A multi-criteria decision-making approach to analyse the viability of blockchain in software development projects

Abstract

Keywords

1 Introduction

2 Related work and motivation

3.1 SAW

3.3 MABAC

3.4 COPRAS

3.5 TOPSIS

3.6 Oreste

3.7 GRA

3.8 VIKOR

3.9 PROMETHEE

3.10 Normalisation

Table 1
Decision matrix for the proposed system

Q1 Q2 Q3 Q4 Q5 Q6

No Blockchain 40 40 5 0 0 0

Hybrid Blockchain 20 20 40 40 40 30

Private blockchain 30 30 15 50 20 50

Public Blockchain 10 10 40 10 40 20

References

Blockchain technology is getting famous, and use cases of blockchain range from financial services to the Metaverse. It is considered a platform for web 3.0. As a result, many industries are planning to adopt blockchain. A simple public blockchain is not suitable for most business scenarios, so hybrid and private blockchains came into existence, but it is important to decide which type of blockchain should be adopted during the project planning phase. Various models can be found in the literature to determine if blockchain should be adopted and, if so, which type of blockchain should be adopted. However, these models are already becoming obsolete as they determine the usage of blockchain using simple yes or no. In order to overcome these problems, all these models are converted from binary-based selection to fuzzy-based selection, and decision matrices are created. Various multi-criteria decision analysis methods are applied, and final results are obtained. In addition, a novel model is presented, and a MATLAB application is developed to let the user determine if blockchain can be integrated with any technology or not. This application can be used as a standard in the project’s planning phase and helps avoid losses to the industry.

Blockchain decision making distributed ledger SAW TOPSIS

Stuart Haber and W. Scott Stornetta [1] first introduced blockchain or its form to refer to a connected and cryptographically secure chain of blocks. Blockchain ledgers are immutable ‘blocks’ of data shared between numerous users. Bitcoin [2] has been the most high-profile application of blockchain technology. Even more admirable opportunities may arise due to the transformation of commercial transactions and information interchange on the one hand and the elimination of costly layers of verification overhead on the other. Among the potential benefits is creating a single immutable record of data and the elimination of costly and error-prone auditing by humans.

The first generation of blockchain solutions includes cryptocurrencies such as Bitcoin, Litecoin [3], and Zcash [4]. They were (and continue to be) ideal for financial transactions. This makes them limited in nature. These were mainly public blockchains. Ethereum [5], Hyperledger [6], and Corda [7], which could be considered ’generation 2.0’ of blockchain technology, facilitate financial transactions and enable users to establish what is known as Smart Contracts [5]. Applications like supply chain management is a feature of Generation 2.0. This generation includes both hybrid and private blockchains.

Blockchain addresses the issue of storing, transporting, and transacting valuable goods online without the use of any intermediary and to transfer assets freely. For any use case involving more than one party, the conditions for any contract can be hard-coded into the blockchain using a Smart Contract which will be executed only when predefined conditions are met. A risk can be reduced by using a blockchain application created purely for processing contracts, notifying all parties, and processing payments. Additionally, the technology might be used to improve the processing and verification of claims. This can be accomplished by using blockchain application platforms such as Ethereum.

Due to the robust developer community, blockchain technology has gained the security, stability, and standardisation required to disrupt the enterprise sector. Microsoft and other companies are introducing technologies such as the CoCo Framework and Blockchain as a Service (BaaS) to enable users to construct applications more simply than ever imagined feasible and establish trust between corporations. Not only that, now blockchain is being expected to be widely used in gaming and metaverse platforms.

As groundbreaking as blockchain technology is, it cannot be utilised for everything. Most enterprise systems are antagonistic to being public, transparent, or distributed. Additionally, a blockchain-based system takes longer to execute transactions and consumes a high amount of resources compared to typical systems. This makes blockchain not suitable for some systems. Some models are proposed in the literature to decide whether blockchain should be used in any system or which type of blockchain should be used. However, these models do not consider a combination of criteria for decision making.

Inspired by Bitcoin’s widespread acceptance and influence, blockchain technology has gained much more interest from academics and businesses, allowing untrusted individuals to communicate with others in a verifiable manner without the need for a trusted central authority. Software projects cost millions, and lousy technology decisions can lead to money and workforce loss. Choosing the correct technologies is crucial to the software product’s success, making it easier to design and leaving opportunity for future scalability, update, and maintenance. Selecting the type of blockchain is not easy as it requires the decision-making of various parameters. Multi-Criteria Decision-making (MCDM) has the potential to improve all areas of decision-making in engineering, from design to manufacturing. So the motivation of this paper is to use MCDM techniques to help software engineers and designers decide which type of blockchain should be used, which can help the potential losses.

To solve this issue, this paper proposes implementing the widely used models in MATLAB as a graphical user interface that helps answer the questions asked in the model not as a yes or no but as a probability or percentage. Then based on the answers, various MCDM algorithms are being called, and the most suitable choice of blockchain is being presented as a result. This paper also proposes a model and a way to know the potential of any research topic when combined with blockchain, which is also implemented in MATLAB.

