Parameter reduction analysis under interval-valued m-polar fuzzy soft information

This paper formalizes a novel model that is able to use both interval representations, parameterizations, partial memberships and multi-polarity. These are differing modalities of uncertain knowledge that are supported by many models in the literature. The new structure that embraces all these features simultaneously is called interval-valued multi-polar fuzzy soft set (IVmFSS, for short). An enhanced combination of interval-valued m-polar fuzzy (IVmF) sets and soft sets produces this model. As such, the theory of IVmFSSs constitutes both an interval-valued multipolar-fuzzy generalization of soft set theory; a multipolar generalization of interval-valued fuzzy soft set theory; and an interval-valued generalization of multi-polar fuzzy set theory. Some fundamental operations for IVmFSSs, including intersection, union, complement, “OR”, “AND”, are explored and investigated through examples. An algorithm is developed to solve decision-making problems having data in interval-valued m-polar fuzzy soft form. It is applied to two numerical examples. In addition, three parameter reduction approaches and their algorithmic formulation are proposed for IVmFSSs. They are respectively called parameter reduction based on optimal choice, rank based parameter reduction, and normal parameter reduction. Moreover, these outcomes are compared with existing interval-valued fuzzy methods; relatedly, a comparative analysis among reduction approaches is investigated. Two real case studies for the selection of best site for an airport construction and best rotavator are studied.


Introduction
Interval representations, parameterizations, partial memberships and multi-polarity are various modalities of uncertain knowledge. They have been combined in a myriad of forms in the literature (Akram et al. 2018;Alcantud et al. 2020b;Atanassov 1986;Chen et al. 2014;Jiang et al. 2010;Maji et al. 2001;Molodtsov 1999;Roy and Maji 2007;Yang et al. 2009;Zadeh 1965). The main purpose of this paper is the formalization of a model that takes all these features into account. We call it interval-valued multi-polar fuzzy soft set, also interval-valued m-polar fuzzy soft set or IVmFSS for short. Then we prove its versality with several theoretical and applied developments, inclusive of fundamental operations, parameter reductions, and applications to decision-making. Soft set theory (Molodtsov 1999) was designed to overcome the lack of parameterization tools in traditional uncertainty theories, including probability theory (Varadhan 2001), fuzzy set theory (Zadeh 1965), intuitionistic fuzzy set theory (Atanassov 1986), and rough set theory (Pawlak 1982). However, the theory of soft sets is not a generalization of previous mathematical theories: the concept behind the production of soft sets is strikingly different from classical models for handling uncertainties. Real and potential applications of soft set theory have been reported in different domains such as game theory, decisionmaking, measurement theory, medical diagnosis, etc.
The inception of soft sets inspired numerous researchers who were attracted by this powerful idea, for example, for the purpose of studying its fundamentals, hybrid structures or its interactions with other disciplines. In this way basic properties of soft set theory were presented by Maji et al. (2003). Ali et al. (2009) argued that some features of soft sets discussed in (Maji et al. 2003) were not true in general. They thus produced some novel formal properties of soft sets and verified the applicability of De Morgan's laws for soft sets. Maji and Roy (2002) solved a decision-making problem based upon soft sets. In their analysis they realized that in a given problem whose structure pertains to soft set theory, there might exist one or more parameters that have no effect on the optimal decision. Such parameters may be safely removed from the dataset in order to reduce the simulation cost. Accordingly they took advantage of the idea of attribute reduction in rough set theory ( Pawlak 1982) in order to launch the investigation of parameter reduction in soft set theory. Several additional investigations on the reduction of parameters in soft sets have been reported. For instance, Chen et al. (2005) and Kong et al. (2008) launched parameterization and normal parameter reductions for soft sets, respectively, to overcome the deficiencies of decision-making applications in (Maji and Roy 2002). Ma et al. (2011) developed a novel efficient normal parameter reduction method in order to further improve (Chen et al. 2005;Maji and Roy 2002;Pawlak and Skowron 2007). Following their strategy, Ali Ali (2012) introduced an alternative approach for the reduction of parameters in soft sets. Now a practical distinction is in order. In normal parameter reduction approaches, the reduction set in a given problem can be reused when the expert needs to add some new parameters. However this is not the case of parameter reduction based on optimal choice, as it relies on the preservation of the the decision object. The application scope of parameter reduction based on optimal choice is wider than the former case, but this comes at the cost of a natural inconvenience: when new parameters are added to the set of parameters, an altogether new reduction process is needed, because the optimal object may have changed. In short, the reduction set cannot be reused in the case of parameter reduction based on optimal choice when new parameters are added to the required set of parameters. Clearly we find the same difficulty in rank based parameter reduction, for it keeps the ranking order of all objects.
Soft set theory developed beyond the original formulation soon. Roy and Maji Roy and Maji (2007) proposed a more general model, namely, fuzzy soft sets, which allows to model and solve a different sort of decision-making cases. Alcantud et al. (2017) presented an alternative approach to solve decision-making problems based on fuzzy soft sets. The idea of attributes reduction has also received considerable attention by many researchers in the analyses of fuzzy soft sets and its extensions. For example, Feng et al. (2010b) proposed the idea of parameter reduction of fuzzy soft sets based on concept of level soft sets. With the help of level soft sets in an intuitionistic fuzzy environment, Jiang et al. (2011) developed a reduction approach for the intuitionistic fuzzy soft set model that had been introduced by Maji et al. (2001), Maji et al. (2004). The combination of the soft set and Motivated by all these factors, the aim of this research article is to produce a framework to deal with IVmF soft information involving vagueness and imprecision in a parameterized setting. We introduce a novel hybrid model that combines the advantages of both IVmF sets (Mahapatra et al. 2020) and soft sets (Molodtsov 1999), and we call it IVmFSS. Actually, IVmFSS theory developed in this research article can be viewed both as a multipolar generalization of the IVFSS model (Yang et al. 2009), as an IVmF generalization of the soft set model (Molodtsov 1999), or as an interval-valued generalization of the mF soft set model ( Akram et al. 2018). The reason is that the IVmFSS model here proposed is based on all mF soft sets, IVmF sets, and IVFSSs. Further, three parameter reduction approaches for this novel hybrid model are presented along with their algorithms. Moreover, two real case studies (this is, an airport site selection and selection of rotavator) are explored. Finally, a comparison of the proposed model with some existing intervalvalued fuzzy methods is discussed, and a comparative analysis among the reduction approaches is also provided.
The structure of this research article is as follows. Section 2 first recalls some definitions like soft sets, mF sets, mF soft sets and IVmF sets. Then it presents the concept of IVmFSS and briefly discusses the basic operations in their context. Numerical examples help to understand all the new concepts. Section 3 studies the three parameter reduction approaches for the new hybrid model. Section 4 explores two real case studies of an airport site selection and rotavator selection and it contains a solution of these critical problems by the methodology of IVmFSSs and their parameter reduction techniques. Section 5 discusses a comparative analysis among the reduction approaches, and it also investigates a comparison of the proposed model with some existing interval-valued fuzzy methods. Finally in Sect. 6, we provide the conclusions and future research directions.

