Location Intelligence in a Big Data Environment

Location information is a common element in Big Data scenarios, whether it is explicitly expressed as coordinates from a GPS sensor or implicitly included in a postal code, a city or a landmark name. It can be used to track assets or categorize entities based on proximity but more importantly it can be used link otherwise unrelated data sets. However, for these purposes geospatial processing capabilities are required on Big Data platforms. While spatial data types, spatial indexes and the corresponding operators are well-established in conventional databases, their support to date was rather limited in Big Data environments. The presentation will cover the fundamentals of spatial processing using MapReduce technologies and architectural concepts for their integration with object-relational databases. We will include examples of how to use geo-enrichment and map visualization to gain insight into huge, variable datasets characteristic of Big Data scenarios.

Using Graph Analytics on Oracle NoSQL to Predict Customer Behaviour

With the new “Oracle Big Data Spatial and Graph” offering on the Oracle NoSQL database or Apache HBase Oracle now provides an optimized, schema-less data storage model for so called property graphs. The product includes an in-memory analytics engine with a suite of 35 parallel-enabled algorithms as well as a set of developer APIs to access the graph data and perform graph operations. 
In our presentation we will look at the fundamentals of graph databases, the architectural concepts of Oracle’s property graph implementation as well as the basic mode of operation. This will cover the end-to-end process from construction of graph data to navigation across the graph to graph analysis. A simplified example will be used to illustrate the identification of key influencers in a customer database who, in the case of customer churn in a telecommunication company, are likely to have an impact on additional customers. Finally, a few options for data visualization will be presented.