In this paper, Section 2 discusses the related work and explains some of the models found in the literature. Experimental setup and explanation of the MCDM algorithms and normalisation techniques are explained in Section 3. It was challenging to show the results of every model, so only the results obtained from the proposed model and explanation of the result are presented in Section 4. Finally, Section 5 concludes the paper.

In Suichies Model [8], the authors suggested abandoning the notion of blockchain as a ready-to-use consumer solution. It is suggested that blockchain is mostly a supporting infrastructure. Furthermore, like with any infrastructure, it should be concealed until it breaks down. Blockchain technology should provide a robust, open, and interoperable standard. It should only appear when anything goes wrong. The authors proposed to consider blockchain solutions in a larger context. While solutions may have unique requirements, utilising blockchain as a foundation does demand the application of some fundamental laws. Its implementation in application can be seen in Fig. 1b.

Fig. 1

Graphical User Interface of all the models along with proposed model.

Birch-Brown-Parulava model [9] is simple and divides the type of blockchain into double permissionless, permissionless, permissioned and double permissioned. It divided the choices in two ways communication choice and consensus choice. With at most four selections in this model, it is easy to know which type of blockchain is suitable for a particular use case.Its implementation in application can be seen in Fig. 1c.

IBM model is intended to assist decision-makers in determining when to adopt blockchain. Additionally, it guides users through the blockchain go-to-market process. This simplified model illustrates the circumstances in which blockchain integration may be a suitable match to be adopted. IBM created Hyperledger Fabric to give private blockchain solutions to industries, so this model focuses mainly on the usage of private blockchains [10]. Its implementation in application can be seen in Fig. 1d.

T. Koens & E. Poll model [11] not just suggests the use of blockchain, but it helps determine which type of technology should be used to store the data. This model can be used to determine if blockchain is necessary from a technological standpoint and can be extended to include non-technical factors that influence blockchain adoption, such as philosophical views and economic incentives. This model is better as it provides suggestions after each selection. This model does not explicitly exclude blockchain technology. Its implementation in application can be seen in Fig. 1e.

Morgen E. Peck model [12] lets the user decide whether to use permissioned or public blockchain. When the categorisation was not done properly, permissioned blockchains included private and hybrid blockchains. In the article, the authors suggested that blockchain should not be used in most scenarios, and if higher transaction speed is needed, then it is better not to use blockchain. Its implementation in application can be seen in Fig. 1f.

J. Gardner model [12] is good, and it does not reject the use of blockchain altogether. It differentiates between using and not using blockchain and categorises the type of blockchain once it is decided to use blockchain. It decides the use of permissioned or public blockchain based on reading and write access control. It also suggests that if a user needs to use a private blockchain, they can use any distributed ledger that can solve a particular use case. So it is more open as compared to other models. Its implementation in application can be seen in Fig. 1g.

Karl Wüst and Arthur Gervais model [13], the authors also present a systematic process for determining the best blockchain solution for a given application. The choices in this model are very clean as it has simple selections and straightforward solutions for a set of selections. The authors also tried to explain the usage of the model, taking the use cases of supply chain management, decentralised autonomous organisations, and domestic and cross-border payments. Its implementation in application can be seen in Fig. 1h.

There are other less famous models like Lixar model, Nandwani model, Verslype model, Best of ICO model, Chand model, Verified ICOs, Mueller model, Pahl model, Lin model [14], world economic forum, Greenspan, Henkel model, Lewis model, Xu model [15], Deloitte model, Maul model [16]. Either these models are discussed as a personal blog, or they are not used much in the literature. Authors in [11] discussed these models in detail.

3 Experimental setup

Multiple-criteria decision-making (MCDM) is a subfield of operations research that examines decision-making under conditions of conflicting criteria in both daily life and industry. Typically, cost is a significant issue, and some measure of quality is a secondary criterion that frequently conflicts with cost. Taking the example of portfolio management, managers seek high returns while minimising risks; nevertheless, high-return equities often involve a significant chance of loss. Similarly, client happiness and service cost are fundamentally at odds in the service sector. The problem should be adequately structured and evaluated using numerous criteria when the stakes are high, for example creating a complex software project, planning is one of the essential parts [17].

Good problem structuring and explicit consideration of numerous factors lead to better judgments. It had come a long way since the early 1960s when the discipline of multiple-criteria decision-making was born [18]. Various techniques and procedures, many aided by decision-making software, have been created for use in fields ranging from politics and business to the environment and energy [19].

As blockchain technology is continuously evolving, and it is considered that it will change how a person uses the software product, especially with the introduction of DaPPS, so it is necessary to know whether blockchain can be implemented in the new system. It will not be straightforward answered as presented in earlier models. Nobody thought that blockchain technology would be adopted by gaming and Metaverse industry. We designed most of the famous models discussed in the literature in MATLAB.

However, instead of simple yes or no, we have an extended option to select the possibility of any choice, i.e., selection can be made in terms of percentages. Selection made by the user will be converted to weights, and a decision matrix will be formed based on the model. After that, various MCDM algorithms are being called with different normalisation techniques to obtain the results of all those algorithms. Finally, the result of all those algorithms is based on making the final decision. These algorithms and normalisation techniques are discussed below.

