68 W. Li and W. Xu / Multigranulation Decision-theoretic Rough Set in Ordered Information System
1. Introduction
Rough set theory proposed by Pawlak [17], is an extension of the classical set theory and could be
regarded as a mathematical and soft computing tool to handle imprecision, vagueness and uncertainty
in data analysis. This relatively new soft computing methodology has received great attention in recent
years, and its effectiveness has been confirmed successful applications in many science and engineering
fields, such as pattern recognition, data mining, image processing, medical diagnosis and so on. Rough
set theory is built on the basis of the classification mechanism, it is classified as the equivalence relation
in a specific universe, and the equivalence relation constitutes a partition of the universe. A concept,
or more precisely the extension of a concept, is represented by a subset of a universe of objects and is
approximated by a pair of definable concepts of a logic language. The main idea of rough set theory is the
use of a known knowledge in knowledge base to approximate the inaccurate and uncertain knowledge.
Due to the existence of uncertainty and complexity of particular problems, several extensions of the
rough set model have been proposed in terms of various requirements, such as the variable precision
rough set model [46, 43], rough set model based on tolerance relation [11], the Bayesian rough set model
[25], the decision-theoretic rough set model [37], the fuzzy rough set model and the rough fuzzy set
model [3] and many others investigations [19, 28]. In many circumstances, relations in ordered infor-
mation systems are not equivalence relations, but partial relations. Such as the dominance relation. It is
vital to propose an extension called the dominance-based rough set approach [26] to take account into
the ordering properties of criteria. The innovation is mainly based on substitution of the indiscernibility
relation by a dominance relation. Studies have been made on properties and algorithmic implementa-
tions of dominance-based rough set approach. In recent years, researchers have enriched the ordered
information system theories and obtained many achievements. For instance, Shao et al. further explored
an extension of the dominance relation in an inconsistent ordered information system [21]. Xu et al.
constructed a method of attribute reduction based on evidence theory in an ordered information system
[31], and others [2, 35].
Pawlak and its generalized rough sets are constructed based on one set of information granule, which
are induced from a partition or a covering. In 1985, Hobbs proposed the concept of granularity [8], and
Zadeh first explored the concept of granular computing [45] between 1996 and 1997. They all think that
information granules refer to pieces, classes, and groups into which complex information are divided in
accordance with the characteristics and processes of the understanding and decision-making. Currently,
granular computing is an emerging computing paradigm of information processing. It concerns the
processing of complex information entities called information granules [27]. Information granules, as
encountered in natural language, are implicit in their nature. To make them fully operational so that they
become effectively used in the analysis and design of intelligent systems, we need to make information
granules explicit. This is possible through a prudent formalization available within the realm of granular
computing. Pal et al. presented the relationship among granular computing, rough entropy and object
extraction [16]. Skowron et al. introduced basic notions related to granular computing on the information
granule syntax and semantics as well as the inclusion and closeness (similarity) relations of granules [24],
the foundations of rough-neural computing [23]. Yao first promoted the relationship between information
granulation and rough set approximation theory [38]. Peters et al. proposed an approach to measures
of information granules based on rough set theory [18]. In order to make rough set theory have a wider
range of applications, Qian et al. extended Pawlak’s single-granulation rough set to a multigranulation
rough set model [20]. And later, many researchers have extended the multigranulation rough sets. Xu et