Interval-valued mF soft sets
This section firstly recalls some basic notions briefly, including soft set and IVmF sets which are very helpful in the remaining study of the paper. Secondly, IVmFSS model is presented with its basic properties and decision-making mechanism.
Definition 1 (Molodtsov 1999) Let U be an initial universal set and Z be a set of parameters related to the objects of the universe. Let C Z then a soft set ðw; CÞ is defined as below: ðw; CÞ ¼ fðc; wðcÞÞjc 2 C; w 2 PðUÞg; where w : C ! PðUÞ is a mapping where P(U) denotes the collection of all subsets of U.
Definition 2 (Chen et al. 2014) A function A : U ! ½0; 1 m is called an mF set (or a ½0; 1 m -set ) and it is represented as (U, A) where A ¼ p 1 A; p 2 A; . . .p m ð o A ) and p i A is the i th part of A.
Definition 3 (Akram et al. 2018) Let U be an initial universe, Z a set of parameters and C Z. Then, a pair ðc; CÞ is said to be an mF soft set on U, which is given as below: ðc; CÞ ¼ fðu; p i C ðuÞÞ : u 2 Uand 2 Cg: be an interval number where 0 M L M U 1: We denote the collection of all sub-intervals of the closed interval [0, 1] by D[0, 1] . If both the lower and the upper bound of an interval are same, that is, [M, M], then M 2 ½0; 1: For interval numbers Let U ¼ fu 1 ; u 2 ; Á Á Á ; u n g be a universe having 'n' objects, and C ¼ fc 1 ; c 2 ; Á Á Á ; c r g be a set of parameters. Then, the tabular arrangement of an IVmFSS ðg; CÞ is given below (see Table 1).
Example 1 Let U ¼ fu 1 ; u 2 ; . . .; u 5 g be a set of five hotels and Z ¼ fc 1 ¼ material; c 2 ¼ location; c 3 ¼ beauty; c 4 ¼ priceg be a universe of parameters and C ¼ fc 1 ; c 3 ; c 4 g Z.
These parameters are further characterized as follows.
• The parameter ''Location'' includes centrality, neighborhood and commercial development. • The parameter ''Beauty'' includes landscape, furniture and wall decorations.
Then, the tabular representation of an IV3FSS ðg; CÞ is displayed in Table 2.
From Table 1, one can easily see that the evaluations for every object regarding parameters are not clear unless the lower and upper bounds of these evaluations are provided. For instance, one cannot describe the exact membership value about wooden material used in the hotel u 1 in the first pole of first cell of Table 2, that is, the least membership degree bound regarding wooden material is 0.6 and most membership degree bound is 0.8. Definition 6 Let ðg; CÞ be an IVmFSS over a universe U and gðcÞ be the IVmF set of parameter c, then a set of all IVmF sets in IVmFSS ðg; CÞ are called IVmF class of ðg; CÞ, and is denoted by Cl ðg;CÞ , then Cl ðg;CÞ ¼ fgðcÞjc 2 Cg: Definition 7 Let U be a universal set, Z be a universe of parameters and C 1 ; C 2 Z. For two IVmFSSs ðg 1 ; C 1 Þ and ðg 2 ; C 2 Þ, we say that ðg 1 ; C 1 Þ is called the IVmF soft subset of ðg 2 ; C 2 Þ and is denoted by ðg 1 ; C 1 Þbðg 2 ; C 2 Þ if 1. C 1 C 2 , 2. g 1 ðcÞ is an IVmF subset of g 2 ðcÞ; for all c 2 C 1 .