In the decision theory, simple additive weighting (SAW) is the most widely used and most straightforward approach for assessing a set of choices against a set of decision criteria. SAW is a multi-attribute technique based on the notion of weighted summation. The system will calculate a weighted total of each option’s performance on all alternative criteria. The alternative with the greatest score will be considered the best and will be recommended [20].

As a starting point, consider that a particular MCDM challenge has y choices and x decision criteria which all are benefit criteria, i.e., as the value increases, the choice becomes better. Assume that m_b signifies the criterion C_b’s relative weight of importance and that p_ab denotes the performance value of alternative O_a when assessed against criterion C_b. Then, the overall (i.e., when all criteria are assessed concurrently) relevance of alternative P_a is defined as follows: $P_{a}^{SAW - score} = \sum_{b = 1}^{x} m_{b} p_{ab}, for a = 1, 2, 3, 4 \dots y$ (1)

The optimal option in the maximising situation is the one that results in the highest overall performance value.

3.2 CODAS

Two metrics are used in this technique to ascertain the desirability of options. The primary and most important metric is the Euclidean distance between alternatives and the negative ideal. Using this form of distance needs a criterion space with an L²-norm indifference. The taxicab distance, which is connected to the L¹-norm indifference space, is used as a additional metric. It is self-evident that the option further away from the negative-ideal solution is preferable. When two options are incomparable in Euclidean distance, this technique uses the Taxicab distance as a supplementary metric. While the CODAS prefers the L²-norm indifference space, it may evaluate two other varieties [21].

MABAC [22] is a relatively new method for MCDM difficulties. The MABAC method calculates the distance between each alternative as well as the border approximation area (BAA). It has several unique characteristics. In this the computing results are stable, the calculating equations are simple, it accounts for latent gains and losses and it can be combined with other approaches. The MABAC method’s fundamental premise is represented in describing the alternative’s distance from the approximate boundary domain. The approximate area of the border indicates the mean value for all choices. If the alternative is more significant than that value, it will have a positive value and vice versa.

Zavadskas et al. proposed the COmplex PRoportional ASsessment (COPRAS) approach as a novel way of MCDM [23]. It is used to evaluate maximising and reducing criteria values using multiple criteria. The COPRAS approach is frequently utilised by its inventors, followers, and professionals for quantitative multi-criteria evaluation of complex systems. The COPRAS approach yields a solution and ratios to the ideal and worst-ideal solutions, so it may be considered a compromise method.

The Technique for Order of Preference by Similarity to the Ideal Solution (TOPSIS) was created by Ching-Lai Hwang and Yoon [24], with additional refinements by Yoon [25], Hwang, Lai, and Liu [26]. It is a type of compensatory aggregation in which a group of options is evaluated by assigning weights to each criterion then normalising the scores for each criterion and finally determining the geometric distance between the ideal choice and each alternative with the best score in each criterion. TOPSIS presupposes that perhaps the criteria are increasing or decreasing monotonously. Normalisation is commonly required because the dimensions of the parameters or criteria are frequently incompatible [26, 27].

Compensatory techniques, such as TOPSIS, allow for trade-offs amongst criteria, where a high score on one might compensate for a poor rating on another. This modelling technique is more realistic than non-compensatory approaches, which incorporate or exclude alternative solutions based on tight cut-off values. Numerous distance metrics are applicable. Traditionally, TOPSIS calculated ideal and nadir solutions using the Euclidean norm (minimisation of the square root of the sum of squared distances), a second power metric (E_u). TOPSIS2 is a version that expresses distance in terms of the least absolute value, a first power metric (S_c). Another often-used measure is the Tchebychev metric, which is based on the minimum-maximum difference. This is equivalent to an infinite power term (P_Inf) [28].

In recent research, two new models for TOPSIS: the complex Pythagorean fuzzy TOPSIS technique and the complex Pythagorean fuzzy ELECTRE I approach are presented [29]. These strategies facilitate the resolution of complex group decision-making issues involving multiple criteria and complex Pythagorean fuzzy data. Similarly, the authors in [30] presented strategies to solve issues with the typical crisp risk priority numbers and the fuzzy failure modes and effects analysis techniques used in risk ranking. An innovative solution is provided where authors combine the properties of complex spherical fuzzy sets with the potential of the TOPSIS method [31].

Oreste was given as a more generic technique applicable even in the absence of quantitative data and does not need the determination of criterion weights. Roubens first suggested the Oreste (organisation, rangement, and synthese de données relationnelles) technique maximising quality in operations research, and it has been widely employed in that sector since then. The approach was described in a tutorial article [32] and shown through a case study using an automobile and a computer selecting system. The Oreste technique is divided into two distinct components. The first section establishes a chronological arrangement for the tests. Using preference relations, the experiments are ranked in order of significance. In the second half of Oreste, an incomplete preference structure is formed, e.g., by incorporating indifference (I) and incomparability (R) thresholds, therefore weakening some aspects of the complete order produced in the first half. If experiment a is somewhat superior than experiment b on certain parameters and vice versa on others, they are indistinguishable. In this instance, one cannot view one experiment to be generally superior to another via a selection method unless one weighs the criteria against which the trials perform significantly differently. Both experiments are considered as equally valid at this stage. On the other hand, if a is statistically superior to b on certain criteria but statistically inferior on others, a contradictory scenario develops. The studies are too dissimilar to be compared. This is referred to as incomparability [37].

