Can Graph Databases Maximize The Value Of EHS Data?

What do web-based services such as Facebook, Amazon, Expedia and eBay have in common? They all have 'recommendations' features embedded in their database architecture, which creates relationships between data from past behaviour such as purchases and searched items. From statements such as 'You may also like...' to 'Customers who bought this item also looked at...,' one can see how this feature can drive sales. What if we could apply this feature to EHS data? Such use is already available to some extent in the form of predictive analytics. Although not entirely phrased in the same way, predictive analytics can make inferences based on trends such as human behaviour and even seemingly quotidian metrics such as weather data. Pair this with artificial intelligence (AI) and machine learning (ML), and you have a decision-making tool that can extract valuable insights from the ever-growing EHS data and associated data relationships.

EHS teams may find it challenging to generate valuable insights from the ever-growing volume and complexity of data from EHS mobile applications, IoT and industrial wearable devices. The issue is especially troublesome for wearables data, which may include location data, photos, videos and vital signs—which are not precisely suited for storage in the traditional tabular format. The Alliance for Internet of Things Innovation, a European Commission initiative launched to support the creation of an IoT ecosystem, estimates that there were over 14 billion connected devices by the end of 2019. This is a lot of data to contend with and will present a problem to firms, as they scale up their infrastructure to avoid being overwhelmed. A technology that may address the issue of digesting unstructured data and the scalability of database infrastructure is graph database technology.

So, what is a graph database? A graph database is a database that stores data in a form akin to hand-sketched mind maps on a whiteboard instead of in columns and rows. One of the main advantages of graph databases is that they intrinsically store the relationships among data points. Hence, making it easier to process nested data—data within multiple tables and sub-tables—which may require lengthy query codes in traditional SQL databases. Analogously, if conventional SQL databases are akin to files in folders stored in shelves, then graph databases are similar to files systematically laid out on a vast expanse of floor, with different coloured strings attached to each file. The strings represent the relationships among data points and pulling on the string of one file pulls all the other files associated with it. While this analogy is an oversimplification, nonetheless, it is a great way of imagining graph databases. Graph databases are what enable Facebook, Amazon, Expedia and eBay to make recommendations based on your past website behaviour.

Therefore, graph databases are excellent for storing vast amounts of complex EHS data, especially mixed data from wearable devices and apps and they work synergistically with AI and ML because of the relationships among data points. Using graph databases, EHS managers are able to identify insightful trends—some that they may not have expected—without having to make hypotheses about the causes of incidents. Additionally, new data stores can be 'bolted' onto the existing structure without having to overhaul the entire database structure as one would for traditional SQL databases. Finally, data processing is much quicker in comparison to conventional databases. Neo4j, developers of a graph database system, simulated a social network of one million individuals. In this simulation, the graph database was able to process as many as five levels of nested data in approximately two seconds, while a traditional database could not complete the operation.

EHS Can Graph Databases Maximize The Value Of EHS Data Verdantix Blog