Download
1 / 14
140 likes | 465 Views
FUZZY SQL FOR STATISTICAL DATABASES. Miroslav Hu d ec INFOSTAT – Bratislava MSIS 200 8. Introduction. Classical SQL and its disadvantages F uzzy improvement Generali sed “ where ” clause Case study for statistical database Conclusion. Classical SQL. select n, a 1 , a 2 ,…a n from T
E N D
FUZZY SQL FOR STATISTICAL DATABASES Miroslav Hudec INFOSTAT – Bratislava MSIS 2008
Introduction • Classical SQL and its disadvantages • Fuzzyimprovement • Generalised“where”clause • Case studyfor statistical database • Conclusion
Classical SQL select n, a1, a2,…an from T where a1>A1 and a2<A2
Classical SQL select n, a1, a2,…an from T where a1>A1-p and a2<A2+q
Fuzzy improvement of the SQL • Accesses relational databases in the unchanged structure • Supports queries based on linguistic expressions on the client side The query is modified as follows: select n, a1, a2,…an from T where a1 is Bigand a2 is Small
Fuzzy sets for queries Big (greather than)Small (less than) Middle (equal,…) ___________ ______________________ The query: The query:The query: select n, a1, a2,…anselect n, a1, a2,…anselect n, a1, a2,…an from T from T from T where a1:>=Ld where a1:<=Lgwhere a1>=Ld and a1<=Lg
Generalised logical condition WHERE clause with fuzzy conditions only: where n denotes number of attributes with fuzzy constraints in a WHERE clause of a query, where and and or are fuzzy logical operators where ai is a database attribute and L is the parameter of a fuzzy set WHERE clause with fuzzy and classical constraints [and/or][atribute_m LIKE “*String”][and/or] [atribute_l<Date]
Calculation of the QCI The QCI values for selected records are calculated in next two steps: 1. Calculation of memebership degree to fuzzy sets 2. Calculation of query satisfaction: - for logical And operator min: i=1,...,n - for logical Or operator max: i=1,...,n
Example selectdistrict, unemployment, area fromT whereunemployment is Big and area is Small. Unemployment is described with„Big value“ fuzzy set andits parameters are: Ld=8% and Lp=10%.Areais represented with „Small value“ fuzzy set with parameters Lp=300km2and Lg=650km2. The query has this form: selectdistrict, unemployment, area fromT whereunemployment >8 and area <650
Conclusion This fuzzy approach supports work with linguistic expressions on the client side, nevertheless it does not change structure and processes onthe server side of relational databases. Fuzzy improving of SQL queries has advantages in cases when the user can not unambiguously separate data he is interested in from data he is not interested in by sharp boundaries or when the user wants to obtain data that are very close to satisfy queries.In other cases classical SQL fulfils the requirements for data. In further use the meaning of the query is not changed only shapes and boundaries of linguistic expressions are changed to catch new requests. The state of art of this approach depends also on the theoretical and practical development of fuzzy database management systems.