Julong Deng of Huazhong University of Science and Technology developed grey relational analysis (GRA). It is a popular model in grey system theory. GRA employs a specific type of data. It categorises situations as black or white depending on the amount of information. However, in reality, neither of these events happen. In fact, circumstances containing just incomplete knowledge are described as grey, foggy, or fuzzy. In engineering, the Taguchi-based GRA model is prevalent. Deng Julong introduced GRA in 1982 as part of grey system theory [34]. A grey system is one in which part of the information is known, and part is uncertain. Grey numbers are interval-valued unknowns in grey systems theory, with the width of the interval indicating knowledge precision [35] and using this concept, information quantity and quality range from zero to one hundred per cent from black to white. Because ambiguity exists, one is constantly in the midst, between the extremes, in the grey region. Grey analysis then provides explicit comments concerning system solutions. A system with no information has no solution. A system with complete information has one solution. Grey systems provide a range of options in the midway. However, it provides ways for selecting a suitable solution, an appropriate answer for real-world issues. Like Harvard Business School professor Joseph L. Badaracco, many notable researchers and business executives were motivated by the notion.

The VIKOR method is one of the most famous MCDM strategy. Assuming that compromise is appropriate for conflict resolution, the decision-maker wishes the closest solution to the ideal, and the options are evaluated using all defined criteria, it was devised by Serafim Opricovic. VIKOR compares possibilities and finds the best compromise. Po-Lung Yu and Milan Zeleny pioneered the concept of a compromise solution in MCDM in 1973 [36, 37]. S. Opricovic devised the fundamental concepts of VIKOR in his 1979 PhD dissertation, and a patent was issued in 1980 [38]. VIKOR was coined in 1990 [39] from the Serbian term “ViseKriterijumska Optimizacija I Kompromisno Resenje,” which translates as “Multicriteria Optimization and Compromise Solution.” The first legitimate applications occurred in 1998 [40]. The 2004 publication had a crucial role in the broad adoption of the VIKOR method [41].

The PROMETHEE and GAIA [42] approaches are more commonly referred to as the “Preference Ranking Organisation METHod for Enriching Evaluations” and its descriptive complement “Geometrical Analysis for Interactive Assistance” GAIA [43]. The PROMETHEE method was developed by Professor Jean-Pierre Brans in 1982 [44]. It was further developed and implemented by Professor Jean-Pierre Brans and Professor Bertrand Mareschal, including enhancements such as GAIA.

PROMETHEE [45] is a prescriptive technique that gives the decision-maker both an entire and partial ranking of activities. PROMETHEE has been effectively employed in various decision-making scenarios throughout the world. In 2010, a non-exhaustive list of scientific articles about PROMETHEE method extensions, applications, and discussions was published [46].

The GAIA and PROMETHEE technique does not prescribe a "correct" course of action, but rather aids decision-makers in choosing the alternative that best matches their aim and understanding of the issue. An extensive and rational methodology for generating a choice problem, identifying and measuring its action clusters, conflicts and synergies, and highlighting the key alternatives and their structural reasoning is provided.

There are numerous definitions for data normalisation, which vary according to the subject of study. When it comes to database normalisation, for example, data attributes are organised into tables so that data management consistency and efficiency can be improved. The most frequently used term in statistics and its applications is converting values measured on different scales to a similar scale, frequently prior to aggregating or averaging them. The sum method divides the sum of the ratings for each alternative by the sum for all alternatives.

For sum normalisation, the formula to normalise benefit criterion and cost criterion is given below, respectively: $s_{ij} = \frac{t_{ij}}{\sum_{i = 1}^{m} t_{ij}}, s_{ij} = \frac{1 / t_{ij}}{\sum_{i = 1}^{m} 1 / t_{ij}}$ (2) These ratings are divided by the maximum possible performance rating for each attribute. The normalised value t_ij for benefit qualities is calculated as follows: $t_{ij} = \frac{y_{ij}}{y_{j}^{max}}$ (3)

For cost attributes, t_ij is computed as $t_{ij} = 1 - \frac{y_{ij}}{y_{j}^{max}}$ (4)

Min-max normalisation is one of the most frequently used methods of data normalising. This function calculates characteristics’ maximum and minimum performance ratings. For each feature, the minimum value is converted to a 0, the maximum value is converted to a 1, and all other values are converted to a decimal between 0 and 1. The benefit and cost criteria are given below: $s_{ij} = \frac{t_{ij} - t_{min}}{t_{max} - t_{min}}, s_{ij} = \frac{t_{max} - t_{ij}}{t_{max} - t_{min}}$ (5)

Each performance rating in the decision matrix is divided by its norm during vector normalisation. $t_{ij} = \frac{x_{ij}}{\sqrt{\sum_{i = 1}^{m} y_{ij}^{2}}}, t_{ij} = \frac{(1 / y_{ij})}{\sqrt{\sum_{i = 1}^{m} (1 / y_{ij}^{2})}}$ (6) where y_ij denotes the performance rating of the ith alternative for attribute $C_{j}, y_{j}^{max}$ denotes the highest performance rating for attribute C_j, and $y_{j}^{min}$ denotes the minimum performance rating for attribute C_j [47].

