International Journal of Engineering Trends and Technology (IJETT) – Volume 6 Number 3- Dec 2013 Database Oriented Validation of Theorems on Approximations of Classifications in Optimistic and Pessimistic Multigranulation Based Rough Sets R. Raghavan SITE, VIT University, Vellore-632014, Tamil Nadu, India Abstract— The concept granular computing was in ability to classify objects. The concept of introduced by Zadeh which deals with granules and approximations of classifications was introduced and includes all methods and tools which provide flexibility studied by Busse [1]. In his work [1] he revealed some and adaptability in the decision at which knowledge or important and interesting results. After analysis it has information is extracted and represented. According to granular computing an equivalence relation on the universe can be considered as a granulation, and a partition on the universe can be considered as a been found that not all the properties of basic rough sets can be extended to the category of classifications. In [1] Busse established four theorems as properties of granulation space. Two types of multigranulation of rough classifications and the obtained results could be used in sets were defined from the view of granular computing. rule generation. The unification process of the four One is called to be the optimistic multigranular rough set theorems of Busse was done by Tripathy et al [7, 8] in was introduced by Qian et al [4]. During evolution there the idea that they established two theorems of the was only one type of multigranulation and it was not necessary and sufficient type, from which several named as optimistic multigranulation rough sets. After the results including the four theorems of Busse can be improvement of a second type of multigranulation, the first one was called optimistic multigranulation rough set and the second one was called as pessimistic multigranulation rough sets. For granulations of higher derived. Also, their results confirm to the prediction of Pawlak. From the view of information systems, the role of these four theorems was found to be useful in order (more than two), the definitions and properties are deriving rules from such systems. It was observed that similar. Using multigranular rough sets several theorems only five cases were considered by Busse out of eleven with respect to approximations of classifications was cases as far as the types of classifications, on the other introduced in [11]. In this paper the validation for the hand, the other types of classifications reduce either theorems on approximations of classification in [11] is directly or indirectly to the five cases considered by provided with a database support for both optimistic and Busse. Another interesting aspect of the results in [7, 8] pessimistic multigranular rough sets. is the enumeration of possible types [2, 3, 9] of elements in a classification, which is based upon the Keywords— validation, approximations of classification, types of rough sets introduced by Pawlak [3] and rough carried out further by Tripathy et al [7, 8]. From the sets, optimistic multi-granular rough sets, pessimistic multi-granular rough sets view of granular computing the above results on approximation of classifications were based upon single I. INTRODUCTION granulation. Two types of multigranulations using In rough set theory role of classification of universes rough sets were recently introduced in the literature. were considered to be more important due to its usage ISSN: 2231-5381 http://www.ijettjournal.org Page 125 International Journal of Engineering Trends and Technology (IJETT) – Volume 6 Number 3- Dec 2013 These are termed as optimistic [4] and pessimistic [5] R, S R . The pessimistic multigranular