2018 4 th International Conference for Convergence in Technology (I2CT) 978-1-5386-5232-9/18/$31.00 ©2018 IEEE Defining Fuzzy Membership Functions for Fuzzy Data Warehouses PPG Dinesh Asanka Department of Computer Science and Engeenring University of Moratuwa Moratuwa, Srilanka dineshasanka@gmail.com Amal Shehan Perera Department of Computer Science and Engeenring University of Moratuwa Moratuwa, Srilanka shehan@cse.mrt.ac.lk Abstract— Data warehouse is used in many organizations as a tool to gain competitive advantage over their competitors. However, in most data warehouse implementations, lot of assumptions are made. By making assumptions, outcome of data warehouse is little far from the truth. Also, these assumptions will veracity of data warehouse is questionable. To avoid veracity of data warehouse, fuzzy logic can be incorporated. In fuzzy logic, fuzzy membership function plays a huge role hence in case of fuzzy data warehouse, fuzzy membership plays a key role as well. In data warehouse, different types of fuzzy membership functions can be introduced. arbitrary, data driven, linguistic, derived, survey-based membership functions are introduced in this research paper for different cases in data warehouse. Keywords— Data Warehousing, Fuzzy, Membership Function I. INTRODUCTION Data warehouse has become an important strategic component in information system of the fiercely competitive industry. Date warehouse is now spreaded to various sectors such as Agriculture [1] [2] [3] [4] [5] [6] [7] [8] [9] [10], Customer Relation Management (CRM) [11], Banking [12] [13], Healthcare etc. Though the data warehousing is not considered as an emerging technology, only recently that industry has adopted data warehousing into their business due to various issues. Over the years, analytics of data warehousing is limited to crisp value analysis. Though there were few attempts made to introduce veracity aspects of data challenge into the data warehousing, it is important to emphasis that end to end aspects of data warehousing is not fully considered due to various technical and practical limitations. However, there are attempts made to introduce veracity aspects to many sectors by using the fuzzy theory. Fuzzy logic is proposed to mitigate uncertainty in many domains such as agriculture [14], medicine [15] [16], power systems [17], production [18], sports [19], transportation [20] etc. In the field of data warehousing, for analytics purposes crisp values are used. For example, when there is a need to analyze some measures (assume sales) with the age of the customer. Customer age can be configured as nominal values such as low, middle and high. Depending on the domain and the situation, ranges for low, middle and high will be different. When the nominal labels are used, it is obvious that all the values in the range when the analysis was made from nominal values is not correct. To introduce veracity, fuzzy logic can be used. For example, 30 years of age can be considered as 0.3, 0.7 weightages are configured for medium and low respectively whereas in case of crisp set analyze, 30 years of age will be labeled as Medium and only Medium. By doing this, young or old contribution of the age is ignored. In this research paper, design strategies for data warehouse is introduced by looking at various aspects of data warehouses such as dimensions and fact tables and different scenarios. In this research paper, current research status of the fuzzy data warehouse is discussed in the State of Art – Fuzzy Data warehouse. Methodology is discussed in the following section while configuring of fuzzy membership is discussed in the next section. In the next two sections, fuzzy data warehouse design is discussed in detail. Finally, the conclusion and future work is discussed. II. STATE OF ART – FUZZY DATA WAREHOUSE As discussed in the introduction section, there are lot of domains which have discussed on usage of data warehouse. Since most of the research papers are published in recent years, it can be concluded that data warehousing is still a popular and used technology in the industry today. Also, fuzzy logic is also not a novel to industry as there are few fuzzy logic implementations are made as discussed in the previous section. In this section, current research status was identified with respect to all relevant areas. Literature review was divided into main two areas, fuzzy databases and fuzzy data warehouse design. Since many techniques of fuzzy databases can be utilized into the data warehouse, it was decided to analyses research areas of fuzzy databases. Unlike fuzzy data warehousing, there are few attempts made to design fuzzy databases in the relational and transactional databases. Research paper titled A Fuzzy Representation of Data for Relational Databases [21] it has suggested relational algebra operations consists of the same four parts as traditional relational algebra operation. To prove the concepts, this paper has come up with an implementation to selection of baseball team. In case of baseball team selection, there are expert places for each place. Some players are better