4 Results and discussions

All the discussed blockchain models are implemented in MATLAB, and all the discussed algorithms are executed to get the suggestion about using blockchain. We also proposed a model, as can be seen in Fig. 1i.

We included six questions whose answers cannot be simple yes or no. However, it covers the expectation for future use as well. The first question is about how much the system is closed, and the answer can be selected in terms of percentage. This question helps to know the transparency that the project owner wants to have for the general public. The companies like Facebook do not reveal anything about the code and what they are tracking, and on the other hand, projects built on Ethereum are fully open. So by this question project owner can predict how much he can make the project open.

The second question is about the known users. If the project users are completely unknown, then to make trust between them, either a trusted third party is required, or the only other method available is blockchain. So this is one of the most important decision criteria. The third question asks about the number of parties involved other than users. This question helps to know whether, in a project, more than one party is involved. Taking the example of supply chain management, there are at least three parties involved, i.e. producer, supplier and customer. So when multiple parties are involved, there is a need to maintain trust between these parties, which can be done using smart contracts of blockchain. So this is also an important decision criterion for the use of blockchain and the type of blockchain.

The next question is about the in house consensus. In any business scenario or transaction, the consensus among the parties is critical. For private projects, consensus can be reached by signing the transactions, and an endorsement policy can be set based on the use case. An example of endorsement policy is when more than half the people agree, and consensus is reached. Similarly, for public projects, consensus will be more difficult. The consensus is reached using a proof of work consensus mechanism, taking the example of Bitcoin. So this question is also essential for deciding the type of blockchain.

The next question is about smart contracts. This question is added so that any possibility of using blockchain is not missed. If any project needs to use a smart contract, the only way is blockchain. Finally, the last question is about auditing. Even if any project does not need blockchain for all the other features of blockchain, but they can still use it for auditing. Take the example of the toxic water or air released by a factory if the PH level and AQI of the output is stored in the blockchain, then it will be easy to audit the factory as blockchain stores data in an immutable ledger, so this makes the auditing authorities believe in the correctness of data.

The weight for MCDM algorithms will be decided by the selection made by the user, and we set the decision matrix for this system as shown in Table 1. The result of running the application will be in the form of suggestions from best to worst as well as the output of each algorithm. Now according to questions, values are assigned, i.e., for which question which type of blockchain is most preferable. For the first question, if the system is not transparent, that means the most appropriate choice is not using blockchain, but that does not rule out the other possibilities like using private or hybrid blockchain.

Similarly, for the second question most preferable choice is not to use blockchain. Now moving to the third question, which asks for the number of parties, now as the number of parties increases, then it means we will be moving towards either public or hybrid blockchain. A large number of participants can also be present in private blockchains, so it also has a significant value. The fourth question is about in house consensus. Now this question is irrelevant if we don’t want blockchain, but it is important to decide between public, private and hybrid blockchains. As the number of in house participants for consensus increase chances of moving towards private blockchain increase. Question five is related to smart contracts which is again not relavent if don’t need blockchain but for both public and hybrid blockchain it is one of the most important question. Last question motivates the user to use private blockchain but it does not rule out the logging in public or hybrid network.

The selections are made as shown in Fig. 1i and based on that, and the results are obtained. These results are shown in Table 2. For each algorithm, results are obtained using all the discussed normalisation methods. For Orestre and PROMETHEE, results are shown in Fig. 2 and Table 3, respectively. For each normalisation method, the third column, i.e. Ai, represents the order of choices where 1 means no blockchain, 2 means hybrid, 3 means private and 4 means public blockchain. Results of all these methods are combined to get the most appropriate sequence, which is displayed as the final result in the application window.

Table 2

Results obtained for different MCDM mehtods and different normalization techniques