Definition 9
The complement of an IVmFSS ðg; CÞ over a nonempty universe U is represented by ðg; CÞ $ ¼ ðg $ ; :CÞ, which is given as ðg; CÞ $ ¼ ðg $ ; :CÞ ¼ fh:c; g $ ð:cÞj:c 2 Cig; where g $ ð:cÞ is an IVmF set of U, 8c 2 C. In other words, for all :c 2 :C and u 2 U. Here :C represent the not set of parameters which holds opposite meaning corresponding to each parameter c 2 C.
Example 4 Assume data in Example 1 again, where an IV3FSS is given in Table 2. Then, by Definition 9 its complement is computed in Table 5.
We now investigate the DeMorgan's laws for IVmFSSs.
Definition 13 Let U ¼ fu 1 ; u 2 ; Á Á Á ; u n g be a universe, Z ¼ fc 1 ; c 2 ; Á Á Á ; c r g a universal set of parameters and C Z. For an IVmFSS ðg; CÞ, the accumulated score of an object regarding an arbitrary parameter for all given poles is denoted by ðb cj ðu i ÞÞ, and is computed by     Definition 14 Let U ¼ fu 1 ; u 2 ; Á Á Á ; u n g be an initial universal set, Z ¼ fc 1 ; c 2 ; Á Á Á ; c r g a universal set of parameters and C Z. For an IVmFSS ðg; CÞ, we define the final score S i for every element u i of the universe which is given as This novel proposed decision-making method under IVmFSS model is supported by an algorithm below (see Algorithm 1).
Example 7 Let U ¼ fu 1 ; u 2 ; u 3 ; u 4 ; u 5 ; u 6 g be the set of six laptops, Z ¼ fc 1 ¼ costly; c 2 ¼ beauty; c 3 ¼ design; c 4 ¼ technology; c 5 ¼ materialg a set of parameters and C ¼ fc 1 ; c 2 ; c 3 g Z. Then, an IVmFSS ðg; CÞ is displayed in Table 10. We now apply Algorithm 1 to IVmFSS ðg; CÞ. By using Equations (3) and (4), the scores of lower and upper bounds for every pole (interval) are given in Tables 11 and 12 . First cell of the Table 11 is computed as below.
From Definition 13, the tabular representation for the accumulated scores of membership values of IV3FSS ðg; CÞ with respect to each parameter is displayed by Table 13.
By Definition 14, the final score of every laptop u i is displayed in Table 14. First entry of final score table is computed as: ¼ ðÀ1:3Þ þ ð0:9Þ þ ð3:3Þ ¼ 2:9: From Table 14, the object having highest score is S 4 ¼ 6:3. Thus, it can be used as decision object.
By using Equations (3) and (4), the scores of lower and upper bounds for every pole (interval) are given in Tables 16 and 17 . First cell of the Table 16 is computed as below. Similarly, one can readily compute the remaining values, which are displayed in Table 16 and 17 . From Definition 13, the tabular representation for the accumulated scores of membership values of IV4FSS ðg; CÞ with respect to each parameter is displayed by Table 18.
By Definition 14, the final score of every car u i is displayed in Table 19. First entry of final score table is computed as: c 1 c 3 u 1 3.6 0.5 u 2 À 0.9 1.0 u 3 À 1.9 À 1.0 u 4 1.6 À 1.0 From the Table 19, the object having highest score is u 1 because S 1 ¼ 4:1. Thus, u 1 can be selected as decision.
In the following, to find the optimal decision based on IVmF soft sets, we construct a flowchart diagram which describe the proposed mathematical method more precisely and feasibly.
By the analysis of above examples, one can easily see that developed decision-making method under IVmFSSs is useful and reliable. Although, decision-making hybrid methods involving soft sets as one of their component may contain some redundant parameters. To remove this drawback in the developed hybrid model, we now provide three parameter reductions approaches for IVmFSSs.

