides a straightforward approach to analyzing continuous relationships or system behavior, even with limited data. In Gray theory, GRA is a technique used to evaluate the influence of different factors and their interactions within a Gray system. [29] GRA (Gray Relational Analysis) and Taguchi methods were combined to optimize the EDM process parameters for Ni-based super alloy (Inconel-718). The best values ​​for material removal rate (MRR), electrode wire rate (EWR), machining time, and shape tolerance (such as squareness and flatness) were 12 amps IP, 400 ms ton, and 10,400 ms TAF. With an effort to achieve the maximum MRR and the minimum form tolerance, the optimal solutions for the output responses were calculated. The results are verified by real experiments and confirm their effectiveness. The experimental findings have shown significant improvements in performance. As a result, the Gray relational approach simplifies the optimization process by transforming multiple response variables into a single response standard through normalization. [30] shows that when Influence of variables such as Ip and Tone on the surface integrity of AISI D2 tool steel is investigated, the EDM machining process becomes more stable with improved machining performance. RSM-based modeling and optimization were used to identify the optimal machining parameters. A similar study on the surface integrity of the same material was recently carried out using RSM as a multi-objective optimization technique in combination with GRA. The successful application of the GRA method led to an increase in MRR and a decrease in TWR. PCA-based optimization provides better results than constrained optimization and multiple-response S/N ratio methods because it effectively addresses the correlation between multiple responses.

1. INTRODUCTION
2. Materials a nd Method
3. Analysis a nd Discussion

Table 1. Alternatives

Table 2. Evaluation parameters

Table 3.Data Marketplace Implementation

The table 3 titled "Data Marketplace Implementation" presents numerical values for five alternatives (A1 to A5) across four criteria (C1 to C4). Each cell contains a quantitative measure, likely reflecting performance, cost, or relevance related to implementing a data marketplace. For example, A1 scores highest in C1 (53.0), while A5 shows the highest value in C2 (186.41). These figures could represent key factors such as data volume, transaction frequency, latency, or cost metrics that influence the effectiveness or feasibility of the data marketplace alternatives. The variation across alternatives and criteria highlights the need for multi-criteria analysis to select the best implementation option.

FIGURE 1.Data Marketplace Implementation

Figure 1 illustrates the Data Marketplace Implementation by comparing five alternatives (A1 to A5) across four criteria (C1 to C4). Each alternative exhibits distinct values for these criteria, reflecting diverse strengths and weaknesses. For instance, A1 leads in C1 with 53.0, whereas A5 significantly outperforms others in C2 at 186.41. Alternatives A2 and A3 demonstrate relatively balanced performance across all criteria, while A4 shows moderate values. These variations suggest trade-offs in implementing data marketplaces, where factors like cost, efficiency, or data throughput might differ.

TABLE 4.Normalized Data

Table 4 presents the normalized data for the five alternatives (A1 to A5) across four criteria (C1 to C4). Normalization scales the values to a common range, facilitating fair comparison. A1 scores the highest in C1 with a value of 1.0, indicating the best performance for this criterion. Conversely, A5 leads in C2 with a maximum normalized score of 1.0, highlighting its strength there. A4 dominates both C3 and C4 with top scores of 1.0. Other alternatives show mixed performance, with some criteria scoring zero, reflecting weaker areas. This normalized data aids in multi-criteria decision-making by balancing diverse evaluation factors.

TABLE 5.Deviation sequence

Table 5 shows the deviation sequence of alternatives (A1 to A5) from the ideal values across four criteria (C1 to C4). The deviations represent how far each alternative's normalized value is from the best possible score of 1. For instance, A1 has zero deviation in C1, indicating it is the benchmark for that criterion. In contrast, A5 has the highest deviation in C1, showing the weakest performance. Similarly, A5 has zero deviation in C2, meaning it performs best there. Lower deviation values signify better alignment with ideal performance, guiding decision-makers in evaluating and ranking the alternatives effectively.