MAX			SUM			VEC			MAX-MIN			Dea
	Q	iQ	Ai	Q	iQ	Ai	Q	iQ	Ai	Q	iQ	Ai	Q	iQ	Ai
SAW	0.774	33.46	3	0.347	34.74	3	0.587	34.28	3	0.73	35.06	3	0.849	28.3	3
	0.657	28.42	2	0.287	28.7	2	0.485	28.3	2	0.588	28.23	2	0.794	26.47	2
	0.452	19.54	4	0.194	19.37	4	0.328	19.14	4	0.417	20.02	1	0.682	22.73	1
	0.43	18.58	1	0.172	17.18	1	0.313	18.28	1	0.348	16.7	4	0.675	22.5	4
CODAS	1.175	63.73	3	0.609	63.67	3	0.98	64.92	3	1.207	54.1	3	0.557	56.84	3
	0.303	34.99	2	0.152	34.63	2	0.215	33.76	2	0.19	29.59	2	0.044	27.54	2
	-0.719	1.28	1	-0.367	1.7	4	-0.581	1.32	1	-0.36	16.31	1	-0.164	15.62	1
	-0.758	0	4	-0.394	0	1	-0.614	0	4	-1.037	0	4	-0.437	0	4
MABAC	0.235	57.95	3	0.107	56.15	3	0.183	59.55	3	0.261	55.29	3	0.112	58.04	3
	0.119	38.31	2	0.047	36.83	2	0.08	37.26	2	0.118	34.72	2	0.056	39.69	2
	-0.087	3.74	4	-0.047	7.01	4	-0.076	3.19	4	-0.052	10	1	-0.056	2.28	1
	-0.109	0	1	-0.069	0	1	-0.091	0	1	-0.122	0	4	-0.063	0	4
COPRAS	0.774	33.46	3	0.347	34.74	3	0.587	34.28	3	0.73	35.06	3	0.849	28.3	3
	0.657	28.42	2	0.287	28.7	2	0.485	28.3	2	0.588	28.23	2	0.794	26.47	2
	0.452	19.54	4	0.194	19.37	4	0.328	19.14	4	0.417	20.02	1	0.682	22.73	1
	0.43	18.58	1	0.172	17.18	1	0.313	18.28	1	0.348	16.7	4	0.675	22.5	4
TOPSIS(Sc)	0.744	35.61	3	0.769	36.99	3	0.76	36.58	3	0.73	35.06	3	0.713	33.96	3
	0.612	29.29	2	0.614	29.55	2	0.605	29.11	2	0.588	28.23	2	0.609	28.98	2
	0.379	18.15	4	0.376	18.08	4	0.368	17.69	4	0.417	20.02	1	0.396	18.84	1
	0.354	16.95	1	0.32	15.38	1	0.345	16.62	1	0.348	16.7	4	0.383	18.22	4
TOPSIS(Eu)	0.756	35.86	3	0.788	37.47	3	0.778	37.05	3	0.734	35.24	3	0.714	33.86	3
	0.565	26.81	2	0.574	27.3	2	0.563	26.81	2	0.531	25.47	2	0.553	26.24	2
	0.394	18.71	4	0.393	18.69	4	0.384	18.29	4	0.46	22.09	1	0.447	21.18	1
	0.392	18.62	1	0.348	16.54	1	0.375	17.84	1	0.358	17.2	4	0.395	18.73	4
TOPSIS(Inf)	0.827	37.68	3	0.857	39.39	3	0.851	38.75	3	0.794	37.68	3	0.782	36.44	3
	0.6	27.32	2	0.6	27.58	2	0.6	27.33	2	0.536	25.44	2	0.536	24.96	2
	0.4	18.21	4	0.4	18.39	4	0.4	18.22	4	0.438	20.78	1	0.438	20.39	1
	0.369	16.79	1	0.318	14.64	1	0.345	15.7	1	0.339	16.1	4	0.391	18.21	4
GRA(T)	1.635	36.66	3	1.564	35.78	3	1.604	36.36	3	1.688	37.34	3	1.55	35.19	3
	1.238	27.77	2	1.203	27.52	2	1.214	27.53	2	1.256	27.77	2	1.223	27.77	2
	0.819	18.36	1	0.812	18.58	1	0.821	18.6	1	0.846	18.71	1	0.852	19.35	1
	0.767	17.21	4	0.792	18.13	4	0.772	17.51	4	0.731	16.18	4	0.779	17.68	4

Fig. 2

Results for ORESTE and Intensivity of performance indicator.

Table 3

Results obtained for PROMETHEE method

	V-Shape			Linear			Gaus			Linear:Gaus
	Q	iQ	Ai	Q	iQ	Ai	Q	iQ	Ai	Q	iQ	Ai
	1.279	53.03	3	1.159	61.83	3	0.996	55.54	3	0.959	53.62	3
	0.327	32.17	2	0.133	29.23	2	0.265	33.12	2	0.287	33.57	2
PROMETHEE	-0.465	14.81	1	-0.505	8.94	1	-0.446	11.34	1	-0.408	12.81	1
	-1.141	0	4	-0.787	0	4	-0.816	0	4	-0.837	0	4

As can be seen from Table 2 alternative three, i.e. private blockchain, has the first rank for MAX-method of normalization of SAW method. The performance indicator, i.e. Q, is 0.774, and the contribution rate, i.e. iQ is 33.45%. Similarly, for SAW we checked for different types of normalization techniques. SAW is one of the MCDM techniques, and the application doesn’t just predict results based on one algorithm or one normalization technique of the algorithm. In the application, results are found for SAW, CODAS, MABAC, COPRAS, TOPSIS(Eu) and, TOPSIS(Inf), GRA(T) using different normalization techniques, i.e. MAX, SUM, VEC, MAX-MIN and dea.

The PROMETHEE method implementation requires preference functions. The PROMETHEE preference function is a function used to define deviations between possibilities for each criterion. So in the application, our results for PROMETHEE are not based on the normalization function but on preference functions. These results are shown separately in Table 3. The final results also consider the results obtained from this method.

In Oreste, evaluation findings are based on sensitivity analysis, and the values of uncertainty parameters have a direct impact on the evaluation outcomes. The values of uncertainty parameters are adjusted based on the application. In practice, uncertainty parameters may be computed based on the range of indicator values and the data quality. Multiple experiments are performed to capture more indicator data in order to establish uncertainty parameters with greater reliability. A sensitivity analysis is used to determine the final value based on variations in the results. These results are shown in Fig. 2. Based on all the results obtained from all the techniques final result is evaluated. These results are in the form of preference for the type of blockchain. These results are presented in the application itself, as can be seen in Fig. 1i.