Parameter reductions of IVmFSSs
An approach to reduce the parameter set to acquire a minimal subset of parameter set that provides a decision similar to the whole set of parameter is called parameter reduction. In this section, we investigate three kinds of parameter reductions of IVmFSSs to handle different reduction situations.

Parameter reduction based on optimal choice
We discuss the parameter reduction based on optimal choices and then give an algorithm for this parameter reduction approach, which is explained through an example.
Definition 15 Let U ¼ fu 1 ; u 2 ; Á Á Á ; u n g be a universe, Z ¼ fc 1 ; c 2 ; Á Á Á ; c m g a set of parameters and C Z. For an IVmFSS ðg; CÞ, we denote a subset O C U as a set having optimal values of final score S i . For any A C, if O CÀA ¼ O C , then A is said to be dispensable in C, else, A is said to be indispensable in the favorable parameter set C. The set C of parameters is said to be independent if each A & C is indispensable in C, otherwise, C is dependent. Any set B & C is called a parameter reduction based on optimal choice (PR-OC, henceforth) of C if it satisfy the following axioms.
1. B is independent (it means B & C which is minimal and keeps the decision unchanged).
Using the Definition 15, we present an algorithm for PR-OC which remove the irrelevant parameters while preserving the decision invariant.
Example 9 Consider Example 7 where C ¼ fc 1 ; c 2 ; c 3 g Z. We apply Algorithm 2 to the IVmFSS ðg; CÞ. Using Table 20, we deduce that for B 0 ¼ fc 2 ; c 3 g, we obtain Hence, one PR-OC of IVmFSS ðg; CÞ is given by C À B 0 ¼ fc 1 g which is displayed in Table 20. From Table 20, one can easily see that u 4 is the decision after reduction. Definitely, the subset fc 1 g & C is smallest which preserves decision invariant.

Rank based parameter reduction
Nowadays, most of the practical problems are mainly solved to find the rank of all the alternatives under consideration. Since, the objects other than optimal choice are not considered in PR-OC technique. To tackle this issue, we define a novel parameter reduction which preserves the rank of all the objects and develop an algorithmic method which maintains the ranking order of all the objects after reduction.
called a rank based parameter reduction (RB-PR, in short) of Z if it satisfy the following two conditions.
1. B is independent (it means B & Z is smallest which keeps the ranking order of all the objects invariant, including optimal decision object). 2. D B ¼ D Z .
Using Definition 17, we now present an algorithm(see Algorithm 3) for the RB-PR method that reduces the set of parameters while preserving the original rank fation.
From Definition 13, the tabular representation for the accumulated scores of membership values of IV3FSS ðg; CÞ with respect to each parameter is displayed by Table 24.
By Definition 14, the final score of every laptop u i is displayed in Table 25. First entry of final score table is computed as: From the Table 25, one can readily see that the object having highest score is u 4 because S 4 ¼ 6:8. Thus, u 4 is the optimal decision object. Furthermore, using Table 25, it can readily computed that D C ¼ ffu 4 g 6:8 ; fu 5 g 0:8 ; fu 3 g 0:3 ; fu 2 g À3:7 ; fu 2 g À4:2 g: By applying Algorithm 3, we now compute a minimal subset of C which preserves the rank of all objects of the universe. Thus, for B 0 ¼ fc 5 g, we obtain D CÀB 0 ¼ ffu 4 g 6:9 ; fu 5 g 5:4 ; fu 3 g À2:1 ; fu 2 g À3:1 ; fu 2 g À7:1 g with D CÀB 0 ¼ D C . Notice that partition and rank of elements of the universe are same after reduction. Thus, fc 3 ; c 4 g is the only RB-PR of IV3FSS ðg; CÞ as displayed by Table 26. Clearly, fc 3 ; c 4 g & C is the only minimal subset which keeps the ranking order of all objects invariant.