Table 6.Grey relation coefficient

Table 6 presents the Grey Relation Coefficients for alternatives A1 to A5 across criteria C1 to C4. These coefficients measure the closeness of each alternative to the ideal reference, with values closer to 1 indicating stronger similarity and better performance. For example, A1 has a perfect coefficient of 1.0 in C1, showing it excels in that criterion, while A5 scores a perfect 1.0 in C2, indicating its dominance there. Conversely, lower coefficients such as A5’s 0.3333 in C1 reflect weaker performance. These coefficients help in assessing the overall quality and ranking of alternatives in the data marketplace implementation.

TABLE 7.GRG

Table 7 displays the Grey Relational Grade (GRG) values for alternatives A1 to A5, summarizing their overall performance across all criteria. The GRG aggregates the Grey Relation Coefficients into a single score, where higher values indicate closer similarity to the ideal solution and better overall ranking. Here, A4 achieves the highest GRG of 0.6919, suggesting it is the most optimal choice in the data marketplace implementation, followed closely by A5 with 0.6743. Alternatives A2 and A3 show moderate performance, while A1, despite strong performance in some criteria, ranks lower with a GRG of 0.6065.

FIGURE 2. GRG

Figure 2 illustrates the Grey Relational Grade (GRG) values for five alternatives (A1 to A5) in the data marketplace implementation. The GRG reflects the overall performance of each alternative by combining multiple criteria into a single score. Among them, A4 has the highest GRG of approximately 0.692, indicating it is the most favorable option. A5 closely follows with a GRG of 0.674. Alternatives A1 and A3 show moderate performance, while A2 has the lowest GRG value of 0.390, suggesting it is the least optimal among the options evaluated.

TABLE 8. Rank

Table 8 presents the ranking of the five alternatives based on their Grey Relational Grade (GRG) values.

Alternative A4 ranks first, indicating it is the best-performing option in the data marketplace implementation. Following closely, A5 holds the second position, demonstrating strong performance. A1 comes in third, showing moderate effectiveness. A3 is ranked fourth, while A2 ranks last, suggesting it is the least favorable choice among the alternatives. This ranking helps prioritize the alternatives for decision-making based on their overall performance across multiple evaluation criteria.

FIGURE 3. Rank

Figure 3 illustrates the ranking of alternatives based on their performance scores. Alternative A4 holds the top position (rank 1), indicating it is the most suitable choice in the data marketplace implementation. Alternative A5 follows closely with rank 2, showing strong overall performance. A1 ranks third, reflecting a moderate standing. A3 and A2 are ranked fourth and fifth respectively, with A2 being the lowest performer. This ranking visually prioritizes the options, guiding decision-makers to select the best alternatives according to their comprehensive evaluation metrics.

1. Conclusion

The growth and implementation of data markets are transforming how organizations acquire, share, and monetize data assets. Traditional approaches to data exchange are often plagued by inefficiencies, trust issues, and limited control over data providers. By adopting frameworks aligned with FAIR principles, data markets foster discovery, accessibility, interoperability, and reusability, which are critical to unlocking the untapped potential of corporate data. The democratization of data empowers a broad range of employees across organizational hierarchies to use data for informed decision-making, improving overall operational efficiency and innovation. Machine learning applications, in particular, benefit significantly from real-time access to high-quality, relevant training data—a need for predictive accuracy in dynamic environments such as ride-sharing and personalized marketing. Data markets designed with real-time support and reputation management help address this need, enabling seamless data flows while maintaining data security and regulatory compliance.

The integration of smart contracts through block chain technology further enhances transparency and automates the enforcement of data usage policies, alleviating concerns about unauthorized data manipulation. Despite the promise, challenges remain. Users often have limited control over downstream data usage, requiring marketplaces to build and maintain strong trust mechanisms. Centralized models, while efficient, can introduce concerns about scalability, control, and dependency, prompting the exploration of decentralized or federated alternatives that prioritize data sovereignty and privacy. Furthermore, the shift from physical to electronic marketplaces driven by widespread internet and mobile adoption underscores the importance of digital marketing strategies in emerging marketplace ecosystems, particularly in countries like Indonesia. By incentivizing data providers through monetization models such as subscription or pay-per-use, marketplaces align economic interests with data sharing, creating value for all participants. The study also emphasizes the need for more research on emerging market structures, evaluation metrics, and multi-criteria decision-making approaches that can handle subjective user perceptions and uncertain data quality. As data increasingly becomes a strategic asset, data markets will play a key role in shaping the future of digital commerce, AI innovation, and industrial growth, bridging the gap between data availability and actionable insights.

