Product Catalogue

The main types of master data for industrial enterprises are product catalogs of manufactured products, spare parts, materials and other types of resources. This data is used in various enterprise systems: ERP, procurement portal, MES, PLM, CMMS, EAM, accounting systems, etc. When integrating these systems, it is necessary to establish a correspondence between the objects of the product catalogs used in them. This is best done by building a single corporate product catalog in the MDM system.

It is also possible to exchange such data with external organizations: for example, when transferring technical specifications and RFPs for purchased products to suppliers, or when processing received technical and commercial proposals. Since each supplier can use their own product catalog, different from the one adopted by the company (if the enterprise already has one), the task of comparing product list items between external sources and the company's own catalog arises.

In both cases, the main task is to automate the comparison of product items between different catalogs, which can contain hundreds of thousands of product names. To solve this problem, you can use the tool for extracting semantic characteristics from text, which is part of the EKG Platform. This tool converts text strings like "Engine 3200 kW 500 rpm SDN3-1776" into structured sets of characteristics: power = 3200 kW, rotation speed 500 rpm, class "Engine", model SDN3-1776. Having received product characteristics from different sources, it is easy to set up the process of automatic comparison of items with manual confirmation if the degree of coincidence is below the specified level.

EKG Platform allows you to create hierarchical equipment catalogs, specify a set of attributes for each class, and use catalogs of units of measurement of physical quantities. Each product can be included in several classes according to one or more hierarchies; the attribute composition of product properties is inherited from all classes to which it belongs.

The functions of cleaning (normalization), checking (validation), and combining (consolidation) of data will help to build a single, reference product catalogue, which can be used to exchange data between all information systems of the enterprise. EKG Platform allows you to implement a centralized process of entering and coordinating elements of the product catalogue, synchronizing the data of this directory with all applications.

See also: Structuring and cleaning product catalogs with DataVera EKG Platform.

Assets: technical locations and equipment units

Another frequently encountered task of industrial enterprises is the creation of a single set of data on technical locations and equipment units. This task may arise when creating an asset management system, when implementing maintenance and repair management (CMMS) systems, resource management, etc. A single asset catalog is also necessary for creating various monitoring systems, an enterprise situation center. Since asset data is important for many systems, it is advisable to build their catalog in the MDM system and synchronize it with all application solutions. On the MDM side, you can implement the process of making changes to the asset structure and distributing the agreed changes to all applications.

Asset data must necessarily include their structural and functional relationships. Structural links describe the relative positions of assets: for example, unit X is located in workshop Y, pump P is part of unit X, sensor T is installed on pump P. Positional codes of technical locations are often used, which reflect the position of each technical location in the structural hierarchy of enterprise assets - such codes can be automatically generated and verified in MDM. Functional links describe the mutual dependencies of technical locations that arise during enterprise operation: transformer T1 supplies electricity to unit X, the working fluid from unit X enters tank R. Describing such links helps solve problems of predicting failure chains, assessing the impact of incidents, building optimization models to determine optimal equipment operating modes, and saving energy and resources.

Information about equipment units includes inventory numbers, passport and operating values ​​of technical characteristics of specific devices, service life, technical condition indices. It is mandatory to store the history of installation of equipment units in technical locations, information on passing inspections and examinations, performing scheduled and emergency repairs. This data can also be copied from the CMMS to the MDM system, for example, to make it available for use in data analysis and machine learning tools.

For many industries, there are reference information models, such as the CIM model for electric power enterprises (IEC 61970, IEC 61968 and other standards). The MDM system built on the EKG Platform makes it easy to import and use such models, as it implements the ontological modeling specifications on which these standards are based.

What industrial enterprise data should be stored in MDM?

Any industrial or mining enterprise is a huge financial and economic complex with a large volume and variety of data. The ability to solve optimization problems, use artificial intelligence tools and organize data-driven management depends on how synchronized these data are between different sources and how much they can be trusted in terms of quality. The higher the level of enterprise management tasks that need to be solved based on data, the higher the requirements for the composition, quality and degree of integration of this data. An MDM system of an industrial enterprise may contain:

  • Catalogues of the products, items, materials
  • Information on assets – technical locations and units of equipment
  • Information on the enterprise's counterparties – its suppliers, dealers, consumers
  • Organizational structure and personnel information.