lower multigranulation. approximation and pessimistic multigranular upper The result of Tripathy et al [7, 8] is extended to approximation of X with respect to R and S in U as the multigranulation context. In [6] the study about the types of basic rough sets was done from the view of optimistic multigranulation rough set and from the pessimistic multigranulation point of view was done in R S X { x | [x]R X and [x]S X} and R S X ~ ( R S (~ X )). III. PROPERTIES OF MULTIGRANULATIONS [10]. A classification F = { X1 , X 2 , ..., X n } of a universe U is n such that X i X j = for i j and X k U . k 1 Some properties of multigranulations which shall be used in this work to establish the results is given below. 3.1 classification F was defined by Busse as PROPERTIES OF OPTIMISTIC MULTIGRANULAR ROUGH SETS The following properties of the optimistic multigranular rough sets were established in [4]. RF {RX1 , RX 2 , ..., RX n } (R + S)(X) X (R + S)(X) The approximations (lower and upper) of and RF {RX1, RX 2 ,...RX n } a . (3.1) (R +S)( ) = = (R +S)( ) , (3.2) For any two equivalence relations R and S (R +S)(U) = U = (R +S)(U) over U, the lower and upper optimistic multigranular (R + S)(~ X) =~ (R + S)(X) (3.3) rough approximations and pessimistic multigranular (R + S)(X) = RX SX (3.4) ( R S )( X ) R X S X (3.5) approximations are defined in a natural manner as R +SF={R +SX1 ,R +S X 2 ,...,R +SX n } and …..1.1 R +SF={R +SX1 ,R +S X 2 ,...R +SX n } R*SF={R*SX1,R*SX 2 ,...,R*SX n } and ……1.2 R*SF={R*SX1,R*SX 2 ,...R*SX n } II. DEFINITIONS (R + S)(X) = (S + R)(X), (3.6) (R + S)(X) = (S + R)(X) (R + S)(X Y) (R + S)X (R + S)Y (3.7) (R + S) (XUY) (R + S)X U (R + S)Y (3.8) (R + S)(XUY) (R + S)XU (R + S)Y (3.9) (R + S) (X Y) (R + S)X (R + S)Y (3.10) 3.2 2.1 DEFINITION: Let K= (U, R) be a knowledge base, R be a family of equivalence relations, X U and R , S R . The optimistic multigranular lower PROPERTIES OF PESSIMISTIC MULTIGRANULAR ROUGH SETS The following properties of the pessimistic multigranular rough sets which are parallel to the above said properties were established in approximation and optimistic multigranular upper approximation of X with respect to R and S in U as R S X { x | [x]R X or [x]S X} and R S X ~ ( R S (~ X )). 2.2 DEFINITION: Let K= (U, R) be a knowledge base, R be a family of equivalence relations, X U and ISSN: 2231-5381 http://www.ijettjournal.org Page 126 [5]. International Journal of Engineering Trends and Technology (IJETT) – Volume 6 Number 3- Dec 2013 (R*S)(X) X (R *S)(X) (3.11) (i) R S ( X ) i (R*S)() = = (R*S)(), (R *S)(U) = U = (R*S)(U) (3.12) (R*S)(~ X) =~ (R *S)(X) (3.13) (R*S)(X) = RX SX (3.14) (R*S)(X) = RXUSX (3.15) (R*S)(X) = (S*R)(X), (R*S)(X) = (S*R)(X) (3.16) (R*S)(X Y) (R*S)X (R*S)Y (3.17) (R*S)(X Y) (R*S)X (R*S)Y (3.18) (R*S)(X Y) (R*S)X (R*S)Y (3.19) (R*S)(X Y) (R*S)X (R*S)Y (3.20) R S Xj U. j I C i I (ii) R * S( X j ) R * SX j U. jI jIC PROOF: R + S( X j ) jI x Usuch that [x]R X j or[x]S X j . jI jI Thus [x]R ( X j ) = or [x]S ( X j ) = . jIC jIC C C So, [x]R ( X j ) or [x]S ( X j ) . jIC jIC C Hence, x R + S( X j ) . C jI IV.THEOREMS ON APPROXIMATIONS OF CLASSIFICATIONS IN MULTIGRANULATIONS C C This implies that x ( R + S ( X j ) ) j I C . The notation used in this section is denoted as N n {1, 2, 3,....n } . So, x R + S( X j ). jI C For any I N n , I C denotes the complement of I in N n . This proves that R + S ( THEOREM 4.1: For any I N n The proof of (ii) is similar. R S ( X i ) = U, if and only if iI R S( Xi ) U iI if and only if Xj )= jI C R S ( U \ Xi ) = iI ~ R S ( Xi ) = iI R S ( Xi ) = U iI The proof for pessimistic case is similar. ISSN: 2231-5381 ON THEOREMS OF APPROXIMATIONS OF CLASSIFICATIONS IN Facult y Name Sam PROOF: THEOREM 4.2: For any I N n VALIDATION The following