The Anatomy of AI

Here is a conceptual analogy to help illustrate how taxonomies and ontologies support AI. While inspired by the human musculoskeletal system, this analogy is not intended to represent anatomical accuracy, but rather to illustrate how taxonomies provide foundational structure and ontologies enable flexible, contextual connections of knowledge assets within AI systems.

Just like the musculoskeletal system gives structure, support, and coherence to the human body, taxonomies and ontologies provide the structural framework that organizes and contextualizes knowledge assets for AI. Here is the analogy: the spine and the bones represent the taxonomies, in other words, the hierarchical, backbone structure for categorizing and organizing concepts that describe an organization’s core knowledge assets. Similarly, the joints, ligaments, and muscles represent the ontologies that provide the flexibility to connect related concepts across assets in an organization’s knowledge domain. 

Depending on the organization’s domain or industry, certain types of knowledge assets become more relevant or strategically important. In the case of a healthcare organization, key knowledge assets may include content such as patients’ electronic health records, clinical guidelines and protocols, multidisciplinary case reviews, and research publications, as well as data such as diagnostic data and clinical trial data. Taxonomies that capture and group together key concepts, such as illnesses, symptoms, treatments, outcomes, medicines, clinical specialties can be used to tag and structure these assets. Continuing with the same scenario, an ontology in a healthcare organization can incorporate those key concepts (entities) from the taxonomy, along with their properties and relationships, to enable alignment and consistent interpretation of knowledge assets across systems. Both taxonomies and ontologies in healthcare organizations make it possible to connect, for instance, a patient’s health record with diagnostic data and previous case reviews for other patients based on the same (or similar) conditions, including illnesses, symptoms, treatments, and medicines. As a result, healthcare professionals can quickly access the information they need to make well-informed decisions about a patient’s care.

Where AI is Failing

On multiple occasions, AI has repeatedly failed to provide reliable information to employees, customers, and patients, undermining their confidence in the AI supported system and sometimes leading to serious organizational consequences. You may be familiar with the case in which a chatbot of a medical association was unintentionally giving harmful advice to people with eating disorders. Or maybe you heard in the news about the bank with a faulty AI system that misclassified thousands of transactions as fraudulent due to a programming error, resulting in significant customer dissatisfaction and harming the organization’s reputation. There was also a case in which an AI-powered translation system failed to accurately assess asylum seekers’ applications, raising serious concerns about its fairness and accuracy, and potentially affecting critical life decisions for those applicants. In each of these cases, had the corresponding AI systems effectively aggregated both unstructured and structured knowledge assets, and reliably linked them to encoded expert knowledge and relevant business context, these cases would have produced very different and positive outcomes. By leveraging taxonomies and ontologies to aggregate key knowledge assets, the result of these cases would have been much more closely aligned with intended objectives, ultimately, benefiting the end users as it was initially intended. 

How Taxonomies And Ontologies Enable Explainable AI

When knowledge assets are consistently tagged with taxonomies and related via ontologies, AI systems can trace how a decision was made. This means that end users can understand the reasoning path, supported by defined relationships. This also means that bias and hallucinations can be more easily detected by auditing the semantic structure behind the results.

As illustrated in the healthcare organization example, diagnoses can be tagged with medical industry taxonomies, while ontologies can help create relationships among symptoms, treatments, and outcomes. This can help physicians tailor treatments to individual patient needs by leveraging past patient cases and the collective expertise from other physicians. Similarly, a retail organization can enhance its customer service by implementing a chatbot that is linked to structured product taxonomies and ontologies to help deliver consistent and explainable answers about products to customers. More consistent and trustworthy customer interactions result in streamlining end user support and strengthening brand confidence.

Do We Really Need Taxonomies and Ontologies to be Successful With AI?

The examples above illustrate that explainability in AI really matters. Whether end users are patients, bank customers, or any individuals requesting specific products or services, they all want more transparent, trustworthy, and human-centered AI experiences. Taxonomies and ontologies help provide structure and connectedness to content, documents, data, expert knowledge and overall business context, so that they all are machine readable and findable by AI systems at the moment of need, ultimately creating meaningful interactions for end users.  

Conclusion

Just like bones, joints, ligaments, and muscles in the human body, taxonomies and ontologies provide the essential structure and connection that allow AI systems to stand up to testing, be reliable, and perform with clarity. At EK we have extensive experience identifying key knowledge assets as well as designing and implementing taxonomies and ontologies to successfully support AI initiatives. If you want to improve the Knowledge Intelligence (KI) of your existing or future AI applications and need help with your taxonomy and ontology efforts, don’t hesitate to get in touch with us. 

The post How Taxonomies and Ontologies Enable Explainable AI appeared first on Enterprise Knowledge.

Enterprise Knowledge recently completed the first round of our new webinar series The Semantic Exchange, which offers participants an opportunity to engage in Q&A with EK’s Semantic Design thought leaders. Participants were able to engage with EK’s experts on topics such as the value of enterprise semantic architecture, best practices for generating buy-in for semantics across an organization, and techniques for semantic solution implementation. The series sparked thoughtful discussion on how to understand and address real-world semantic challenges. 

To view any of the recorded sessions and their corresponding published work – use the links below:

If you are interested in bringing semantics and data modeling solutions to your organization, contact us here!

The post The Semantic Exchange Webinar Series Recap appeared first on Enterprise Knowledge.

