Distributed Action-Rule Discovery Based on Attribute Correlation and Vertical Data Partitioning
-
Published:2024-02-03
Issue:3
Volume:14
Page:1270
-
ISSN:2076-3417
-
Container-title:Applied Sciences
-
language:en
-
Short-container-title:Applied Sciences
Author:
Benedict Aileen C.1ORCID, Ras Zbigniew W.12ORCID
Affiliation:
1. Computer Science Department, University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223, USA 2. Polish-Japanese Academy of Information Technology, Institute of Computer Science, 02-008 Warsaw, Poland
Abstract
The paper concerns the problem of action-rule extraction when datasets are large. Such rules can be used to construct a knowledge base in a recommendation system. One of the popular approaches to construct action rules in such cases is to partition the dataset horizontally (personalization) and vertically. Different clustering strategies can be used for this purpose. Action rules extracted from vertical clusters can be combined and used as knowledge discovered from the horizontal clusters of the initial dataset. The number of extracted rules strongly depends on the methods used to complete that task. In this study, we chose a software package called SCARI recently developed by Sikora and his colleagues. It follows a rule-based strategy for action-rule extraction that requires prior extraction of classification rules and generates a relatively small number of rules in comparison to object-based strategies, which discover action rules directly from datasets. Correlation between attributes was used to cluster them. We used an agglomerative strategy to cluster attributes of a dataset and present the results by using a dendrogram. Each level of the dendrogram shows a vertical partition schema for the initial dataset. From all partitions, for each level, action rules are extracted and then concatenated. Their precision, the lightness, and the number of rules are presented and compared. Lightness shows how many action rules can be applied on average for each tuple in a dataset.
Subject
Fluid Flow and Transfer Processes,Computer Science Applications,Process Chemistry and Technology,General Engineering,Instrumentation,General Materials Science
Reference41 articles.
1. Kuang, J., Daniel, A., Johnston, J., and Raś, Z.W. (2014, January 9–13). Hierarchically structured recommender system for improving nps of a company. Proceedings of the Rough Sets and Current Trends in Computing: 9th International Conference, RSCTC 2014, Granada and Madrid, Spain. Proceedings 9. 2. Tarnowska, K., and Ras, Z. (2021). NLP-based customer loyalty improvement recommender system (CLIRS2). Big Data Cogn. Comput., 5. 3. Developing customer attrition management system: Discovering action rules for making recommendations to retain customers;Duan;Appl. Intell.,2023 4. Al-Mardini, M., Hajja, A., Clover, L., Olaleye, D., Park, Y., Paulson, J., and Xiao, Y. (2016, January 13–16). Reduction of hospital readmissions through clustering based actionable knowledge mining. Proceedings of the 2016 IEEE/WIC/ACM International Conference on Web Intelligence (WI), Omaha, NE, USA. 5. Extraction of actionable knowledge to reduce hospital readmissions through patients personalization;Mardini;Inf. Sci.,2019
|
|