faculty details database of an university with assumed data is used to validate the theorems of approximations of classifications in multigranulation rough sets. R S ( X j ) . jI C RS ( V. X j) U. MULTIGRANULATIONS ROUGH SET THEORY Xj )=. jI C RS( j I C Ram Div isio n NW Grade AP Highes t Degree M.C.A Pr Ph.D APJ M.sc., IS Shyam SE Peter AI AS P Ph.D Roger ES P Ph.D Native State Tamil Nadu Andhra Pradesh Tamil Nadu Tamil Nadu Tamil Nadu , http://www.ijettjournal.org Page 127 International Journal of Engineering Trends and Technology (IJETT) – Volume 6 Number 3- Dec 2013 AI APJ M.Sc., Mishra ES APJ M.Sc., Albert Andhra Pradesh Tamil Nadu Hari IS SP Ph.D Orissa John SE AP M.C.A Smith NW ASP Ph.D West Bengal Orissa Linz AI SP Ph.D Orissa Keny SE P Ph.D Karnataka ASP Ph.D APJ M.Sc., Tamil Nadu Karnataka Willia ES ms Martin IS APJ M.Sc., Karnataka Lakma SE n Sita AI ASP Ph.D AP M.Tech West Bengal Karnataka Fatima ES AP M.Tech Muker jee Preeth a IS SP Ph.D SE SP Ph.D Jacob NW West Bengal West Bengal Karnataka {shyam, albert, mishra jacob,sam, john,sita, fatima, ram, peter, roger, hari,smith, keny, linz, williams, mukherjee, preetha} / Highest Degree {{shyam, albert, mishra jacob},{sam, john},{sita, fatima},{ram, peter, roger, hari, smith, keny, linz, williams, mukherjee, preeth}} / Native State {{sam,shyam, roger, mishra, williams},{ram, peter},{hari, smith, linz},{peter, john, lakman, fatima, mukherjee},{keny, martin, jacob,sita, pretha}} / Division {{sam,smith, jacob},{shyam, john, keny, lakman, pretha},{peter, albert, linz, sita}, {roger, mishra, williams, fatima},{ram, hari, martin, mukherjee}} Let us define a classification F namely grade from the faculty details table as given below F / Grade {{shyam, albert, mishra, martin, jacob}, {sam, john,sita, fatima},{peter,smith, williams, lakman},{linz, mukherjee, preetha, hari}, {ram, roger, keny}} F {X1, X 2 , X 3 , X 4 , X 5} N n {1, 2, 3, 4, 5} The database which is stated above is used for the I N validation. Here NW indicates networks, SE indicates I {1, 2, 3, 4} Software I C {5} engineering, AI indicates Artificial Intelligence, ES indicated Embedded Systems, IS indicated Information Systems. Similarly AP indicates Assistant Professor, APJ indicates Assistant Professor (Junior), ASP indicates Associate Professor, SP n Let Let U/Grade be the classification named F U/ R be U / Native state U/S be U / Division For validating theorem 4.1 indicates Senior Professor, Pr indicates Professor. The universe and other equivalence relations are defined below. ISSN: 2231-5381 http://www.ijettjournal.org Page 128 International Journal of Engineering Trends and Technology (IJETT) – Volume 6 Number 3- Dec 2013 5.1 VALIDATION OF THEOREM 4.1 Validation of Theorem 4.2 OPTIMISTIC CASE For any I N n For any I N n R S( iI Xi ) ,if and only if , R S( Xi ) R SX j R S( jIC X j ) jI C iI (ii) R * S( X j ) R * SX j U. jI C jI Take R S( iI Xi ) R S(X1 X 2 X3 X 4 ) F {X1, X 2 , X 3 , X 4 , X 5} ~ (R S(~ (X1 X 2 X 3 X 4 ))) N n {1, 2, 3, 4, 5} ~ (R S(X 5 )) ~ ([x]R X5 or [x]S X 5 ) I N n I {1, 2} Here fifth classification X 5 is {ram, roger, keny} I C {3, 4, 5} Let Let U/Grade be the classification named F R S( jIC X j ) R S(X 5 ) [x]R X 5 or [x]S X 5 U/R be U / Highest Degree U/ S be U / Native state for validating theorem 4.2 {}or{} ~ ( or ) {} So R S( iI X i ) ,if and only if OPTIMISTIC CASE R S( jIC X j ) R S( iI Xi ) R S(X1 X 2 ) Hence Validated [x]R {X1 X 2 }or[x]S {X1 X 2 } {sam, john}or{} {sam, john} PESSIMISTIC CASE To Pr ove R SX R *S( iI X i ) , if and only if j R S(X3 X 4 X 5 ) jIC R *S( jIC X j ) ~ R S(~ (X 3 X 4 X 5 )) Take ~ R S(X1 X 2 ) R *S( iI X i ) R *S(X1 X 2 X 3 X 4 ) ~ ({sam, john}or } ~ (R *S(~ (X1 X 2 X 3 X 4 ))) ~ ({sam, john}) ~ (R *S(X 5 )) ~ ([x]R X 5 and [x]S X 5 ) Here fifth classification X 5 is {ram, roger, keny} ~ ( and ) So R S( X i ) iI R *S( jIC X j ) R *S(X 5 ) [x]R X 5 and [x]S X 5 R SX j jIC Hence Validated {}and{} {} So R *S( iI X i ) ,if and only if R *S( jIC X j ) Hence Validated ISSN: 2231-5381 http://www.ijettjournal.org Page 129 International Journal of Engineering Trends and Technology (IJETT) – Volume 6 Number 3- Dec 2013 PESSIMISTIC CASE To Prove R *S( Xi ) R *SX j jIC iI Take I {3, 4,5} IC {1, 2} R *S( iI X i ) R *S(X3 X 4 X 5 ) [x]R {X3 X 4 X 5}and[x]S {X 3 X 4 X 5} {ram, peter, roger, hari,smith, linz, keny, williams, lakman, mukherjee, pretha}and{hari,smith, linz} {hari,smith, linz} R *SX j R *S(X1 X 2 ) jIC ~ R *S(~ (X1 X 2 )) ~ R *S(X3 X 4 X 5 ) ~ ({hari, smith, linz}) So R *S( X j ) jI R *SX j jIC Hence Validated V. CONCLUSION In this paper the validation for the theorems on approximations of classification was successfully validated with a database support for both optimistic and pessimistic multigranular rough sets. This visualizes how the internal segments of theorems had the flow with the database used. Such practical implementation with a database gives more understandability to the analytical work carried in the previous paper. Studies in Computational Intelligence, vol.174, Rough Set Theory: A True Landmark in Data Analysis, Springer Verlag, (2009), pp.85 - 136. [8] Tripathy, B.K., Ojha, J., Mohanty, D. and Prakash Kumar, Ch. M.S: On rough definability and types of approximation of classifications, vol.35, no.3, (2010), pp.197-215. [9] Tripathy, B.K. and Mitra, A.: Topological properties of rough sets and their applications, International Journal of Granular Computing, Rough Sets and Intelligent Systems (IJGCRSIS), (Switzerland),vol.1, no.4, (2010),pp.355-369 . [10] Tripathy, B.K. and Nagaraju, M.: On Some Topological Properties of Pessimistic Multigranular Rough Sets, International Journal of Intelligent Systems and Applications, Vol.4, No.8, (2012), pp.10-17. [11] Tripathy, B.K. and Raghavan, R.: Some Algebraic properties of Multigranulations and an Analysis of Multigranular Approximations of Classifications, International Journal of Information Technology and Computer Science, Volume 7, 2013, pp. 63-70. VII.BIOGRAPHY R. Raghavan is an Assistant Professor (Senior) in the School of Information Technology and Engineering (SITE), VIT University at Vellore in India. He obtained his masters in computer applications from University of Madras. He completed his M.S., (By Research) in information technology from school of information technology and engineering, VIT University. He is a life member of CSI. His current research interest includes Rough Sets and Systems, Knowledge Engineering, Granular Computing, Intelligent Systems, image processing. VI. REFERENCES [1] Grzymala Busse, J.: Knowledge acquisition under uncertainty- a rough set approach, Journal of Intelligent and Robotics systems, 1, (1988), pp. 3 -1 6. [2] Pawlak, Z.: Rough Classifications, international Journal of Man Machine Studies, 20, (1983), pp.469 – 483. [3] Pawlak, Z.: Theoretical aspects of reasoning about data, Kluwer academic publishers (London), (1991). [4] Qian, Y.H and Liang, J.Y.: Rough set method based on Multi-granulations, Proceedings of the 5th IEEE Conference on Cognitive Informatics, vol.1, (2006),pp.297 – 304. [5] Qian, Y.H., Liang, J.Y and Dang, C.Y.: Pessimistic rough decision, in: Proceedings of RST 2010, Zhoushan, China, (2010), pp. 440-449. [6] Raghavan, R. and Tripathy, B.K.: On Some Topological Properties of Multigranular Rough Sets, Journal of Advances in Applied science Research, Volume 2(3), 2011, pp.536-543. [7] Tripathy, B.K.: On Approximation of classifications, rough equalities and rough equivalences, ISSN: 2231-5381 http://www.ijettjournal.org Page 130