The Solution

EK partnered with the professional association to establish the foundational content strategy, semantic models, and solution architecture to enable their goals for content discovery and analytics. First, EK conducted a current state analysis and target state definition, as well as a semantic maturity assessment. This helped EK understand the factors that could be leveraged to help the association realize its goals. EK subsequently completed three parallel workstreams:

1. Content Assessment: EK audited a sample of assets on priority web properties to understand the condition of the association’s content and semantic practices. EK identified recommendations for how to enhance the performance, governance, and discoverability of content. Based on these recommendations, EK provided step-by-step procedures to support the association in completing a comprehensive audit to enhance their content quality and aid in future findability enhancement and content personalization efforts.
2. Taxonomy and Ontology Development: EK developed an enterprise taxonomy and ontology framework for the association—to provide a standardized vocabulary for use across the association’s systems, and increase the maturity of the association’s semantic models. The enterprise taxonomy included 12 facets to support 12 metadata fields, with a cumulative total of over 900 concepts. An ontology identified key relationships between the different taxonomy facets, establishing a foundation for identifying related content and supporting auto-tagging.
3. Semantic Layer Architecture: EK provided recommendations for maturing the association’s tooling and integrations in support of their goals. Specifically, EK developed a solution architecture to integrate taxonomy, ontology, and auto-tagging across content, asset, and learning management systems, in order to inform a variety of content analytics, discovery, recommendation, and assembly applications. This architecture was designed to form the basis of a semantic layer that the association could later use to connect and relate content enterprise-wide. The architecture included the addition of a taxonomy and ontology management system (TOMS) to centralize semantic model management and to introduce auto-tagging capabilities. Alongside years of experience in tool evaluation, EK leveraged their proprietary TOMS evaluation matrix to score candidate vendors and TOMS solutions, supporting the association in selecting a tool that was the best fit for their needs.
4. Auto-Tagging Proof of Concept: Building on these efforts, EK conducted an auto-tagging proof of concept (PoC), to support the association in applying the taxonomy to their content. The PoC automatically tagged all content assets in 2 priority CMSes with concepts from 2 prioritized topic taxonomy facets. The EK team prepared the processing pipeline for the auto-tagging effort, including pre-processing the content and conducting analysis of the tags to gauge quality and improvement over time.

To determine the exact level of improvement, EK worked with subject matter experts to establish a gold standard set of expected tags for a sample of content assets. The tags produced by the auto-tagger were compared to the expected tag set, to generate measures of recall, precision, and accuracy. EK used the analytics to inform adjustments to the taxonomy facets and to fine-tune and improve the auto-tagger’s performance over successive rounds.

To support the association in continuing to grow and leverage their semantic maturity, EK provided a detailed semantic maturity implementation roadmap. The roadmap identified five target outcomes for semantic enrichment, including: enhancing analytics to provide insights into content use and content gaps; and recommending content by using content tags to suggest related resources. For each outcome, EK detailed the requisite goals, business value, tasks, and dependencies—providing the association with the guidance they needed to realize each outcome and further advance their semantic maturity.

The EK Difference

EK was uniquely positioned to help the association improve their semantic maturity. As thought leaders in the semantic space, EK had the expertise and experience to assess the association’s semantic maturity, identify opportunities for growth, and define a vision and roadmap to help the association realize its business priorities. Further, EK has a deep understanding of the semantic technology landscape. This positioned EK to deliver tailored solutions that reflect the specific needs of the association, ensuring the solutions contribute to the association’s long-term technology roadmap.

EK leveraged a holistic approach to assessing and advancing the association’s semantic maturity. EK’s proprietary semantic maturity assessment accounts for the varied factors that influence an organization’s semantic maturity, including considerations for people, process, content, models, and technology. This positions the association to develop the capabilities required for semantic maturity across all contributing factors. Building off of the semantic maturity assessment, EK delivered end-to-end services that supported the entire semantic lifecycle, from strategy through design, implementation, and governance. This provided the association with the semantic infrastructure to realize near-term value; for instance, developing an enterprise taxonomy and applying it to their content assets using auto-tagging. By using proprietary, industry-leading approaches, EK was able to deliver these end-to-end services with tangible results within 4 months.

The Results

EK delivered a semantic strategy and solution architecture, as well as a content clean-up strategy and initial taxonomy and ontology designs, that helped the professional association establish a foundation for realizing their goals. This effort culminated in the implementation of an auto-tagging PoC. The PoC included configuring the selected TOMS, establishing system integrations, and developing processing pipelines and quality evaluations. Ultimately, the PoC captured tags for over 23,000 content assets using more than 600 concepts from 2 priority taxonomy facets. This foundational work helped the professional association establish the initial components required for a semantic layer. A final roadmap and recommendations report provided detailed next steps, with specific tasks, dependencies, and pilots, to guide the professional association in leveraging and extending their foundational semantic layer. The first engagement was deemed a success by association leadership, and the roadmap was approved for phased implementation, which EK is now supporting. This continued partnership is enabling the association to begin realizing its goals of enhancing member engagement with content by improving content discovery and overall user experience.

Want to improve your organization’s content discovery capabilities? Interested in learning more about the semantic layer? Learn more from our experience or contact us today!

 

Download Flyer

 

The post Semantic Layer for Content Discovery, Personalization, and AI Readiness appeared first on Enterprise Knowledge.

If you’re reading this, you’re likely already convinced of the value of tags and filters when it comes to creating, saving, and retrieving content. You may even be involved in developing taxonomies at your organization. When working on initiatives aimed at enhancing targeted search and enabling content-level tagging, you’ll find that the systems that house and categorize your content each come with their own strengths and limitations regarding the extent to which they support tagging and filters. We have touched on system limitations in a past blog on Taxonomy Implementation Best Practices, and this blog revisits and expounds upon system limitations in greater detail. 

While some fortunate teams have the coveted balance of leadership support, generous budget, and ample time to shop around for a solution that supports robust taxonomy tagging and customizable filters, the reality for most of us entails having to make the best of whatever systems are already in place. Even teams that invest in a Taxonomy and Ontology Management System (TOMS) often have to navigate system limitations when integrating the tool with the existing systems. The systems that consume the taxonomy terms and filters tend to come with shortcomings that pose obstacles to optimal taxonomy implementation. Ideally, the consuming system supports hierarchical facets, captures semantic context like synonyms, supports an unlimited number of tags, and allows for advanced user permissioning, but the reality often falls short on at least one of these features. 

