In an ever connecting world, where everything is linked to everything, enterprise data are still mostly stored in silos. And this is not only a technical limitation but rather an organizational mindset. Graphs help to break these silos, make information wider accessible and help gain new insights.
The most prominent example is the friend-of-a-friend concept. Social networks but also recommendation systems use this paradigm to propose people you may also know or products you may also like. Graphs provide additional semantics as well as wider views across organizations (business units, departments, business partners) along the value chains and across systems. This leads to significantly higher transparency and new insights for operations and advanced analytics.
A knowledge graph is a collection of knowledge for a certain domain. It can span across domains in enterprise knowledge graphs. The content is
collected by data integration architecture: Data is automatically loaded from enterprise sources, from web resources and by direct input through UIs.
refined by machine learning algorithms: ML algorithms can categorize, cluster and group as well as create semantic relationships of data.
curated and governed by subject matter experts: Domain experts define the content available in a knowledge graph and define access requirements for end users.
In graph theory for information technology, a graph is semantic representation of knowledge through nodes and edges. Nodes (or vertices) usually represent data points while edges connect them physically. These connections can be either directed (uni-/bidirectional) or non-directed.
In a graph the data points are physically connected via edges, which makes querying extremely fast. When querying through SQL, data is collected by comparing indexes across different tables. This becomes very slow with an increasing amount of data and their heterogeneity. For instance, looking to get a complete customer view for insurance, we need to collect data about a customer, contracts, offers, claims, about products, about relationships, … A graph is capable to collect these information faster than RDMBS via graph traversal, i.e. by checking each node in a graph that is connected via edges to the starting node. On top of that, graph data models are much more flexible since nodes with new labels (new data objects) or new properties can easily be added.
Various graph databases exist, with the most prominent being property graph databases and RDF (Resource Description Framework) stores. In enterprise architecture, property graphs are wider spread because they provide higher elasticity when it comes to data modelling.
A property graph consists of nodes and edges, each of which can be qualified with further properties. That allows a data model to be highly flexible, as properties can simply be added on demand and properties that are not valid for certain nodes are simply non-existent (instead of null).
Due to its ability to link heterogeneous data, disparate data sources are made easier accessible to unveil hidden connections between data. Complex relationships will emerge that provide deeper insights in ever-growing amounts of data. By relating big data sets to master data companies will create smarter data that will drive business transformations to new heights. Easier access to a smart data foundation will also support data scientist in their work. A lot of ground work for AI solutions can be made using graph ML algorithms.
The key value levers of graphs are as follow:
The strengths of the graph lie in the connection of heterogenous data.
Uncover fraud, e.g. in insurance, using real-time analytics across many different data sources and objects.
Inventorize your corporate assets using graph technology for added transparency and faster search, e.g. across claims, offers and contracts.
Create higher transparency in complex interdependencies central to managing networks and IT infrastructure.
Eliminate silos in master data and reference data and create better inter-connected insights across various data objects in real time. Enable 360° views of customers or complex supply chain hierarchies for materials.
Better understand social connections, infer new relationships and identify opinion leaders about your products or services.
Map users, assets, relationships and authorizations to create higher transparency in identify management.
Connect the dots for regulatory compliance (GDPR, BCBS 239, FRTB…). Manage your risk by connecting several data points and check against regulations.
Build the base for your AI applications and leverage ML capabilities of most graph products. Graph ML will increase your models even more.
We break down silos, set the right incentives, reduce complexity, and build lean processes to deliver customer value.
We help our customers shift from to define effective operating models to reduce costs and to improve the experience of both clients and employees.
A siloed and technically focused workforce needs to transition to an interconnected, collaborative and customer– centric operating model.
Augmenting the workforce through intelligent automation and data-driven insights are key levers to improve the customer experience and improving the operational effectiveness.
We have developed our own solutions to help you drive automation and data-driven insights along your value chain.
Machine Learning is a very potent approach to solving tough problems covering areas such as smart automation and predictive complex forecasting. We offer consultancy service to help you build machine learning capabilities. Additionally, our suite of machine learning products helps you to accelerate your adoption journey.
We shape your Machine Learning vision and identify opportunities along your value chain as input for your transformation roadmap. Our focus lies on improving your customer focus and operational excellence. Additionally, we help you explore disruptive new products and services.
Our product line includes cutting-edge data analytics and machine learning solutions to help our customers automate unstructured data extraction and take the right decisions.