Description: A method is proposed for calculating parameters of generalized association rules based on the parameters of simple association rules that are part of an aggregated association. Under the generalized associative rules we mean the logical dependencies between the attributes of objects in databases, where attributes can take values from some set. The apparatus of the algebra of finite predicates was used to describe such attributes. The level of support for a generalized cover is determined by summing the levels of support for simple covers. In order to find the level of confidence of a generalized rule, we decompose it into simple ones. We show that the decomposition of an aggregated association rule can be made both from right to left and vice versa. In contrast to the method of searching for generalized associations using the taxonomy of features, the proposed method does not require additional scanning of the database to calculate the characteristics of the rule, but allows calculating them analytically using the constructed tree of covers. Using the union of attribute values allows defining relationships that are included in higher levels of generalization, since the set support increases as the occurrence of the group of attribute values is counted. The main disadvantage of the method of combining attribute values using taxonomy is the fact that merging is possible only by moving to a higher level of hierarchy. This leads to less “interesting” (“utility”) rules, since in this case they belong to groups of attribute values. The method of finding generalized association rules by combining the branches of the cover tree avoids this disadvantage. The proposed method allows a more flexible approach to the construction of generalized associative rules. Association of attributes of objects can be carried out not for the whole group, but for some attribute values.
Keywords: associative rules, parameters of associative dependency, support, confidence, improvement
1. Srikant, R. and Agrawal, R. (1997), Mining generalized association rules, Future Generation Computer Systems, Vol. 13, Issues 2–3, pp. 161-180.
2. Sitnikov, D., Titova, E., Ryabov, O. and D’Cruz, B. (2005), A method for generating aggregated association between discrete data features, WIT Transactions on Information and Communication Technologies, Vol. 35, pp. 25-34.
3. Marchenko, O.O. and Rossada, T.V. (2017), “Aktualʹni problemy Data Mining” [Actual problems of Data Mining], KPI, Kyiv, 150 p.
4. Kornilkov, A.P. and Khabibulina, T.V. (2014), “O realizatsii poiska assotsiativnykh pravil sredstvami yazyka programmirovaniya PHP” [On the implementation for association rule mining in the programming language PHP], Modern technology and technology, No. 5, available at: www.technology.snauka.ru/2014/05/3659.
5. Fam, K.Kh. and Kvyatkovskaya, I.Yu. (2015), “Primeneniye assotsiativnykh pravil v informatsionno-analiticheskoy sisteme otsenki kachestva predostavleniya telekommunikatsionnykh uslug” [Application of associative rules in the information and analytical system for assessing the quality of telecommunications services], Scientific Bulletin of the Novosibirsk State Technical University, No. 2 (59), pp. 33-42.
6. Zheliznyak, I.Y. (2017), “Pravyla pobudovy asotsiatyvnykh pravyl na prykladi fizychnykh pokaznykiv patsiyenta” [Rules of construction of associative rules on the example of physical indicators of the patient], Scientific Bulletin of NLTU of Ukraine, Vol. 27, No. 9, pp. 107-110.
7. Ju, C., Bao, F., Xu, C. and Fu, X. (2015), A Novel Method of Interestingness Measures for Association Rules Mining Based on Profit, Discrete Dynamics in Nature and Society, Article ID 868634, 10 p.
8. Blanchard, J., Guillet, F., Gras, R. and Briand, H. (2005), Using information-theoretic measures to assess association rule interestingness, 5th IEEE International Conference on Data Mining ICDM’05, IEEE Computer Society, United States, pp. 66-73.
9. Prajapati, D.J., Sanjay, Garg and Chauhan, N.C. (2017), Interesting association rule mining with consistent and inconsistent rule detection from big sales data in distributed environment, Future Computing and Informatics Journal, Vol. 2, Issue 1, pp. 19-30.
10. Titova, O., Sitnikov, D., Ryabov, O. and Kovalenko, A. (2018), Assessment of extended aggregated association rules, The 9th IEEE International Conference on Dependable Systems, Services and Technologies, DESSERT’201, Kyiv, Ukraine.
11. Barsegyan, A.A. Kupriyanov, M.S., Kholod, I.I., Tess, M.D. and Yelizarov, S.I. (2009), “Analiz dannykh i protsessov” [Data and process analysis], BHV-Petersburg, Sankt-Petersburg, 512 p.