To an extent, these limitations should be expected, because the systems in question, be they content management systems, collaboration platforms, or learning management systems, have to fulfill a vast range of business needs. For these systems, tagging or search filters may be one of many desirable features, rather than their core function. 

In this blog, I invite two audiences to consider system limitations for taxonomy implementation: firstly, those who have the opportunity to shop around for systems to consume their taxonomy, and secondly, those who have to work around the limitations of the systems they already have in place. While this is by no means an exhaustive list, I’ll discuss some of the main considerations around taxonomy implementation for content management. 

All too often, systems are unable to support a fundamental element of taxonomies: hierarchical relationships. In these instances, the client finds themselves having to use tags or filters that don’t reflect broader and narrower relationships, and falsely represent very granular and very broad concepts at the same level.

For instance, you may have a tool that supports tagging, which is a great start for making content easier to find, but you aren’t able to indicate that tags like “Senior Health” and “Maternal Health” should be considered more specific than the broader “Health and Wellness” tag (Figure 1).

Why it Matters

The importance of hierarchies lies in the very definition of what separates a taxonomy from a mere controlled vocabulary. Because a taxonomy is a hierarchy of concepts accompanied by semantic context, one of the primary reasons a taxonomy is valuable is because of its ability to express broader and narrower relationships in a human-readable as well as machine-readable format. 

Hierarchies are especially valuable for supporting inheritance, in which attributes or properties defined at a higher level in the taxonomy are applied to lower-level elements, and, in turn, whatever qualities are true of the child concept are also true of the parent, or higher-level concepts.

In Figure 2, having hierarchical tagging would mean we wouldn’t have to tag a product with “To-Go Items,” “Beverages,” “Espresso Beverages,” and “Cafe Latte,” because we can see that Cafe Lattes are a type of Espresso Beverage, which is a type of Beverage, which are categorized under To-Go Items. 

This can be especially valuable because hierarchical tagging can help reduce the amount of work that has to go into using tags to store and retrieve content, as well as helping the user understand the significance of their tags. Without hierarchical tagging, a user may have to tag a piece of content with both the narrowest concept as well as broader concepts, rather than just apply one tag – which results in excessive tagging, depleting the tags’ value.

 

Recommendation: Develop detailed tagging guidance to compensate for system limitations, accompanied by an easily-accessible visual of the taxonomy hierarchy for users.  

System Limitation 2: Semantic Context Not Supported

In many systems, tagging – even hierarchical tagging – may be supported, but the tags themselves aren’t accompanied by any additional semantic context to provide additional guidance around what the tag means. For taxonomies adhering to Simple Knowledge Organization System standards, each taxonomy concept (or tag) should include not just the term itself (or, preferred label), but also be accompanied by alternative labels (equivalent terms or synonyms), a definition, and, if applicable, scope notes to provide additional context around intended usage or disambiguation between other concepts. However, many systems lack the ability to store or leverage this valuable context. In many cases, a tag is just a tag, and system users have to rely on implicit knowledge to interpret the broader meanings associated with each tag.

Why it Matters

Although good taxonomy design entails carefully selecting self-explanatory, clear concept labels, there can still be a certain amount of implicit knowledge required to accurately interpret, apply, and leverage tags. Ideally, anyone using a taxonomy should be able to also review alternative labels or scope notes as well as definitions, but this is often unavailable.  Alternative labels, definitions, and scope notes are essential for documenting otherwise implicit knowledge that, without which, could result in tags being inappropriately applied to content. This can result in undesirable outcomes like scope drift, incorrect tagging, and more. 

The semantic context provided by a robust taxonomy is also essential for future-proofing a taxonomy. With industries experiencing greater degrees of uncertainty and worker turnover, it’s more important than ever to avoid taking institutional knowledge for granted. For instance, a team may be initially aligned on the terminology and meaning for a taxonomy of 50 concepts, which are also tagged to content. However, let’s say that the team responsible for the taxonomy undergoes a reorganization at their company, and some individuals move to different teams, others retire, and others move to another company. Whoever inherits ownership of the taxonomy may no longer have access to the knowledge that contextualizes the meaning of those taxonomy terms. Over time, the terms lose their meaning, and are subject to different interpretations and usage.

In Figure 3, we can see that both pieces of content (a blog and a help article) mention “Tips.” Without the additional context provided by the alternative labels and definitions, it would be reasonable to have both pieces of content tagged with “Tips.” However, being able to access this context reveals that only the second article should have this tag.

 As illustrated in Figure 4, suppose a company isn’t consistent about using the term “Seller Protection” or “Merchant Protection.” Perhaps the Legal team calls it “Merchant Protection,” but customer-facing content tends to say “Seller Protection,” and internal customer support workers refer to it as “Merchant Protection.” In a case where content has to be manually tagged, just having a tag for “Seller Protection” without also indicating that “Merchant Protection” is an equivalent term would result in content not being tagged, even when it should be.

 

Recommendation: Ensure that key semantic context such as alternative labels and definitions is available to users, highlighting areas of frequent confusion.

System Limitation 3: Tag Number Limits

Another issue that users frequently encounter is where a system supports tagging, but imposes limitations on the quantity of tags supported. These limits primarily manifest themselves in two ways: firstly, by imposing a limit on the total number of tags that the system can support, or secondly, putting a limit on the number of tags that can be applied to a single piece of content. Some systems may impose both kinds of limitations.

