Understanding Taxonomy and Metadata

EK defines a taxonomy as a “controlled vocabulary used to describe or characterize explicit concepts of information, for the purpose of capturing, managing, and presenting information.” Metadata, therefore, can be thought of as the building blocks of a taxonomy; it is “descriptive data about a content item” or, in other words, characteristics of a document that allow you to categorize it. When working with clients, I like to have them visualize taxonomies as a series of drop down lists that can be used to categorize and tag their content. Some dropdown lists will have multiple levels (hierarchies) that allow users to determine the level of specificity and depth that is applicable, while other fields are a single level list (flat). For example, one of our clients enlisted our services to design a legal taxonomy that could be used to help them find critical contracts across their global offices. Metadata that made up this legal taxonomy included fields such as “Subject Matter” (a list of hierarchical terms 3-levels deep that captured what a contract was about), “Geographic Location” (a flat list of terms that captured the different global regions a contract could be executed by the organization) and “Document Type” (a hierarchical list of terms 2-levels deep that captured the various types of legal documents utilized by the organization). The terms that made up each of these metadata fields were used to tag, find, and search for legal documents in their system.

Understanding Search Filters

As my colleague Jenni Doughty pointed out in her blog “How Do I Implement a Taxonomy?”, one of the primary determinations that needs to be made when implementing a taxonomy is deciding how to implement each metadata field into the system, particularly when the use-cases, hierarchy, and quantity of terms within each metadata field can vary greatly. For SharePoint Online, a crucial consideration is how to best translate metadata into search filters, or the options that pop up on the left-hand side of a search results page that a user can select to narrow down their search results (see a snapshot of Enterprise Knowledge’s open-access Knowledge Base below for an example of an “Article Type” search filter).

Search filters are a powerful tool in your organization’s Knowledge Management (KM) arsenal, as they allow users to go beyond employing keywords in their search queries to track down the content item(s) that will help them take a necessary action. Search filters are largely driven by taxonomy fields; not only do search filters themselves directly align to taxonomy fields and their terms, but they are also fueled by content being tagged with the same terms that are captured in filters on the search results page (It is important to note, however, that there are search refiners for non-taxonomy metadata fields in SharePoint, such as date ranges, that allow a user to limit the search by a date metadata field). Ultimately, search filters not only make search an interactive experience for users, but gives users the ability to quickly and easily adjust their search results within a system like SharePoint Online. For more information about the technical side of search filtering, see my colleague’s blog, How to Optimize Search Relevance: Boosting and Filtering.

Metadata as Search Filters, Out-of-the-Box SharePoint Online

For out-of-the-box SharePoint Online (or SharePoint without add-ons or customizations), it is not currently possible to capture hierarchy in search filters, meaning that nested terms and parent-child relationships are not able to be visually or systematically depicted within search filters. Because of this, it is pivotal that your organization not only determines what metadata fields will best serve users as search filters, but also, whether displaying a single, selected level of a hierarchical field will provide users with an optimal and intuitive search experience.

Let’s put this into practice. For many of our clients, a metadata field for “Topic” is a great candidate for a search filter, as many users want to narrow down their search results by selecting terms that capture what a content item is about. Topical metadata fields, however, often contain up to 4-levels of hierarchy, are rooted in parent-child relationships, and can contain hundreds of different terms. For instance, your “Topic” metadata field might have term relationships that span 4-levels, like the following:

Because out-of-the-box SharePoint Online does not capture hierarchies or nested terms in its search filters, implementing every term in this type of field as a search filter would result in a single level list of hundreds of terms that would not only ignore the relationships that exist in your taxonomy, but would be visually burdensome and confusing for a user to sift through in order to find what they need. If we stick to our example with the 4-level term relationships that might exist in your “Topic” metadata field, this would mean that the terms detailed in the table above, if directly translated into search filters in SharePoint Online, would each be treated as individual entities in a flat list:

As depicted in the image above, when this full set of terms in “Topic” is implemented into SharePoint Online, the lack of ability to nest child-terms underneath parent terms makes it difficult for users to discern that any term relationships exist at all. Furthermore, in a real-world scenario where it is likely that an organization’s “Topic” metadata field contains hundreds of unique terms at various levels, it becomes difficult for users to digest the quantity and breadth of terms when there is no logical order to them in terms of their concept or level of specificity.