REFERENCE

1. Eichler, Rebecca, Christoph Gröger, Eva Hoos, Christoph Stach, Holger Schwarz, and Bernhard Mitschang. "Introducing the enterprise data marketplace: a platform for democratizing company data." Journal of Big Data 10, no. 1 (2023): 173.
2. Agarwal, Anish, MuntherDahleh, and Tuhin Sarkar. "A marketplace for data: An algorithmic solution." In Proceedings of the 2019 ACM Conference on Economics and Computation, pp. 701-726. 2019.
3. Abbas, AntragamaEwa, WirawanAgahari, Montijn Van de Ven, AnnekeZuiderwijk, and Mark De Reuver. "Business data sharing through data marketplaces: A systematic literature review." Journal of Theoretical and Applied Electronic Commerce Research 16, no. 7 (2021): 3321-3339.
4. Prihastomo, Yoga, Achmad Nizar Hidayanto, and HarjantoPrabowo. "The key success factors in e-marketplace implementation: A systematic literature review." In 2018 International Conference on Information Management and Technology (ICIMTech), pp. 443-448. IEEE, 2018.
5. Fernandez, Raul Castro, Pranav Subramaniam, and Michael J. Franklin. "Data market platforms: Trading data assets to solve data problems." arXiv preprint arXiv:2002.01047 (2020).
6. More, Stefan, and Lukas Alber. "You shall not compute on my data: access policies for privacy-preserving data marketplaces and an implementation for a distributed market using MPC." In Proceedings of the 17th International Conference on Availability, Reliability and Security, pp. 1-8. 2022.
7. Pala, Mesut, Francis Edum-Fotwe, KirtiRuikar, Nathan Doughty, and Chris Peters. "Post-implementation analysis of a B2B e-marketplace." Journal of Information Technology in Construction 24 (2019).
8. Jahnke, Nils, IlkaJussen, Thorsten Schoormann, and Frederik Möller. "Designing Federated Data Marketplaces in Industrial Production: Findings from a Prototypical Implementation." (2024).
9. Mehmood, Faisal, Shabir Ahmad, and DoHyeun Kim. "Design and implementation of an interworking IoT platform and marketplace in cloud of things." Sustainability 11, no. 21 (2019): 5952.
10. Zhao, Xiangchen, Kurian KaryakulamSajan, GowriSankar Ramachandran, and BhaskarKrishnamachari. "Demo abstract: the intelligent IoT integrator data marketplace-version 1." In 2020 IEEE/ACM Fifth International Conference on Internet-of-Things Design and Implementation (IoTDI), pp. 270-271. IEEE, 2020.
11. Haqqoni, Muhammad Ghozy Al, and SetiaPramana. "Implementation of marketplace data in the production of Consumer Price Index in Indonesia." Data Science 5, no. 2 (2022): 79-95.
12. Gordon, Nathaniel G., and Gregory Falco. "An On-Orbit Data Marketplace for Distributed Space Domain Awareness." IEEE Access (2025).
13. Nagababu. K, “Machine Learning Techniques in Fracture Mechanics a Comparative Study of Linear Regression, Random Forest, and Ada Boost Model” Journal of Artificial intelligence and Machine Learning., 2024, vol. 2, no. 2, pp. 1–13. doi: http://dx.doi.org/10.55124/jaim.v2i2.257
14. Sibarani, Hendra Jonathan. "Digital Marketing Implementation on Development and Prospective Digital Business (case Study on Marketplace in Indonesia)." Malaysian E Commerce Journal 5, no. 2 (2021): 64-68.
15. Jaiswal, Ashutosh, Yongdae Kim, and Maria Gini. "Design and implementation of a secure multi-agent marketplace." Electronic Commerce Research and Applications 3, no. 4 (2004): 355-368.