Why it Matters

Generally speaking, limiting the number of tags you maintain in your system can actually be one of many ways to adhere to best practices in taxonomy. We frequently encounter situations where the number of tags within a system have proliferated well beyond a manageable quantity, and users may only meaningfully engage with several dozen tags out of thousands. In these instances, we often find that many of the tags are duplicates, represent concepts that are no longer applicable or relevant, or go into an excess level of granularity. However, there can be situations in which a higher number of tags may be necessary, such as when a particularly diverse range of content or data needs to be retrievable and a certain degree of specificity is required to ensure precision in search.

In one instance, we worked with a client using a CMS that could only allow for a total of 15 tags to be assigned to a single piece of content. While, in most cases, 15 tags or fewer would be an appropriate maximum number of tags to assign to a document to avoid over-tagging, this limitation could be problematic for long-form content, like complex user agreements or long scientific articles. In another instance, a system could only support 500 total tags – which would be appropriate enough for a repository of simple help articles, but may be insufficient for a database of complex healthcare topics.

Suppose a content repository houses hundreds of long, highly complex user manuals like the one shown in Figure 5. However, the content repository only allows for 12 tags to be applied to any given content item. In this situation, a 12-tag limit would be woefully inadequate.

 

Recommendation: Explore ways to accompany complex documents with additional context to complement the limited tags.

System Limitation 4: Inadequate Options for Roles and Permissions

When implementing a taxonomy with a system, whether that means configuring search filters or tagging content, it’s crucial to be able to configure user roles and permissions. On many teams, content roles are distinct; you have content strategists, content authors, content managers, a taxonomist, and so on. Furthermore, a mature taxonomy requires governance, where well-defined roles dictate who is responsible for deciding upon and carrying out updates to a taxonomy, and effective governance is difficult to achieve without a system that supports distinct permissions to accompany these defined roles. However, not all systems have customizable roles to support governance or even basic taxonomy management. In some cases, new tags can be freely entered, or anyone can have the ability to apply tags to content.

Why it Matters

Anyone who’s been involved in content tagging knows how challenging it can be to successfully guide a group of taggers to consistently tag content the same way. Some may take an exhaustive approach, tagging content with concepts that may only be tangentially relevant or only mentioned in passing. Others may be more cautious, only applying tags when a concept acts as one of the core themes of a piece of content. While drafting detailed tagging guidelines and providing robust tagging training can go a long way in mitigating these differences in tagging approaches, this is not a foolproof approach. 

These issues can be only further exacerbated when systems allow for any individual to add new tags to a system without any formal governance review. In one instance, we worked with a nonprofit client where it was easy to add new tags to a system, and over time, the total number of tags grew to over 2,000. In another case, at a large financial technology organization I worked with, there were no system limitations on how many entries could be added to a drop-down menu of customer contact reasons, which resulted in a list of over 1,400 entries, the majority of which were duplicates. In situations like these, tags lose the value they were meant to provide, and may even foster more confusion than if there were no tags at all.

In the example shown in Figure 6, a system where any users can create a new filter or tag without any user roles restricting permissions can easily lead to a proliferation of overlapping or duplicated terms. Here, we see a drop-down menu of “Account Questions,” where users were able to create new tags without having to review which tags are already in place, resulting in six different selections about creating an account. Lack of controls around tags or filter creation can quickly result in an unwieldy tool that is difficult to use. If tags or filters become bloated enough, users may even select a random term or only pick the top term rather than scroll through hundreds of entries.

Figure 7 illustrates a situation where two different content authors were asked to tag content. On the left, one tagger got carried away, providing far more tags than were necessary, and misinterpreting “Brand Health” to mean physical health. On the right, a different tagger was more strategic and selective, and only applied tags that would be necessary to retrieve the intended report. User controls help ensure that only qualified individuals can create, apply, and manage tags, and are essential to prevent an uncontrolled proliferation of tags, enable consistent tagging, and support ongoing governance.

 

Recommendation: Provide detailed tagging guidelines as well as training to ensure aligned approaches. For systems where new entries can be freely added, communicate frequently about the importance of preventing accidental entries, and schedule frequent clean-ups.

Conclusion

When navigating the transition from designing a taxonomy to implementing it in the intended systems, it can be common to encounter a gap between ideal implementation (hierarchical tagging without system-imposed limits, controlled by tight role-based user permissions), and reality. A certain degree of flexibility, creativity, and above all, effective documentation will be necessary to develop effective strategies to get as much out of your taxonomy as possible in spite of any challenging system limitations. Developing a detailed set of use cases that the taxonomy is intended to address is essential, so you can evaluate which system limitations are acceptable and which should be considered deal-breakers for the success of your project. 

Are you in the process of implementing a taxonomy at your organization, but struggling to navigate any of these system limitations? We have extensive experience navigating these system limitations to guide organizations to successful, impactful taxonomy implementations, and would be happy to help. Contact us to learn more.

The post Navigating System Limitations for Taxonomy Implementation appeared first on Enterprise Knowledge.

Just as a semantic layer passes through degrees of maturity and complexity as it is developed and operationalized, so too does a taxonomy. While a taxonomy comprises only one facet of a fully realized Semantic Layer, every incremental increase in its granularity and scope can have a compounding effect in terms of unlocking additional solutions for the organization. While it can be tempting to assume that only a fully mature taxonomy is capable of delivering measurable value for the organization that developed it, each iteration of a taxonomy provides value that should be acknowledged, quantified, and celebrated to advocate for continued support of the taxonomy’s ongoing development.  

Taxonomy Maturity Stages

A taxonomy’s maturity can be measured across five levels: Basic, Foundational, Operational, Institutional, and Transformational. Taken as a snapshot from our full semantic layer maturity framework, the following diagram illustrates each of these levels in terms of their taxonomy components, technical manifestation, and what valuable outcomes can be expected from each at a high level. 

Basic Taxonomy

A basic taxonomy lacks depth, and is essentially a folksonomy (an informal, non-hierarchical classification system where users apply public tags). At this stage, a basic taxonomy is only inconsistently applied across departments. 