Like the proposed model, all the discussed models are developed, and results can be obtained from these models. For research purposes, we added another option to know any application’s potential, as shown in Fig. 1a. Here user can give a research topic, and based on the literature found in the Scopus database, it calculates the percentage of the topic with blockchain and finds its percentage concerning all the papers of blockchain. By this, users can get an idea of how much a topic is studied, and the given topic has some potential or not.

5 Conclusion

This paper explains the significance of the exponentially growing popularity of blockchain technology. Moreover, it was explained why blockchain might be unsuitable for some applications. Various models to determine whether blockchain should be used or not are explained next. All the models found in the literature are converted from simple yes-no type questions to probability-based questions. A new model is proposed, which consists of six questions in which users can answer any question based on the probability. These probabilities are used to decide the weight of the criterion for MCDM algorithms. A decision matrix is created for all the models, including the proposed model. All these questions are explained, and finally, the results of running MCDM algorithms based on the selected weights and decision matrix is presented in the form of a table and for ORESTRE in the form of graphs. Blockchain is changing rapidly, and it has very diverse adoptions from financial services to NFTs to Metaverse. Finally, our scheme serves as a practical guide for blockchain projects seeking to choose the most appropriate technology for a given circumstance.

No Blockchain

Hybrid Blockchain

Private blockchain

Public Blockchain

Stuart Haber , Stornetta Scott , How to timestamp a digitaldocument, Journal of Cryptology 3(2) (1991), 99–111.

Satoshi Nakamoto , Bitcoin: A Peer-to-Peer Electronic Cash System | Satoshi Nakamoto Institute, Technical report, bitoin.org (2008).

Elliott Branson , Litecoin: The ultimate beginnerś guide for understanding litecoins and what you need to know (2014).

Eli Ben Sasson , Alessandro Chiesa , Christina Garman , Matthew Ian Miers , et al., Zerocash: Decentralized anonymous payments from bitcoin, In 2014 IEEE Symposium on Security and Privacy, IEEE (2014), 459–474.

Gavin , et al., Wood, Ethereum: A secure decentralised generalisedtransaction ledger, Ethereum Project Yellow Paper 151(2014), 1–32.

Elli Androulaki , Artem Barger , Vita Bortnikov , Christian Cachin , Christidis , et al., Hyperledger Fabric: A Distributed Operating System for Permissioned Blockchains, In Proceedings of the Thirteenth EuroSys Conference, EuroSys ’18, page 15, New York, NY, USA, 2018. Association for Computing Machinery.

Mike Hearn , Corda: A distributed ledger, Whitepaper (2016), 1–73.

Bart Suichies , Why Blockchain must die in Bart SuichiesMedium (2016).

David Birch , Richard Brown , Salome Parulava , Towards ambientaccountability in financial services: Shared ledgers, translucenttransactions and the technological legacy ofthe great financial crisis, Journal of Payments Strategy &Systems 10(2) (2016), 118–131.

10.

Kathryn Harrison, , Top 5 questions for choosing a blockchain technology IBM Supply Chain and Blockchain Blog (2018).

11.

Tommy Koens , Erik Poll , What blockchain alternative do you need? In Joaquin Garcia-Alfaro, Jordi Herrera-Joancomartí,, Giovanni Livraga, and Ruben Rios, editors, Data Privacy Management, Cryptocurrencies and Blockchain Technology, Cham, Springer International Publishing (2018), 113–129.

12.

Morgen Peck

, Blockchain world - Do you need a blockchain? Thischart will tell you if the technology can solve your problem, IEEE Spectrum 54(10) (2017), 38–60.

13.

Karl Wüst , Arthur Gervais , Do you need a Blockchain? IACR Cryptology ePrint Archive (i):375 (2017).

14.

Yu-Pin Lin , Joy R Petway , Johnathen Anthony , Hussnain Mukhtar , Liao , et al. Blockchain:, The Evolutionary Next Stefor ICTE-Agriculture, Environments 4(3) (2017).

15.

Xiwei Xu , Ingo Weber , Liming Zhu , Mark Staples , Jan Bosch , et al.A Taxonomy of Blockchain-based Systems for Architecture Design, In 1st IEEE International Conference on Software Architecture (ICSA 2017), Gothenburg, Sweden, Apr 2017. IEEE, IEEE.

16.

Roger Maull , Phil Godsiff , Catherine Mulligan , Alan Brown , BethKewell , Distributed ledger technology: Applications andimplications, Strategic Change 26(5) (2017), 481–489.

17.

Irik Mukhametzyanov , Dragan Pamucar , A sensitivity analysis inmcdm problems: A statistical approach, Decision Making:Applications in Management and Engineering 2 (2018), 1–20.

18.

Roland Weistroffer

, Yan Li , Multiple criteria decisionanalysis software, International Series in Operations Researchand Management Science 233 (2016), 1301–1341.

19.

Angeliki Kylili , Elias Christoforou , Fokaides Paris

, Polycarpos Polycarpou , Multicriteria analysis for the selection of the most appropriate energy crops: the case of Cyprus, International Journal of Sustainable Energy 35(1) (2016), 47–58.

20.