Product Hierarchies and Properties

First and foremost, a product catalog is a hierarchy. Our platform allows for maximum flexibility in constructing a class tree and managing the attribute model. New classes and attributes can be added during system operation, and class nesting can be modified. Thanks to the ontological foundation of our product, each class can have multiple parents and inherit a set of properties from them. Each specific product can be included in multiple classes at once. This enables the use of faceted classification: creating multiple parallel class hierarchies built on different foundations. For example, one hierarchy might describe a product's affiliation with a structural or functional component of a specific product, while another hierarchy describes how that product is used in the company's business processes. A product will receive a set of properties inherited from all the classes it belongs to.

In the age of artificial intelligence, the tasks of creating hierarchies, classifying products, and assigning property values ​​can be automated.

Let's consider the example of a small product catalog for a company that sells tires, wheels, and batteries. Products in different categories have different sets of properties. The values ​​for these properties can be obtained using AI when importing the product list, obtained from external sources through data enrichment, or entered manually. If the value of a property is obtained by the system from an external source, such as found online, such a value is highlighted in green. Hovering over it displays a link to the source.

Property Generation Using Templates

The system can not only extract property values ​​from product names but also, conversely, generate a reference product name based on property values. We can define templates for generating values ​​for some fields on the "Template" tab of the class page. These templates can use not only values ​​from other object properties, but also constants and values ​​from auto-incrementing sequences. Once a template is created, changing any property of a product in this class will automatically update the generated value.

Units and Languages

Properties often reflect the value of a physical quantity. For example, maximum tire pressure can be expressed in kilopascals, bars, or PSI (pounds-force per square inch). Our system uses the QUDT model, which defines physical quantities and conversion factors between them. If we select kilopascals as the primary unit of measurement for the "Maximum Pressure" property, the value will always be stored in this unit in the database. However, in the interface, we can select any other unit applicable to our value. The system will automatically convert one unit to another when viewing a product card, saving it, or setting a filter.

Text property values, such as name or description, can be specified in multiple languages. The system allows you to specify values ​​for each property of a given object in any language defined in the system. These values ​​can then be used when generating documents in a specific language.

Approval Routes

Our system allows you to separate user access to products in different classes. For example, a user can edit products in one catalog branch and only view others. Sometimes it is necessary to implement a process for approving the creation of new product cards or editing existing ones in the MDM system. Users of some groups can be granted the rights to create requests for catalog changes, while others can be granted the right to approve these changes. Approval can occur in several stages, with changes approved sequentially or simultaneously by users of different groups. The initiating user can create a request to add a new product card, which will be highlighted in the list. The approving user sees an indicator showing the number of cards requiring approval in a specific section or in the entire catalog. Clicking the indicator will display only the cards that need to be reviewed. The approving user can accept, reject, or accept the request, or make their own changes and confirm. If the user is the last in the approval flow, the card is given full object status and becomes accessible for use in a product catalog.

Duplicate Search

A typical problem with product catalogs is the presence of duplicates. Users often mistakenly create multiple cards for the same product. Duplicates can also appear when merging different product databases. Our system has a duplicate search mechanism based on strict criteria, such as matching article numbers. These groups of duplicates are then converted into a single reference record, which can be passed back to source systems or to an analytics display. However, there are also situations when duplicates must be identified by imprecise criteria. For example, several mismatched characters in a product name could be a difference in labeling or a typo. In such cases, a human must decide whether to merge such cards. To find candidates for merging, we use fuzzy duplicate search rules that allow a specified degree of matching in key field values, such as names or SKUs. The catalog operator can review the identified candidates and decide in each case whether these objects are valid duplicates. If so, they should be linked with an "Equivalent" relationship to form a single reference object.

Combined with other features of our platform, such as automatic value normalization, data quality control, generation of reference objects using customizable logic, maintaining a history of all changes and the ability to restore the state of objects to any date, data reception from source systems using customizable ETL rules, and distribution of reference objects in synchronous or asynchronous mode, the DataVera MDM platform provides the full functionality necessary for creating master catalogs of goods and services for retail and industrial enterprises.

If your enterprise has tasks for integrating and processing these types of data, we will be happy to provide consulting services and implement an MDM system. The creation of a single, reference set of information of the listed types will allow the implementation of rational management practices based on decision support based on data analysis.