As an example, a single business unit (Marketing) may have begun developing a basic taxonomy that other business units (Sales) may be starting to integrate with their product taxonomy. 

Components and Technical Manifestation at this Level

• Basic taxonomies are only developed for limited, specific use cases, often for a particular team or subset of an organization.
• At this stage of maturity, a taxonomy expresses little granularity, and may have up to three levels of broader/narrower relationships. 
• A basic taxonomy is likely maintained in a spreadsheet, rather than a taxonomy management system (TMS). The taxonomy may be implemented in a rudimentary form, like being expressed in file structures. Taxonomy concepts are not yet tagged to assets. 
• At this stage, the taxonomy functions primarily as a proof of concept. The taxonomy has not yet been widely validated or socialized, and is likely only known by the team building it. It may represent an intentionally narrow scope that can then be scaled as the team builds buy-in with stakeholders. 

Outcomes and Value 

• The basic taxonomy provides an essential foundation to build upon. If it is well-designed, the work invested in this stage can serve as a model for other functional areas of the organization to adopt for their own use cases. 
• At this stage, the value is typically limited to providing a proof of concept to demonstrate what taxonomy is, and working towards establishing consistent terminology within a department.

Foundational Taxonomy

The foundational taxonomy is not yet wholly standardized, but growing momentum helps to drive adoption and standardization across systems and business units. The taxonomy can support simple data enrichment by adding semantic context (like relevant location data, contact information, definitions, or subcategories) to an existing data set. Often, a dedicated taxonomy management solution (TMS) is procured at this stage, and it may be unscalable to proceed to the next level of maturity without one. 

Components and Technical Manifestation at this Level

• The taxonomy is imbued with semantic context such as definitions, scope notes, and alternative labels, along with the expected hierarchical relationships between concepts. A foundational taxonomy exhibits a greater level of granularity beyond the basic level. 
• The taxonomy is no longer only housed in a spreadsheet, and is maintained in a Taxonomy Management Solution (TMS). This makes it easier to ensure that the taxonomy’s format adheres to semantic web frameworks (such as SKOS, the Simple Knowledge Organization System). 
• The addition of this context serves the fundamental purpose of supporting and standardizing semantic understanding within an organization by clarifying and enforcing preferred terms while still capturing alternative terms.  
• Some degree of implementation has been realized – for instance, the tagging of a representative set of content or data assets.
• The taxonomy team actively engages in efforts to socialize and promote the taxonomy project to build awareness and support among stakeholders. 
• A taxonomy governance team has been established for ongoing validation, maintenance, and change management. 

Outcomes and Value

• At this stage, the taxonomy can provide more measurable benefits to the organization. For instance, a foundational taxonomy can support content audits for all content that has been auto-tagged. 
• The taxonomy can support more advanced data analytics – for instance, users can get more granular insights into which topics are the most represented in content. 
• The foundational taxonomy can be scaled to incorporate backlog use cases or other departments in the organization, and can be considered a product to be replicated and more broadly socialized.
• The taxonomy can be enhanced by adding linked models and/or concept mapping.

Operational Taxonomy

The operational taxonomy is standardized, used regularly and consistently across teams, and is integrated with other components or applications. 

At this stage, the taxonomy is integrated with key systems like a content management system (CMS), learning management system (LMS), or similar. Users are able to interact with the taxonomy directly through the system-powered apps they work in, because the systems consume the taxonomy.

Components and Technical Manifestation at this Level

• At this level of maturity, advanced integrations have been realized – for instance, the taxonomy is integrated into search for the organization’s intranet, or the taxonomy’s semantic context has been leveraged as training data for generative AI-powered chatbots.
• At the operational level, the taxonomy acts as a source of truth for multiple use cases, and has been expanded to cover multiple key areas of the organization, such as Customer Operations, Product, and Content Operations. 
• By this stage, content tagging has been seamlessly integrated into the content creation process, in which content creators apply relevant tags prior to publishing, or automatic tagging ensures content is applied to current and newly-published content. 
• A TMS has been acquired, and is implemented with key systems, such as the organization’s LMS, intranet, or CMS. 
• The taxonomy is subject to ongoing governance by a taxonomy governance team, and key stakeholders in the organization are informed of key updates or changes to the taxonomy.

Outcomes and Value 

• The taxonomy is integrated with essential data sources to provide or consume data directly. As a result, users interacting with the systems that are connected to the taxonomy are able to experience the additional structure and clarity provided by the taxonomy via features like search filters, navigational structures, and content tags. 
• The taxonomy can support enhanced data analytics, such as tracking the click-through rate (CTR) of content tagged with particular topics. 

Institutional Taxonomy

The institutional taxonomy is fully integrated into daily operations. Rigorous governance and change management capabilities are in place. 

By now, seamless integrations between the taxonomy and other systems have been established. Ongoing taxonomy maintenance work poses no disruption to day-to-day operations, and updates to the taxonomy are automatically pushed to all impacted systems.

Components and Technical Manifestation at this Level

• The taxonomy, or taxonomies, are fully integrated into daily operations across teams and functional areas – for instance, the taxonomy supports dynamic content delivery for customer support workers, the customer-facing product taxonomy facilitates faceted search for online shopping, and so on. 
• The organization’s use cases are supported by the taxonomy, which supports core goals such as ensuring a shared understanding of key concepts and their meaning, providing a consistent framework for the representation of data across systems, or representing the fundamental components of an organization across systems. 
• Governance roles, policies, and procedures are fully established and follow a regular cadence. 

Outcomes and Value

• At this stage of maturity, the taxonomy has been scaled to the extent that it can be considered an enterprise taxonomy; it covers all foundational areas, is utilized by all business units, and is poised to support key organizational operations. At this stage, the taxonomy drives a key enterprise-level use case. 
• Data connectivity is supported across the organization; the taxonomy unifies language across teams and systems, reducing errors and data discrepancies. 
• Internal as well as external users benefit from taxonomy-enhanced search in the form of query expansion. 