Normal parameter reduction
The reduction approaches discussed above may not be useful in different real situations. That's why, we give another reduction approach called normal parameter reduction (NPR) for IVmFSSs, which handles the issue of added parameters. We propose the notion of NPR and provide its algorithmic approach, that is, how to remove redundant parameters using NPR method.
Definition 18 Let U ¼ fu 1 ; u 2 ; Á Á Á ; u n g be a universe of objects, C & Z ¼ fc 1 ; c 2 ; Á Á Á ; c r g a favorable set of parameters. For an IVmFSS ðg; CÞ, B is said to be dispensable if we compute a set B ¼ fc 1 ; c 2 ; Á Á Á ; c p g & C, which verify the expression given below. X Otherwise, B is called indispensable. A set N & C is said to be NPR of C, if it satisfy the conditions given as follows.
Using Definition 18, we develop the NPR algorithm as below:  Example 11 Let U ¼ fu 1 ; u 2 ; u 3 ; u 4 g be the set of four houses, Z ¼ fc 1 ¼ costly; c 2 ¼ beauty; c 3 ¼ design; c 4 ¼ location; c 5 ¼ materialg a set of parameters and C ¼ fc 1 ; c 2 ; c 5 g Z. Then, an IV3FSS ðg; CÞ is displayed in Table 27. We now apply Algorithm 4 to IV3FSS ðg; CÞ. (3) and (4), the scores of lower and upper bounds for every pole (interval) are given in Tables 28 and 29 . First cell of the Table 28 is computed as below.  Similarly, one can readily compute the remaining values, which are displayed in Table 28 and 29 . From the Definition 13, the tabular representation for the accumulated scores of membership values of IV3FSS ðg; CÞ for each parameter is displayed by Table 30.

By using Equations
By Definition 14, the final score of every house u i is displayed in Table 31. First entry of final score table is computed as: ¼ ðÀ1:2Þ þ ð1:5Þ þ ð1:2Þ ¼ 1:5: Clearly, objects u 1 and u 4 have maximum score that is 1.5. Thus, one from them can be chosen as optimal decision. By Table 31, one can easily observe that for N ¼ fc Thus, C À fc 1 ; c 5 g ¼ fc 2 g is the NPR of IVmFSS ðg; CÞ, which is given by Table 32.

Application to MCDM
This section solves two real decision-making situations using the developed model and discusses the impact of the proposed parameter reduction approaches on them.

