In our data-driven world, businesses are facing an overflow of content and its related information. And making sense of it all is crucial. One powerful tool in this data management toolkit is creating a solid metadata model. This article is based on one session of the course “Metadata and Taxonomies” with Madi Weland Solomon and should give you an overview over the basics of crafting a good metadata model. 

So, in the following paragraphs you will learn why this is so essential, what it involves, and how to do it right. But why? “With the onset of AI products that can quickly build data blocks through entity extraction, the one thing technology cannot do is build coalitions and nurture teamwork. Inspiring and building stakeholder buy-in to create a shared data vision that delivers data integrity and trusted LLMs requires commitment across divisions. Keeping humans in-the-loop throughout this journey is essential in realising data capabilities that can enhance business imperatives. The process starts with a metadata model,” so Madi Weland Solomon. 

As we navigate the maze of modern data, a well-designed metadata model not only makes data easy to find and understand but also lays the foundation for smarter decision-making. Join us on a simplified journey to discover how building a strong metadata model in the ideal case can turn your data challenges into opportunities for success.

Metadata: a Short Recap

But before we dive into the world of metadata models, it is important to recap what metadata is and why it is so important. In our previous article “Key aspects of metadata modeling for organizing and managing content” you can learn about it in detail or refresh your knowledge about it.

In short: Metadata is “data about data” and it asks questions. You could also think about it as the “who, what, when, where, and why” for your data. It can have various roles but in general it is a perspective, narrative and business domain. It serves as the framework that describes content for the following points:

• search,
• retrieval,
• discovery and
• automation.

To sum up metadata is crucial because it adds meaning and organization to the vast amounts of information organizations handle daily – as you may have already recognized.

Metadata Model: the Essentials

Now that we remember what metadata is, we move on to the next question – what is a metadata model? A metadata model, such as a content metadata model, serves as a structured framework documenting the various types of content within a specific project. It goes beyond just listing content types, providing detailed definitions of each element and outlining their relationships.

This model plays a pivotal role in fostering collaboration and coherence among different disciplines involved in a project. By serving as a shared reference point, the content model brings clarity to requirements, influencing the work of designers, developers crafting the content management system (CMS), and content creators alike. It acts as a bridge, facilitating communication and ensuring a common understanding of how different pieces of content relate to each other. In essence, a well-defined metadata model is not just a documentation tool; it is a catalyst for effective teamwork and streamlined processes in content development projects. You can also think of it as the roadmap for your information highway, providing clarity and organization.

A Guide: Step by Step to your Model

As promised in the title of this article we will provide you with a guide on creating a metadata model. In crafting a robust metadata model a structured approach is essential to ensure effectiveness and alignment with organizational goals. This process involves four key steps, each contributing to the model’s comprehensiveness and adaptability.

Step 1: Create a Draft Model

Understanding the Assembly Model: The initial step is to comprehensively collect and analyze how content creators put individual content items together to form diverse content products such as webpages, campaigns and documents. This requires engaging with stakeholders to capture their needs and expectations.

Identifying Content Types: Once the assembly model is clear, identify the distinct content types within the system. These content types represent various configurations of content that are unique enough to warrant differentiation.

Establishing Controlled Vocabularies: Define controlled vocabularies to ensure consistency and accuracy across content types.

Configuring Content Attributes: Define and configure content and metadata elements that constitute each content type. 

Utilizing a Metadata Workbook: Document and organize the metadata attributes for each content type systematically using a Metadata Workbook that serves as a central reference point for the entire metadata creation process.

Step 2: Synchronizing Metadata

Synchronizing Across the Organization: Achieve consistency and coherence by synchronizing metadata across the entire organization with the help of tools such as metadata crosswalks.

Step 3: Understanding Processes and People

Recognizing the Human Factor: Acknowledge that successful metadata creation is a result of a blend of people, processes and technology.

Step 4: Continuous Improvement

Adapting to Real-World Challenges: Recognize that the real-world application of the metadata model will uncover challenges and opportunities for improvement that require an iterative process.

To explain some of the important aspects of the “Step to Step Guide” in more detail, further explanations of processes that are essential for the model will follow in the next paragraphs.

