• Written by Azhar Aliyar, Director

A purchasing manager at a large retail company is deciding which supplier to use for a new product line, and currently bases this decision on a manual review of the purchase order table and the supplier performance records. A knowledge graph, which is a database that stores information as entities and their relationships, such as the relationship between a supplier and a product, can help with this decision. The knowledge graph stores these relationships in a way that allows for complex queries, enabling the purchasing manager to ask questions like "which suppliers have provided products with a high defect rate" and receive answers in real time.

Definition

A knowledge graph is a type of database that stores information as entities, such as suppliers, products, and employees, and their relationships, such as "supplier A provides product B". This allows for complex queries that can traverse these relationships, providing more detailed and relevant answers. For example, a shift supervisor can use a knowledge graph to find all the employees who have worked on a specific machine, by querying the relationships between employees, machines, and work shifts.

The knowledge graph is designed to handle the complexity of large datasets, and to provide answers that are relevant to the specific question being asked. This is in contrast to traditional databases, which often require manual querying and analysis to get the desired information.

Difference from plain retrieval

Most people get wrong the idea that a knowledge graph is just a fancy way of doing plain data retrieval. They are wrong. Plain retrieval is about finding specific data points, such as "what is the price of product A", whereas a knowledge graph is about finding relationships between entities, such as "which products are supplied by supplier A". The knowledge graph provides a more nuanced understanding of the data, by showing how different entities are connected.

For instance, a category manager can use a knowledge graph to compare two suppliers, by querying the relationships between the suppliers, products, and prices. This allows the category manager to see not just the price of each product, but also which products are supplied by each supplier, and how the prices compare.

Applications in a large organisation

A knowledge graph can be applied in various areas of a large organisation, such as supply chain management, employee management, and customer relationship management. For example, a knowledge graph can be used to store information about employees, such as their skills, work experience, and training, and their relationships with other employees, such as who they report to and who they work with. This allows for complex queries, such as "which employees have the skills to work on project A" and "who reports to manager B".

A knowledge graph can also be used to store information about customers, such as their purchase history, preferences, and relationships with other customers, such as who they have referred to the company. This allows for complex queries, such as "which customers have purchased product A and also referred a friend" and "who are the top referrers of new customers". To learn more about how a knowledge graph can be applied in your organisation, you can visit our website.

Limitations

A knowledge graph is not a replacement for traditional databases, but rather a complementary tool that can be used to provide more detailed and relevant answers. It is not suitable for all types of queries, such as simple data retrieval or transactions, and is best used for complex queries that require traversing relationships between entities. For example, a knowledge graph would not be the best choice for a simple query like "what is the price of product A", but would be well-suited for a query like "which suppliers provide products with a high defect rate".

In addition, building and maintaining a knowledge graph requires significant expertise and resources, and is not a project to be taken lightly. It requires careful planning, data integration, and ongoing maintenance to ensure that the graph remains accurate and up-to-date. If you are considering building a knowledge graph for your organisation, you can contact us to discuss your options.

When it is the wrong choice

There are certain situations where a knowledge graph is not the best choice, such as when the data is simple and does not require complex relationships, or when the queries are straightforward and do not require traversing relationships between entities. In these cases, a traditional database or a simpler data storage solution may be more suitable. For example, a small company with a simple product line and few suppliers may not need a knowledge graph to manage their data.

In general, a knowledge graph is best used when there are complex relationships between entities, and when the queries require traversing these relationships to get the desired answers. It is a powerful tool that can provide significant benefits, but it is not a one-size-fits-all solution, and should be carefully evaluated to determine if it is the right choice for your organisation.