A Novel Pattern Storage System to Preserve Patterns in a Pattern Database

Abstract

Various data mining techniques nowadays can generate results popularly known as patterns from large data repositories. However, there is no facility or infrastructure, or model to preserve it as persistent storage of patterns. The proposed storage system provides a model to make these patterns persistent. Most organizations are interested in knowledge or patterns rather than raw data or many unprocessed data because extracted knowledge plays a vital role in making the right decision for the organization's growth. In this paper, Frequent Pattern Mining (FPM) algorithms are used to extract large data set patterns. Result comparison was done using Apriori, Fp-Growth, and Eclat. The proposed model of the Pattern Storage System uses a pattern database to make all patterns persistent. Pattern Storage System (PSS) is used to store FPM algorithms' results in the MongoDB NoSQL database. The work proposed in this paper reduces unnecessary processing on row datasets for which patterns are already available. Pattern retrieval comes very easy with minimum time compared to traditional ways of finding patterns. The proposed model uses a unique way of generation of pattern id which is useful for pattern storage as well as pattern retrieval.