Case Study: Selection of a suitable site for an airport
Choosing an appropriate site for another airport, or assessing how suitably a current site can be extended to give another significant airport, is a complicated procedure. A proportion must be accomplished among air-transport and aeronautical needs and the effect of the airport on its current circumstance. For an aeronautical perspective, the fundamental necessity of an airport is its generally flat area of land adequately enormous to adapt the runways and different services and that this site is in a territory liberated from such obstacles to air route as tall buildings and mountains. For the perspective of air-transport requirements, airport sites should be adequately near to population centers that these are thought to be approached by users easily. However, environmental factors demand that the location should be too much long way from urban areas which will overcome the noise and other destructive impacts on the population to tolerable levels. Moreover, the natural beauty of different areas and other important assets should not be destroyed by the airport. The environmental and aeronautical, nearly necessarily clash, with the conflict getting more serious as the size of the envisaged airport increases for these two sets of requirements. The most unassuming airport facility with an aircraft parking, a building, and a single runway that serves at the same time as terminal, control tower, and administration area can quietly be constructed on a location as little as 75 acres since it needs just a flat, very much depleted area adequate to oblige a short runway and its encompassing safety strip. On the other hand, more modern and huge airport facilities need a large number of runways of huge length, huge terminal aircraft parking areas, and huge territories of land committed to landside access roads and parking. For this kind of airport, a base area of 3000 acres is probably going to be needed. A few significant airports, for example, King Abdul Aziz International Airport close to Jeddah, Saudi Arabia, Charles de Gaulle Airport close to Paris, and Dallas-Fort Worth International Airport in Texas are based on destinations well in overabundance of above mentioned figure. The site selecting procedure for a huge airport can take several months; in some essential circumstances, it has gone on for several years. The difficulty in the process is due to the involvement of several factors. First, evaluate the operational capabilities of the station, especially for weather conditions (such as fog, low visibility, ice, snow, and wind), as well as obstacles to air navigation nearby the airport, especially on the approach and take-off paths. The location of the facility relative to the air traffic control airspace is also practically significant. Furthermore, the capability of available land must be evaluated to adjust the expected configuration of runways and other facilities. The landing must be flat or very gentle, because the runway must be constructed based on the maximum allowable slope, which depends on the performance of the aircraft during landing and takeoff. Also consider the ground access of the airport. Evaluate the distance to population centers, regional highway infrastructure, public transportation facilities (like railways), and the distance to land available for parking. Also consider the nature of the terrain, rock and soil conditions, drainage needs and local land value to estimate development costs. In a site selection process, the effect of an airport development on the environment is very high. The effect of aircraft noise on the surrounding population is usually a very important environmental factor, but in different countries, the effect on the fauna and flora of the area must also be considered, the pollution of local groundwater by chemical runoff, the existence of endangered species or important culture sites, and even bad changes in land use. Several countries now demand environmental analysis of airport development projects, including changes in employment patterns, assessments of population migration, transportation plans and distortions in existing regional land use. Suppose that government of a country planned to construct a new airport on a most suitable site from twenty alternatives. This critical task is given to a team of experts of the field. Let U ¼ fu 1 ; u 2 ; . . .; u 20 g be a set of twenty sites for the selection of most suitable site for an airport construction and Z ¼ fc 1 ¼ size; c 2 ¼ cost; c 3 ¼ environmentalconsequences; c 4 ¼ groundaccesg be a set of parameters from which C ¼ fc 2 ; c 3 ; c 4 g Z are favorable according to the team of experts. These parameters can be further classified as follows: • The parameter ''Size'' includes medium, large, very large.
• The parameter ''Environmental Consequences'' includes distortion of existing regional land use, changes in employment patterns, and evaluations of population relocation. • The parameter ''Ground Access'' includes public transport facilities, the regional highway infrastructure, distance from population centers.
The report collected from the team of experts is in the form of an IV3FSS, which is displayed by Table 33. By using Equations (3) and (4), the scores of lower and upper bounds for every pole (interval) are given in Tables 34 and 35 . First cell of the Table 34 is computed as below. Similarly, one can readily compute the remaining values, which are displayed in Table 34 and 35 . From Definition 13, the tabular representation for the accumulated scores of membership degrees of IV3FSS ðg; CÞ with respect to each parameter is displayed by Table 36.
By Definition 14, the final score of every site u i is displayed Table 37. Clearly, from Table 37 the object having highest score is u 15 because S 15 ¼ 20:4. Thus, u 15 is the most suitable site from all the available alternatives. From the Table 37, it can be readily compute that B 0 ¼ fc 2 g C such that C À B 0 ¼ fc 3 ; c 4 g is the only PR-OC.
Regrettably, there is no RB-PR in present situation. Now we apply our decision-making Algorithm 1 under IVmFSSs, and the reduction Algorithms 2, 3 and 4 , to another situation from agriculture engineering.