Transformational Taxonomy

The transformational taxonomy drives data classification and advanced analytics, informing and enhancing AI-driven processes. At this stage, the taxonomy provides significant functionality supporting an integrated semantic layer. 

Components and Technical Manifestation at this Level

• The taxonomy can support the delivery of personalized, dynamic content for internal or external users for more impactful customer support or marketing outreach campaigns.
• The taxonomy is inextricably tied to other key components of the semantic layer’s operating model. The taxonomy provides data for the knowledge graph, provides a hierarchy for the ontology, categorizes the data in the data catalog, and enriches the business glossary with additional semantic context. These connections help power semantic search, analytics, recommendation systems, discoverability, and other semantic applications. 
• Taxonomy governance roles are embedded in functional groups. Feedback on the taxonomy is shared regularly, introductory taxonomy training is widely available, and there is common understanding of how to both use the taxonomy and provide feedback. 
• Taxonomies are well-supported by defined metrics and reporting and, in turn, provide a source of truth to power consistent reporting and data analytics.  

Outcomes and Value 

• At this stage, the taxonomy (within the broader semantic layer) drives multiple enterprise-level use cases. For instance, this could include self-service performance monitoring to support strategic planning, or facilitating efficient data analytics across previously-siloed datasets. 
• Taxonomy labeling of structured and/or unstructured data powers Machine Learning (ML) and Artificial Intelligence (AI) development and applications. 

Taxonomy Use Cases 

Low Maturity Example

In many instances, EK partners with clients to help develop taxonomies in their earliest stages. Recently, a data and AI platform company engaged EK to lead a taxonomy workshop covering best practices in taxonomy design, validation activities, taxonomy governance, and developing an implementation roadmap. Prior to EK’s engagement, the company was in the process of developing a centralized marketing taxonomy. As the taxonomy was maintained in a shared spreadsheet, lacked a defined governance process, and lacked consistent design guidelines, it met the basic level of maturity. However, after the workshop, the client’s taxonomy design team left with a refreshed understanding of taxonomy design best practices, clarified user personas, an appreciation of the value of semantic web standards, a clear taxonomy development roadmap, and a scaled-down focus on prioritized pilots to build a starter taxonomy. 

By clarifying and narrowing their use cases, identifying their key stakeholders and their roles in taxonomy governance, and reworking the taxonomy to reflect design principles grounded in semantic standards, the taxonomy team was equipped to elevate their taxonomy from a basic level of maturity to work towards becoming foundational. 

High Maturity Example 

EK’s collaboration with a major international retailer illustrates an example of the evolution towards a highly-mature semantic layer supported by a robust taxonomy. EK partnered with the retailer’s Learning Team to develop a Learning Content Database to enable an enterprise view of their learning content. Initially, the organization’s learning team lacked a standardized taxonomy. This made it difficult to identify obsolete content, update outdated content, or address training gaps. Without consistent terminology or content categorization, it was especially challenging to search effectively and identify existing learning content that could be improved, forcing the learning team to waste time creating new content. As a result, store associates struggled to search for the right instructional resources, hindering their ability to learn about new roles, understand procedures, and adhere to compliance requirements. 

To address these issues, EK first partnered with the learning team to develop a standardized taxonomy. The taxonomy crystallized brand-approved language which was then tagged to learning content. Next, EK developed a tailored governance plan to ensure the ongoing maintenance of the taxonomy, and provided guidance around taxonomy implementation to ensure optimal outcomes around reducing time spent searching for content and simplifying the process of tagging content with metadata. With the taxonomy at a sufficient stage of maturity, EK was then able to build the Learning Content Database, which enabled users to locate learning content across previously disparate, disconnected systems, now in a central location. 

Conclusion

Every taxonomy – from the basic starter taxonomy to the highly-developed taxonomy with robust semantic context connected to an ontology – can provide value to its organization. As a taxonomy grows in maturity, each next level of development unlocks increasingly complex solutions. From driving alignment around key terms for products and resources, supporting content audits, enabling complex data analytics across systems, or powering semantic search, the progressive advancement of a taxonomy’s complexity and semantic richness translates to tangible business value. These advancements can also act as a flywheel, where each improvement makes it easier to continue to drive buy-in, secure necessary resources, and achieve greater enhancements. 

If you are looking to learn more about how other organizations have benefitted from advanced taxonomy implementations, read more from our case studies. If you want additional guidance on how to take your organization’s taxonomy to the next level, contact us to learn more about our taxonomy design services and workshops.

The post Semantic Layer Maturity Framework Series: Taxonomy appeared first on Enterprise Knowledge.

Taxonomy for the Semantic Layer

Taxonomies are controlled vocabularies that organize terms or concepts into a hierarchy based on their relationships, serving as key knowledge organization systems within the semantic layer to promote consistent naming conventions and a common understanding of business concepts. Categorizing concepts in a structured and meaningful format via hierarchy clarifies the relationships between terms and enriches their semantic context, streamlining the navigation, findability, and retrieval of information across systems.

Taxonomies are often a foundational component in RDF-based graph development used to structure and classify data for more effective inference and reasoning. As graph technologies evolve, the application of taxonomy is gaining relevance beyond RDF, particularly in the realm of LPGs, where it can play a crucial role in data classification and connectivity for more flexible, scalable, and dynamic graph analytics.

The Role of Taxonomy in LPGs

Even in the flexible world of LPGs, taxonomies help introduce a layer of semantic structure that promotes clarity and consistency for enriching graph analytics:

Taxonomy Labels for Semantic Standardization

Taxonomy offers consistency in how node and edge properties in LPGs are defined and interpreted across diverse data sources. These standardized vocabularies align labels for properties like roles, categories, or statuses to ensure consistent classification across the graph. Taxonomies in LPGs can dynamically evolve alongside the graph structure, serving as flexible reference frameworks that adapt to shifting terminology and heterogeneous data sources. 

