Application areas and usage of frequent subgraph mining algorithms

Graph mining is a subfield of data mining that represents the database as a graph. Graphs are powerful data structures used in various domains of data mining. In recent years, there has been a growing interest in graph mining. This can be attributed to the rise of social media, the digitization of medical research, and the increasing need for structured data in different disciplines. In structured data sets, the removal of repetitive infrastructure allows for interpretation and generalization of the data. Furthermore, users in a specific knowledge domain search for predefined types or domain-specific infrastructures. Graph databases reveal relationships between objects, where nodes represent objects and edges represent relationships between them. A graph database can consist of many small graphs or a single large graph. A database consisting of many small graphs is referred to as an operational graph database, while a database consisting of a single large graph is referred to as a single large graph database. Graph databases facilitate the representation of data from various disciplines such as chemical compounds, genetic structures, computer networks, and social media. Frequent subgraph mining is a data mining technique used to discover repeated patterns and significant relationships in large data sets. This technique employs specialized algorithms and methods to identify frequently occurring subgraphs in the data set. Frequent subgraphs are graphs that surpass a certain frequency threshold in the data set and carry specific patterns of meaning. Frequent subgraph mining has diverse application areas. For instance, it plays a crucial role in database management, social network analysis, bioinformatics, and graph analysis. This technique enables businesses to make better decisions by identifying recurring patterns in the data, optimizing marketing strategies, analyzing network structures, and revealing important information in biological data. Frequent subgraph mining can analyze terabytes of information and preserve data relationships even when they change and evolve. Therefore, graph databases provide an effective tool for frequent subgraph mining. The objective of this study is to explore how frequent subgraph mining can be applied, how frequent subgraphs can be discovered, and how these subgraphs can be valuable for organizations and researchers. The study consists of three stages. In the first stage, real-world data is transformed into a graph database. In the second stage, frequent subgraphs are discovered from the generated graph database. In the third stage, the informational significance of these frequent subgraphs is validated. Furthermore, a state-of-the-art frequent subgraph mining algorithm, GSPAN, specifically designed for operational graph databases, is applied to a sample case. The IMDb website is used as the sample case, specifically focusing on Turkish films released between 2000 and 2022 and their properties. The IMDb database is transformed into an operational graph database. The GSPAN algorithm (reference) is then executed on the operational graph database, resulting in the discovery of frequent subgraphs.