Case Study: Selection of a suitable rotavator machine
In agricultural engineering, a rotavator is a useful machine for seedbed preparation. It is directly connected with a tractor to plow the soil by a set of blades that cuts, mixes, pulverizes and level the soil and makes the ground perfect before planting bulbs and seeds. Thus blades become very essential components in the rotavator. This is an effective agriculture equipment which replaces the disc harrow, cultivator and leveler because it works collectively of these three equipments. It is better than traditional agricultural rotary machines due to its rapid and efficient seedbed preparation. The rotavator is a very reliable source of transmission of the engine power of a tractor directly to the soil without any serious reduction in transmission power loss and wheel slip. Due to this fact, it saves a lot of time and reduces the cost of operation more than other classical tillage machines. In the manufacturing process of a rotavator, major errors can be reduced by the analysis of its components design. The design optimization of a rotavator machine is achieved by decreasing its cost and weight, and by enhancing a field efficiency to high weed removal performance. Soil conditions directly affect the types of blades selected in terms of power requirement of rotary cultivator. For example, clay soil consumes more power than loamy soil. Therefore, soil condition is also an important factor in the development of a suitable rotavator. Other main factors which affect the selection procedure of an appropriate rotavator are: size, cost, fuel consumption and material quality. Nowadays, with the advancement in agriculture sector, agriculture engineers are trying to enhance the rotavator design and its material, with the goal of achieving maximum outputs in minimum time and with the lowest cost. Experts believe that there are different characteristics which should be considered in the selection of a rotavator, such as size, cost, fuel consumption, blades shape, material quality, soil condition etc.
• The parameter ''Material Quality'' includes low, medium, and high.
The report collected from the team of experts is in the form of an IV3FSS, which is displayed by Table 38.
By using Equations (3) and (4), the scores of lower and upper bounds for every pole (interval) are given in Tables 39 and 40 .
From Definition 13, the tabular representation for the accumulated scores of membership degrees of IV3FSS ðg; CÞ with respect to each parameter is displayed by Table 41.
By Definition 14, the final score of every site u i is displayed Table 42. Clearly, from Table 42 the object having highest score is u 1 because S 1 ¼ 7:1. Thus, u 1 is the most suitable rotavator from all the available alternatives. From Table 42, it can be readily computed that B 0 ¼ fc 1 ; c 2 g C such that C À B 0 ¼ fc 4 g is the only PR-OC.
Regrettably, there is no RB-PR and NPR in the present situation. Figure 2 presents a flowchart to compute the parameter reductions of IVmFSSs.

Comparison
This section provides a detailed comparison among the reduction methods developed in Sect. 3, in terms of their respective reduction of computational speed and scope of application. It also gives a comparative discussion between the IVmF model developed in this paper and some existing models.

Comparison of reduction of computational speed and scope of application
Suppose that j represents the number of parameter reductions calculated in one or more data-sets. Basically, j denotes the application scope of the developed reduction approaches in different situations and described as computation speed in reduction techniques.
• PR-OC only preserves the maximum score value, that is, decision object. That is the reason why PR-OC is an easy and fast way to compute reduction set. For example, fc 1 g is the PR-OC in Example 7, fc 1 g is the PR-OC in Example 8, fc 4 g is the PR-OC in Example 10, fc 2 g is the PR-OC in Example 11. Hence, j ¼ 100% because in all presented numerical examples we easily computed PR-OC.  • The RB-PR approach maintains the rank of all alternatives under consideration. Therefore, RB-PR is difficult as compared to PR-OC because this reduction method not only preserves rank of optimal decision object but also sub-optimal choices. For example, fc 3 ; c 4 g is the RB-PR in Example 10 and fc 2 g is the RB-PR in Example 11, It can be easily see that we find no RB-PR in Examples 7 and 8 . Thus, j ¼ 2=4 ¼ 50%. • NPR preserves not only rank but also maintain final score values of objects. The computation of NPR in a given problem is difficult as compared to aforementioned reduction techniques. Clearly, fc 2 g is the only NPR in Example 11. Unfortunately, we compute no NPR in Examples 7, 8 and 10 . Hence, j ¼ 1=4 ¼ 25%: As we know, in the case of NPR the reduction set can be reused if new parameter added in the set of parameters. Now we give an example to examine this issue.  u 9 À 6.0 11.7 À 0.6 u 10 À 4.0 0.7 À 3.6 u 10 À 1.9 À 12.5 0.8 À 6.9  Table 43.