For instance, a professional networking graph may encounter job titles like “HR Manager,” “HR Director,” or “Human Resources Lead.” As new titles emerge or organizational structures change, a controlled job title taxonomy can be updated and applied dynamically, mapping these variations to a preferred label (e.g., “Human Resources Professional”) without requiring schema changes. This enables ongoing accurate grouping, querying, and analysis. This taxonomy-based standardization is foundational for maintaining clarity in LPG-driven analytics.

Taxonomy as Reference Data Modeled in an LPG

LPGs can also embed taxonomies directly as part of the graph itself by modeling them as nodes and edges representing category hierarchies (e.g. for job roles or product types). This approach enriches analytics by treating taxonomies as first-class citizens in the graph, enabling semantic traversal, contextual queries, and dynamic aggregation. For example, consider a retail graph that includes a product taxonomy: “Electronics” → “Laptops” → “Gaming Laptops.” When these categories are modeled as nodes, individual product nodes can link directly to the appropriate taxonomy node. This allows analysts to traverse the category hierarchy, aggregate metrics at different abstraction levels, or infer contextual similarity based on proximity within the taxonomy. 

EK is currently leveraging this approach with an intelligence agency developing an LPG-based graph analytics solution for criminal investigations. This solution requires consistent data classification and linkage for their analysts to effectively aggregate and analyze criminal network data. Taxonomy nodes in the graph, representing types of roles, events, locations, goods, and other categorical data involved in criminal investigations, facilitate graph traversal and analytics.

In contrast to flat property tags or external lookups, embedding taxonomies within the graph enables LPGs to perform classification-aware analysis through native graph traversal, avoiding reliance on fixed, rigid rules. This flexibility is especially important for LPGs, where structure evolves dynamically and can vary across datasets. Taxonomies provide a consistent, adaptable way to maintain meaningful organization without sacrificing flexibility.

Taxonomy in the Context of LPG-Driven Analytics Use Cases

Taxonomies introduce greater structure and clarity for dynamic categorization of complex, interconnected data. The flexibility of taxonomies for LPGs is particularly useful for graph analytics-based use cases, such as recommendation engines, network analysis for fraud detection, and supply chain analytics.

For recommendation engines in the retail space, clear taxonomy categories such as product type, user interest, or brand preference enable an LPG to map interactions between users and products for advanced and adaptive analysis of preferences and trends. These taxonomies can evolve dynamically as new product types or user segments emerge for more accurate recommendations in real-time. In fraud detection for financial domains, LPG nodes representing financial transactions can have properties that specify the fraud risk level or transaction type based on a predefined taxonomy. With risk level classifications, the graph can be searched more efficiently to detect suspicious activities and emerging fraud patterns. For supply chain analysis, applying taxonomies such as region, product type, or shipment status to entities like suppliers or products allows for flexible grouping that can better accommodate evolving product ranges, supplier networks, and logistical operations. This adaptability makes it possible to identify supply chain bottlenecks, optimize routing, and detect emerging risks with greater accuracy.

Conclusion

By incorporating taxonomy in Labeled Property Graphs, organizations can leverage structure while retaining flexibility, making the graph both scalable and adaptive to complex business requirements. This combination of taxonomy-driven classification and the dynamic nature of LPGs provides a powerful semantic foundation for graph analytics applications across industries. Contact EK to learn more about incorporating taxonomy into LPG development to enrich your graph analytics applications.

The post The Role of Taxonomy in Labeled Property Graphs (LPGs) & Graph Analytics appeared first on Enterprise Knowledge.

The post Why Your Taxonomy Needs SKOS  appeared first on Enterprise Knowledge.

The Solution

EK collaborated with the company to improve their internal search platform by enhancing Electronic Lab Notebook (ELN) metadata, thereby increasing the searchability and findability of critical research documents, and created a strategy for leveraging ELNs in AI-powered services such as chatbots and LLM-generated document summaries. EK worked with the business stakeholders to evaluate the most important information within ELNs and understand the document structure, and developed semantic models in their taxonomy management system with more than 960 relevant concepts designed to capture the way their expert chemists understand the experimental activities and molecules referenced in the ELNs. With the help of the client’s technical infrastructure team, EK developed a new corpus analysis and ELN autotagging pipeline that leveraged the taxonomy management system’s built-in document analyzer and integrated the results with their data warehouse and search schema. Through three rounds of testing, EK iteratively improved the extraction of metadata from ELNs using the concepts in the semantic model to provide additional metadata on over 30,000 ELNs to be leveraged within the search platform. EK wireframed 6 new User Interface (UI) features and enhancements for the search platform designed to leverage the additional metadata provided by the autotagging pipeline, including search-as-you-type functionality and improved search filters, and socialized them with the client’s UI/ User Experience (UX) team. Finally, EK supported the client with strategic guidance for leveraging their internal LLM service to create accurate regulatory reports and AI summaries of ELNs within the search platform.

The EK Difference

EK leveraged its understanding of the capabilities and features of enterprise search platforms, and taxonomy management systems’ functionality, to advise the organization on industry standards and best practices for managing its taxonomy and optimizing search with semantics. Furthermore, EK’s experience working with other pharmaceutical institutions and large organizations in the development of semantic models benefited the client by ensuring their semantic models were comprehensively and specifically tailored to meet their needs for the development of their semantic search platform and generative AI use cases. Throughout the engagement, EK incorporated an Agile project approach that focused on iterative development and regular insight gathering from client stakeholders, to quickly prototype enhancements to the autotagging pipeline, semantic models and the search platform that the client could present to internal stakeholders to gain buy-in for future expansion. 