Fostering Metadata Excellence: Harnessing the Power of Controlled Vocabularies

At the heart of a metadata model lies the concept of controlled vocabulary which can be described as an arrangement of words and phrases designed to index and retrieve content systematically. All industries and companies that collect and use data can benefit from a metadata model and from the uniformity that controlled vocabularies offer. The journey of establishing controlled vocabulary begins with the identification of content types. 

Controlled vocabularies can be, for example, look-up lists, drop-down menus, taxonomies, or selection dialogs. They serve as building blocks for basic metadata, where questions guide the formulation of answers using carefully curated vocabularies. So, when creating metadata, just think about the question and figure out the vocabularies you would need to answer.

Optimizing Metadata Organization: Functionality through Categories and Attributes

Organizing metadata elements into categories further refines the model and makes it easier to manage and retrieve relevant information. Examples for metadata elements are product information, technical details, usage information or customer data.

Metadata models often contain their own metadata by way of attributes. These attributes play a crucial role within metadata models, often serving as metadata for the metadata. Elements such as ID, descriptive names, character lengths, and obligations like mandatory or optional contribute to the model.

Ensuring Consistency: Synchronizing Metadata across Organizations

Synchronizing metadata across an organization is a pivotal task and identifying common descriptors, even when names differ, is essential. Metadata may originate in a third-party system, and it is important to identify the source of data and make it coherent throughout the organization. Therefore, the process involves tracing the origins of metadata from systems such as Enterprise Resource Planning (short: ERP) or Customer-Relationship-Management (short: CRM).

Metadata Symbiosis: Harmonizing with Crosswalks and Standardized Vocabularies

Metadata crosswalks also emerge as a valuable tool for synchronizing data among different systems and for identifying essential metadata. By gathering metadata from various sources and standards, organizations gain insights into commonalities, discrepancies, and opportunities for harmonization. This strategic approach reveals how the “same” metadata is described in diverse information systems.

To keep things consistent in a metadata model, it is helpful to rely on existing definitions and vocabularies. Take Dublin Core, for instance – a commonly used metadata schema for describing digital resources. It is like a set of straightforward and standardized rules for explaining documents and objects on the Internet. To make it clear in the metadata model, the elements are being labeled in addition with “dc:”, for example dc: Content Type, dc: Title, or dc: Date.

Ongoing Metadata Evolution: The Need for Continuous Improvements

When it comes to metadata modeling think of it like a mosaic you keep improving. The first design might not work perfectly with real content but that is okay. It is about getting the model ready for whatever comes next, especially in the beginning, because later it becomes harder and more expensive. 

Conclusion

Metadata, described as “data about data”, adds meaning and organization to information, serving as a framework for search, retrieval, discovery, and automation. Essential for documenting content types, relationships and attributes as well as for managing the overwhelming amount of data in our data-driven world is the metadata model. It includes aspects like draft model creation, synchronizing metadata, understanding the human factor and continuous improvement. Particularly important in this process are aspects like controlled vocabularies, organizing metadata elements, consistency through synchronization across organizations and the evolution of metadata models through continuous improvement. So start building your metadata model today for streamlined processes and structured, well-organized data.

Madi Weland Solomon teaches the course on metadata and taxonomy at FH JOANNEUM. Photo: Christopher Andreou

Madi Weland Solomon is Head of Client Solutions & Services at Graphifi, a knowledge graphs, semantics and data management company. She is a creative technologist specializing in DAM, enterprise taxonomy and ontology development, information architecture, and digital strategy. She has held executive roles in large multinational companies and has initiated and led business transformation programs from the ground up. She is a lecturer of the course "Metadata and Taxonomy" at FH JOANNEUM in Austria, and teaches the Data Maturity for DAM course for the Henry Stewart Online Education Series. She is also Managing Director and owner of Dots the Camden Music Shop.

Where to go from here

Key Aspects of Metadata Modelling for Organising and Managing Content

References

This article is based on the course "Metadata and Taxonomies" at FH JOANNEUM in 2023.

New Business Model. Available online at https://www.linkedin.com/pulse/whos-watching-you-metadata-new-business-model-madi-weland-solomon/.

Stewart, H. (2022). Essential Metadata Mapping. Available online at https://www.youtube.com/watch?v=RLsmpdg0bjg.

Cherryleaf. (2019). Podcast 74: What is a taxonomy? Interview with Madi Solomon. Available online at https://www.cherryleaf.com/2019/12/podcast-74-what-is-a-taxonomy-interview-with-madi-solomon/.