Metamodeling Lightweight Data Compression Algorithms and its Application Scenarios

Lossless lightweight data compression is a very important optimization technique in various application domains like database systems, information retrieval or machine learning. Despite this importance, currently, there exists no comprehensive and non-technical abstraction. To overcome this issue, we have developed a systematic approach using metamodeling that focuses on the non-technical concepts of these algorithms. In this paper, we describe CLLATE, the metamodel we developed, and show that each algorithm can be described as a model conforming with CLLATE. Furthermore, we use CLLATE to specify a compression algorithm language CALA, so that lightweight data compression algorithms can be specified and modified in a descriptive and abstract way. Additionally, we present an approach to transform such descriptive algorithms into executable code. As we are going to show, our abstract and non-technical approach offers several advantages.