The Results

EK’s expertise in knowledge extraction, semantic modeling and implementation, along with a user-focused strategy that ensured that improvements to the search platform were grounded in stakeholder needs, enabled EK to effectively provide the client with a major update to their search experience. As a result of the engagement, the client’s newly established autotagging pipeline is enhancing tens of thousands of critical research documents with much-needed additional metadata, enabling dynamic context-aware searches and providing users of the search platform with insight at a glance into what information an ELN contains. The semantic models powering the upgraded search experience allow users to look for information using natural, familiar language by capturing synonyms and alternative spellings of common search terms, ensuring that users can find what they are looking for without having to do multiple searches. The planned enhancements to the search platform will save scientists at the company hours every week from searching for information and judging if specific ELNs are useful for their purposes or not, reducing reliance on individual employee knowledge and the need for the regulatory intelligence team to rediscover institutional knowledge. Furthermore, the company is equipped to move forward towards leveraging the combined power of semantic models and AI to improve the speed and efficiency of document understanding and use. By utilizing improved document metadata provided by the auto-tagging pipeline in conjunction with their internal LLM service, they will be able to generate factual document summaries in the search platform and automate the creation of regulatory reports in a secure, verifiable, and hallucination-free manner. 

 

The post Extracting Knowledge from Documents: Enabling Semantic Search for Pharmaceutical Research and Development appeared first on Enterprise Knowledge.

Enterprise Knowledge CEO Zach Wahl speaks with Ahren Lehnert, Principal Taxonomist at Nike.

In this conversation, Zach and Ahren discuss the future of taxonomy and artificial intelligence (AI), emphasizing both the augmentation of traditional roles and growth to include new ones, and how to demonstrate the value of a taxonomy initiative for your organization more tangibly. They walk through highs and lows of the taxonomy development and refinement process and the importance of avoiding semantic debt to ensure adaptable solutions that stand the test of time. Ahren also shares his thoughts on the state of the market for semantic tools as the influence of AI/ML grows.

If you would like to be a guest on Knowledge Cast, contact Enterprise Knowledge for more information.

The post Knowledge Cast – Ahren Lehnert at Nike appeared first on Enterprise Knowledge.

Whether you are striving to enhance user experiences by improving search or navigation, or by improving asset management or reporting, in a single system or across multiple systems—you need metadata.

But not just any metadata. You need metadata that provides the information and context needed to leverage assets effectively and meaningfully.

For those seeking to extend or enhance the metadata for their organizational assets, it can be difficult to ascertain what metadata should be captured. In this blog post, I provide an overview of the role of metadata, and guidance on what to consider when defining metadata.

The Role of Metadata in a Semantic Layer

Metadata codifies characteristics of an asset—what it is about, how it should be managed and used, and what contexts it is relevant for. For example, a document may have metadata to identify its title, publication date, lifecycle status, document type, and topic.

Additionally, metadata should capture actionable characteristics of an asset, reflecting the characteristics necessary for finding, managing, using, and understanding assets. In other words, if the characteristic is used to support a requisite interaction with the asset, it should be captured as metadata.

By codifying actionable characteristics, metadata enables user experiences by providing the connection between a specific asset and supported interactions.

Metadata also enables AI-supported interactions. As an explicit signal of an asset’s characteristics, metadata ensures that AI-powered tools can reference and leverage the asset when appropriate. For instance, a semantic search engine can be tuned to prioritize information in assets marked with specific Document Types, or a content generation tool can be directed to summarize the most recent, published assets on a topic.

The utility of metadata is limited by any differences in its definition or application. If different terms are used to identify the same topic, it is more difficult to identify assets that are about the same, or related, topics. Similarly, if a topic is applied inconsistently—missing on relevant assets or present on irrelevant assets—it is more difficult to identify similar or related assets. 

This underscores the importance of standardizing metadata. To ensure concepts, like dates and topics, are identified and applied consistently, metadata should be defined as a shared representation; that is, characteristics common to assets across systems should share the same metadata definition, and common concepts, like topics and document types, should be standardized with a taxonomy. This approach controls what data should be captured for an asset and what terms can be used to identify concepts, helping to enforce a shared representation across assets.

When metadata is standardized, it is possible to improve user experiences, such as improving asset discovery within a repository via a common metadata definition and taxonomy or across repositories by establishing a metadata knowledge graph.

How to Define Metadata for a Semantic Layer

The process for defining metadata within your organization may look different, depending on what metadata, taxonomies, or other Semantic Layer components you already have in place. There are fundamental considerations, however, that can be useful regardless of what you have in place to date. These include:

• Determining use cases: Consider what specific use cases you need to support. This will help you focus on the metadata that will be most actionable and impactful. Look to your users to understand what their tasks are and what pain points they have, like finding documents for research or identifying data sets for reporting. 
• Identifying metadata fields: Investigate the ways users describe, look for, or interact with the assets for the use cases you’ve identified. Consider the different kinds of information and context that may be required to support the use cases.
• Allocating metadata fields: Consider which assets and systems will need to use the metadata field. Identify metadata fields that are applicable to all assets on all (in-scope) systems or some assets on all (in-scope) systems. These metadata fields will be the focus for standardization.
• Standardizing metadata fields: Establish definitions for each metadata field and ensure those definitions can be upheld across in-scope systems. Specify the role of each field, the type of data they capture, and, if applicable, whether they accommodate single or multiple values. For metadata fields that leverage a controlled list of terms, design taxonomies to further standardize and enrich the capture of information and context.

Importantly, you don’t have to define all the metadata all at once. These considerations can be applied to a specific, priority use case to start, then revisited and expanded as needed. Even a small initial set of metadata can help transform user experiences. 

Contact us if you are grappling with how to define and standardize your metadata, or are interested in learning more about the Semantic Layer.

The post Metadata Within the Semantic Layer appeared first on Enterprise Knowledge.