Currently, customers do analyses of various market reports on their own or hire experts to make procurement decisions. These experts analyze reports captured from data sources, a process that can be time-consuming and prone to human error. However, by using our intelligent procurement system, this process can become faster and more accurate.

This process involves building and integrating a Watson Knowledge Studio (WKS) model into Watson Discovery. The data extracted from Watson Discovery is fed into a knowledge graph that you can then query to extract relevant info. I’ll quickly go through the steps of this system here, but you can get more details from the GitHub repository.

Entity types

An entity type is how you categorize a real-world thing. An entity mention is an example of a thing of that type.

Relation types

A relation type defines a binary, ordered relationship between two entities. For example, in the following image, SupplyforPeriod is between the Supply and Time_Period entities. It indicates what the supply is for the specified time period. This can be used as search criteria in a query.

To create machine learning model in WKS, you must follow the process shown in the following image. The preannotation steps to bootstrap the human annotations are explained in the ML in Procurement – Bootstrapping Annotation process article.

After the human annotations are submitted and accepted by the reviewer, you can create a model and deploy it to an existing Watson Discovery instance.

Deploying WKS to Watson Discovery

Results with the WKS model ID configured

Methyl Methacrylate (MMA) is an important chemical used in the paint industry. Here, MMA is captured as Product_of_Interest entity info. When a chemical plant that produces MMA is shut down, it is captured as Facility_Production_Down entity info. The following two images show this use case, and this is possible because of a trained WKS model for this domain.

Results without the WKS model ID configured

In this example, MMA is captured as Organization information instead of Product_of_Interest. This is because there is no domain knowledge (WKS model) integrated into Watson Discovery.

Summary

This is a quick overview of our intelligent procurement system, based on Watson Discovery, which allows a customer to receive expert analyses quicker and more accurately. For a more detailed explanation, see the associated code pattern.

Register for IBM Cloud to access Watson Knowledge Studio, Watson Discovery, and other IBM Watson services.

Take a look at the IBM Cloud docs for more information about IBM Cloud Services.

The IBM Code site offers code patterns to help you get started quickly.

Join The Discussion

Your email address will not be published. Required fields are marked *