Business Intelligence & Analytics
petak, 18. listopad 2019., 10:20
Modeling information as graphs is a natural and intuitive way for understanding complex relationships such as interactions in social networks or financial transactions between legal entities. In graph databases, entities are modeled as vertices in a graph structure, while relationships are modeled as edges between them, turning relationships into first-class objects in the graph store. This enables powerful analytics as well as sophisticated pattern matching capabilities.With the Spatial and Graph option to the database, Oracle offers an in-memory analytics engine based on the concepts of property graphs, which comes with a large number of graph algorithms out-of-the box and which also supports graph pattern matching. For pattern matching queries, a powerful, SQL-like query language, PGQL, is included to define the search patterns.In this session, we will explain the architecture of the solution and how the load is distributed between the persistence layer in the database and the in-memory analytics engine. We will look at how inherently schema-less data is stored in the relational database and how the resulting data structures can be used for graph traversal queries, computational analysis in PL/SQL, or pattern matching by means of SQL queries. With a new functionality in Oracle 18c, these pattern matching queries can be automatically generated from PGQL queries, allowing a very efficient and compact formulation of patterns. Finally, we will explain how the entire environment can be deployed on the Oracle Cloud.