Discussion
In this section, we give a comparison of the developed model with certain existing hybrid models and also provide a comparative analysis of the proposed reduction approaches with some existing reduction methods.
1. IVmFSSs constitute a generalized structure of IVmF sets or mF soft sets because they are still a function from set of parameters to the set of IVmF subsets of universal set.  Fig. 3 that optimal and sub-optimal decision objects are different. Since, u 15 is the optimal decision object by the proposed decision-making method, that is, IVmFSSs, Clearly, u 19 is the optimal decision object by the IVFSS model (Yang et al. 2009). Similarly, it can be readily see from the Table 47 that the overall rankings are also different for the proposed IVmFSS model and IVFSS model (Yang et al. 2009). The reason behind these differences is that IVFSSs (Yang et al. 2009) only consider one pole (one membership value) with respect to parameters, that is, in this case for a particular parameter its all properties are not considered which may lead to wrong decision because all the parameters and their further characterizations (poles) are independent. That's why, when we consider all possible further features of a particular parameter, the ranking of optimal and suboptimal choices changed or we can say more exact decision choices computed because in the case of IVmFSSs decision is made on the basis of all the possible information in a given decision-making problem. Thus, IVmFSS model is more generalized because it has strength to consider all possible properties of any parameter in a decision-making problem as compared to IVFSSs (Yang et al. 2009). • When only one parameter and some of its characteristics are considered in decision-making process of the IVmFSSs, it degenerates into IVmF set theory (Mahapatra et al. 2020). We now apply the decision-making method proposed by Mahapatra et al. (2020) to the developed application in Sect. 4 and the obtained results are given in Table 46 and displayed by Fig. 3. From the Table 46 and Fig. 3, it can be easily see that u 15 and u 9 are the optimal decision objects for the proposed IVmFSSs and IVmF sets (Mahapatra et al. 2020), respectively. The reason behind these differences is that in a given problem all the parameters are independent, that is, their membership values are not dependent to each other. Intuitively, if there exist more than one important parameters in a given problem then IVmF set model (Mahapatra et al. 2020) only consider one parameter which may lead to wrong decision because in this case some important parameters may be missed in the decision-making process. Therefore, IVmFSS model is an extension of IVmF set model with respect to parameters and provides more accurate results than IVmF sets (Mahapatra et al. 2020). • Chen et al. (2014) proved that 2-polar fuzzy sets and bipolar fuzzy sets (YinYang bipolar fuzzy sets) are cryptomorphic mathematical tools. According to this strong fact, interval-valued bipolar fuzzy soft set model and IV2FSS model are crypotomorphic mathematical notions. Thus, interval-valued bipolar fuzzy soft set model is a particular case of our proposed IVmFSS model, for m ¼ 2. • When the mF values are fixed, that is, not in the interval form, IVmFSS model degenerates into mF soft set model (Akram et al. 2018).
Thus, our proposed model is a generalization of the IVFSS model (Yang et al. 2009), IVmF set model (Mahapatra et al. 2020) and mF soft set model (Akram et al. 2018).
2. PR-OC approach only preserves the decision object invariant after reduction (it means, the ranking order of sub-optimal object may be varied after reduction). Thus, the reusability of reduction set is lower. RB-PR method removes redundant parameters by keeping the partition and ranking order of all elements under consideration. Thus, the re-usability of reduction set is higher than PR-OC. With similar arguments, it is clear that the re-usability of reduction set is highest in case of NPR method. A comparison between the optimal decision values obtained by proposed IVmFSS model and its PR-OC on the Application explored in Sect. 4 is displayed in Fig. 4. Moreover, reduction set is obtained by applying the parameter reduction techniques discussed in ) for IVFSS and is compared with reduction set computed for IV3FSS in Application 1 (Sect. 4). For more clarification, the obtained results are displayed in Fig. 5.

Conclusions and future directions
The theory of IVFSSs is arising as a helpful expansion of soft sets which is upheld by genuine data-sets. In this study, we have improved upon the hypothetical premise of this theory in directions that are validated by their impact on specific settings and their role in various theories. In this context, we have developed a novel hybrid model, namely, IVmFSSs. It produces a formal generalization of many existing models. It can be regarded as a multi-fuzzy extension of the IVFSS model. Alternatively, it can be considered as a IVmF extension of the soft set model (Molodtsov 1999). At one and the same time, it is an interval-valued extension of the mF soft set model (Akram et al. 2018). Some fundamental operations, including complement, union, intersection, ''AND'', ''OR'' are studied on the IVmFSSs and investigated through examples. An algorithm is developed to handle decision-making situations having data in interval-valued multi-fuzzy soft form, which has been applied on two numerical examples. In addition, three parameter reduction approaches in algorithmic expression are proposed for IVmFSSs, namely, PR-OC, RB-PR and NPR. After that, two real case studies for the selection of best site for an airport construction and best rotavator are explored. Finally, the significance and rationale behind the new hybrid model and its parameter reduction methods are discussed, particularly through a comparative analysis with some existing approaches like IVFSSs ( Yang et al. 2009). We may advance some lines of research whose examination should produce innovative contribution: • Alternative methodologies from additional perspectives for parameter reduction are still possible, regardless of whether they are original or imported from external settings. • Our overall research objectives can be exported to other relevant settings such as spherical fuzzy soft sets (Perveen et al. 2019).
Funding Thanks to the CRUE-CSIC agreement with Springer Nature. J. C. R. Alcantud is grateful to the Junta de Castilla y León and the European Regional Development Fund (Grant CLU-2019-03) for the financial support to the research unit of excellence ''Economics Management for Sustainability'' (GECOS).