Leveraging Knowledge Graphs to Enrich Machine Learning
By combining knowledge graphs and machine learning, organizations can extend the capabilities of ML and ensure that the results derived from their models have strong explainability and reliability.
Current applications of machine learning (ML) are widespread: deciding which trades to execute on Wall Street, determining credit decisions, optimizing inventory, improving product recommendations, predicting whether a user will click on an ad or Google’s ability to improve data center cooling efficiency. And that only scratches the surface. At the heart of what makes ML possible is vast amounts of data, which means that if companies don’t have access to or don’t fully understand the relationships between the data in their data assets, they will miss opportunities. . Knowledge graphs can help.
Why? Companies that strive to define and implement machine learning at the same time find that the easiest part is to implement the algorithms used to make machines intelligent on a data set or problem. So what’s the hardest part? Here’s a hint: data is becoming the key differentiator in the machine learning race. Businesses are striving to transform to stay digitally competitive, and the stakes couldn’t be higher.
Therefore, organizations are turning to knowledge graph technologies to improve data search, information retrieval, and recommendations. By combining knowledge graphs with machine learning, organizations can make machine learning more ubiquitous and effective. According Gartner Research, 23% of organizations have deployed graph techniques in their artificial intelligence (AI) projects. Perhaps other organizations don’t realize that they can combine knowledge graphs with machine learning using platforms that are easy to adopt and scale. This makes machine learning more mainstream and efficient.
See also: How datagraphs provide a competitive advantage
Knowledge graphs and their relationship to machine learning
Knowledge graphs connect and contextualize disparate data. Designed to capture dynamics, knowledge graphs easily accept new data, datasets, definitions, and requirements. As each part of the organization elevates its information as facts in the knowledge graph, more information is acquired and more value is realized. The knowledge graph is often positioned as the semantic data layer within the business data layer. The “semantic” part is the ability to describe the meaning of entities and relationships, whether in simple taxonomies or in-depth ontologies that capture the meaning of data. As a layer in an enterprise architecture, knowledge provides a secure suite of endpoints for consuming the knowledge graph in an ecosystem of off-the-shelf commercial software, analyst tools, data science tools, and more.
Machine learning and artificial intelligence (AI) have burst onto the scene over the past decade, fueled by stories ranging from self-driving cars to top ISPs with uncannily accurate recommendation engines on what that we would like to buy next. The story of the ML process begins and ends with data, and the challenge is clear: Companies that can acquire clean, connected data to train ML models and then use those additional facts will dominate the next decade of technology solutions in nearly all sectors.
How does a knowledge graph meet this challenge? The answer comes down to data. From the start, the data scientist and ML solutions need high-quality information that is correctly interpreted so that it can be used in feature engineering, trained in models, and analyzed in results. Many teams today do it by hand, what they call data wrestling and spend over 70-80% of the developer’s time reworking and reshaping the data. The Knowledge Graph provides connected, aligned and harmonized terminology across all areas of the business and configured for easy consumption in all kinds of ML tools and software systems. The result is far fewer hours spent searching for data, cleaning it, or reshaping it for the ML process. Additionally, metadata about those features and what happened with that data can all be captured in the knowledge graph. This provides a pedigree to the likely facts, predictions, or classifications produced by running the ML models.
The second major area is what happens to the results of ML runs? Traditionally, these values can be used to provide a specific report, feed a particular dashboard, or provide a feedback cycle to additional ML routines. The knowledge graph offers another option: capture these likely facts, predictions and recommendations, weights and scores, and other generated information and augment the graph. ML outputs are not only tagged or labeled, but aligned with the business model they relate to, and specifically linked to the key entities and relationships referenced in the model output.
This ML-to-knowledge graph process can be done in the form of enriching existing knowledge with new information learned by the ML process or by combining multiple ML outputs into a graph for analysis and further development. For example, take an ML model that predicts next month’s manufacturing output and another ML model that predicts supply chain bottlenecks in major ports. These two outputs can be combined into the Enterprise Knowledge Graph which would typically contain Customer 360, Product 360, and other cross-sectional graphs along with logic models and constraints. Such a system would allow the business to make more agile and informed decisions beyond just having the ML reports or the knowledge graph.
Because knowledge graphs provide domain and context information in a machine-readable format, organizations can integrate them into explainable ML approaches, providing more reliable explanations. By combining knowledge graphs and machine learning, organizations can extend the capabilities of machine learning and ensure that the output derived from machine learning models has strong explainability and reliability.