16. Mahmoudi, Amin, Saad Ahmed Javed, Sifeng Liu, and Xiaopeng Deng. "Distinguishing coefficient driven sensitivity analysis of GRA model for intelligent decisions: application in project management." Technological and Economic Development of Economy 26, no. 3 (2020): 621-641.
17. Quan, Huafeng, Shaobo Li, Hongjing Wei, and Jianjun Hu. "Personalized product evaluation based on GRA-TOPSIS and Kansei engineering." Symmetry 11, no. 7 (2019): 867.
18. Kuncan, Melih. "An intelligent approach for bearing fault diagnosis: combination of 1D-LBP and GRA." Ieee Access 8 (2020): 137517-137529.
19. Nguyen, Phi-Hung, Jung-Fa Tsai, Venkata Ajay KUMAR G, and Yi-Chung Hu. "Stock investment of agriculture companies in the Vietnam stock exchange market: An AHP integrated with GRA-TOPSIS-MOORA approaches." The Journal of Asian Finance, Economics and Business 7, no. 7 (2020): 113-121.
20. Wei, Gui-Wu. "GRA method for multiple attribute decision making with incomplete weight information in intuitionistic fuzzy setting." Knowledge-Based Systems 23, no. 3 (2010): 243-247.
21. Zhang, Shi-fang, San-yang Liu, and Ren-he Zhai. "An extended GRA method for MCDM with interval-valued triangular fuzzy assessments and unknown weights." Computers & Industrial Engineering 61, no. 4 (2011): 1336-1341.
22. Ashraf, Shahzaib, Saleem Abdullah, and Tahir Mahmood. "GRA method based on spherical linguistic fuzzy Choquet integral environment and its application in multi-attribute decision-making problems." Mathematical Sciences 12 (2018): 263-275.
23. Mittapally R (2024). Intelligent Framework Selection: Leveraging MCDM in Web Technology Decisions. J Comp Sci Appl Inform Technol. 9(1): 1-9.
24. Baki, Rahmi. "Comparison of Innovation Performances of BRICS Countries through CRITIC and GRA Methods." Gaziantep University Journal of Social Sciences 23, no. 4 (2024): 1561-1570.
25. Patel, Dipal M., Ramesh V. Upadhyay, and D. V. Bhatt. "Design and optimization of shear mode MR damper using GRG and GRA methods: experimental validation." Sādhanā 46 (2021): 1-17.
26. Dhuria, Gaurav Kumar, Rupinder Singh, and Ajay Batish. "Application of a hybrid Taguchi-entropy weight-based GRA method to optimize and neural network approach to predict the machining responses in ultrasonic machining of Ti–6Al–4V." Journal of the Brazilian Society of Mechanical Sciences and Engineering 39 (2017): 2619-2634.
27. Lang, Kun, LijunGu, Zhiying Chen, ChunhuiNiu, Lin Li, and Jinyuan Ma. "Ecological Quality Status Evaluation of Port Sea Areas Based on EW-GRA-TOPSIS Model." Sustainability 15, no. 11 (2023): 8809.
28. Kavimani, V., K. Soorya Prakash, Titus Thankachan, S. Nagaraja, A. K. Jeevanantham, and Jithin P. Jhon. "WEDM parameter optimization for silicon@ r-GO/magneisum composite using taguchi based GRA coupled PCA." Silicon 12 (2020): 1161-1175.
29. Cheng, Zhe, Niaoqing Hu, and Xiaofei Zhang. "Crack level estimation approach for planetary gearbox based on simulation signal and GRA." Journal of Sound and Vibration 331, no. 26 (2012): 5853-5863.
30. Kumar, Sandeep, and S. Dhanabalan. "Form tolerance analysis and multi-parametric optimization of meso deep square hole EDMed on Inconel-718 plate using GRA method." Grey Systems: Theory and Application 11, no. 4 (2021): 664-680.
31. Gangil, Manish, and M. K. Pradhan. "Optimization of machining parameters of EDM for performance characteristics using RSM and GRA." J. Mech. Eng. Biomech 2, no. 4 (2018): 27-33.