West Churchman

and Russell Ackoff

, An Approximate Measure ofValue, Journal of the Operations Research Society of America 2(2) (1954), 172–187.

21.

Jurgita Antucheviciene , Mehdi Keshavarz Ghorabaee , EdmundasKazimieras Zavadskas , Zenonas Turskis , A New CombinativeDistance-based Assessment (Codas) Method For Multi-criteriaDecision-making, Economic Computation and Economic CyberneticsStudies and Research 50(1) (2016), 39–68.

22.

Alireza Alinezhad , Javad Khalili , Mabac method, In New Methods and Applications in Multiple Attribute Decision Making (MADM), Springer (2019), 193–198.

23.

Dr. Ginevičius Habil

Kaklauskas

and Pastat

, U sistemotechninis Iverunimas. v.: Technika 275, Statyba 4(1) (1998), 91–91.

24.

Kwangsun Yoon , Ching Lai Hwang , Multiple attribute decision making : an introduction (1995), 75.

25.

Kwangsun Yoon , A reconciliation among discrete compromisesolutions, Journal of the Operational Research Society 38(3) (1987), 277–286.

26.

Ching Lai Hwang , Young Jou Lai , Ting-Yun Liu , A new approach formultiple objective decision making, Computers & OperationsResearch 20(8) (1993), 889–899.

27.

Edmundas Kazimieras Zavadskas , Algimantas Zakarevicius , JurgitaAntucheviciene , Evaluation of ranking accuracy in multi-criteriadecisions, Informatica 17(4) (2006), 601–618.

28.

Olson

D.L.

, Comparison of weights in TOPSIS models, Mathematical and Computer Modelling 40(7-8) (2004), 721–727.

29.

Akram

, Garg

and Zahid

, Extensions of electre-i and topsismethods for group decision-making under complex pythagorean fuzzyenvironment, Iranian Journal of Fuzzy Systems 17(5) (2020), 147–164.

30.

Muhammad Akram , Anam Luqman and Jose Carlos

, Alcantud, Riskevaluation in failure modes and effects analysis: Hybrid topsis andelectre i solutions with pythagorean fuzzy information, NeuralComput. Appl. 33(11) (2021), 5675–5703.

31.

Muhammad Akram , Cengiz Kahraman , Kiran Zahid , Extension of topsismodel to the decision-making under complex spherical fuzzyinformation, Soft Comput. 25(16) (2021), 10771–10795.

32.

Peter Rousseeuw , Estimation and testing by means of optimallyrobust statistics,rationnelle, Revue Belge de Statistique, d’Informatiqueet de Recherche 22(3) (1980).

33.

Bourguignon

and Massart

D.L.

, The Oreste method formulticriteria decision making in experimental chemistry, Chemometrics and Intelligent Laboratory Systems 22(2) (1994), 241–256.

34.

Deng Ju-Long , Control problems of grey systems, Systems &Control Letters 1(5) (1982), 288–294.

35.

Sifeng Liu , Yingjie Yang , Naiming Xie , Jeffrey Forrest , Newprogress of Grey System Theory in the new millennium, GreySystems: Theory and Application 6(1) (2016), 2–31.

36.

P.L.

, A class of solutions for group decision problems, Management Science 19(8) (1973), 936–946.

37.

Milan Zelany , A concept of compromise solutions and the method ofthe displaced ideal, Computers & Operations Research 1(3) (1974), 479–496.

38.

Lucien Duckstein , Serafim Opricovic , Multiobjective optimizationin river basin development, Water Resources Research 16(1) (1980), 14–20.

39.

Opricovic

Programski paket VIKOR za visekriterijumsko kompromisno rangiranje, In 17th International Symposium on Operational Research SYM-OP-IS (1990).

40.

Serafim Opricovic , Multicriteria optimization of civil engineeringsystems, Faculty of Civil Engineering, Belgrade 2(1) (1998), 5–21.

41.

Serafim Opricovic , Gwo-Hshiung Tzeng , The Compromise solution byMCDM methods: A comparative analysis of VIKOR and TOPSIS, European Journal of Operational Research 156(2) (2004).

42.

José Figueira , Salvatore Greco , Matthias Ehrgott , Multiple criteria decision analysis: state of the art surveys (2005).

43.

Bertrand Mareschal , Jean-Pierre Brans , Geometricalrepresentations for MCDA, European Journal of Operational Research 34(1) (1988), 69–77.

44.

Jean-Pierre Brans , de la décision:l’élaboration d’instruments d’aide a la décision,Université Laval, Faculté des sciences del’administration, L’ing (1982).

45.

Jean Pierre Vincke , Ph Brans , A preference ranking organizationmethod. The PROMETHEE method for MCDM, Management Science 31(6) (1985), 647–656.

46.

Majid Behzadian, , Reza Baradaran Kazemzadeh , Amir Albadvi , Mohammad Aghdasi , PROMETHEE: A comprehensive literature review onmethodologies and applications, European Journal ofOperational Research 200(1) (2010), 198–215.

47.

Aydin Çelen , Comparative analysis of normalization proceduresin TOPSIS method: With an application to Turkish deposit bankingmarket, Informatica (Netherlands) 25(2) (2014), 185–208.