Because of Sharepoint’s out-of-the-box limitation with search filters, EK typically recommends employing a single level of a hierarchical field as a search filter (typically the top level) rather than the full exhaustive term list, particularly if this metadata field will help users narrow down their search results to find or discover the content that they need to do their jobs. Continuing with our 4-level metadata field “Topic,” selecting just the Level 1 terms as search filters would result in a more streamlined, understandable set of filters:

By selecting a single level of a metadata field, such as “Topic” for a search filter, it becomes paramount that your organization’s taxonomy is also implemented in a way that ensures all content items tagged with a more specific, deeper level term that also inherits the tag(s) of the parent level terms. For example, if the most applicable tag for a content item is the topic of “Chickpeas,” we would also ensure that the content item be tagged with the parent terms, which in this use case would include “Small Grains & Legumes,” “Crops,” and “Agriculture.”

By implementing a single level of a hierarchical metadata field into SharePoint Online’s search filters, users are still able to benefit from the ability to narrow search results by concepts such as “Topic” without becoming overly burdened by a long list of terms. In this way, your organization can use this method to help strike the balance between employing SharePoint’s native capabilities and ensuring that users have an optimized search experience.

Search Filter Alternatives: SharePoint Web Part Solution

If you are able to go beyond SharePoint Online’s out-of-the-box capabilities, however, it is worth considering whether translating the full hierarchy of a metadata field into search filters would provide users with enough value to warrant the purchase of a license for a web part solution that enables the nesting of search filters. While SharePoint Online alone does not allow for metadata hierarchy or term relationships within search filters, there are web parts available, such as the Aequos Modern Data Visualizer web part solution (for SharePoint Online modern), that can be purchased and configured with your organization’s SharePoint Online instance. While these types of technical solutions often require some additional customization and development work, when implemented according to best practices, they can give your organization’s users the ability to utilize more dynamic filters that recognize term relationships with the employment of different drop down levels:

Conclusion

Strategically translating your taxonomy into search filters in SharePoint Online will enable your organization to empower users to self-serve within and across your SharePoint sites, while fueling content findability and discoverability that leads to increased productivity and job satisfaction. At Enterprise Knowledge, we have helped numerous clients optimize their SharePoint Online instances, partnering on projects that range from smaller scale department taxonomy designs to the design and development of full SharePoint Online environments. Interested in learning more? Contact us today.

The post Optimizing Your Taxonomy in SharePoint Online: Search Filters appeared first on Enterprise Knowledge.

If you are exploring ways your own enterprise can benefit from implementing AI capabilities, we can help! EK has deep experience in designing and implementing solutions that optimizes the way you use your knowledge, data, and information, and can produce actionable and personalized recommendations for you. Please feel free to contact us for more information.

The post What is the Roadmap to Enterprise AI? appeared first on Enterprise Knowledge.

• Findability – The most common use case for taxonomy is, as we call it findability. In short, making it fast, simple, and intuitive for an end user to find what they’re looking for, either through search, browse, or any combination thereof. Taxonomy plays a number of roles here, from driving site navigation/information architecture, to improving search weighting, to enabling filtering/faceting on search.
• Discoverability – Going beyond findability, discoverability is about making end users aware of information they weren’t necessarily seeking, thereby providing them more complete answers. This is often surfaced via push recommendations. The idea here is that, with consistent taxonomy applied as metadata on content, tools can recommend content with similar metadata, helping users to find more than they were initially seeking.

Both findability and discoverability translate to more information getting to the user, ideally faster and more completely. This means less time looking for information and more time acting on a complete set of information. Moreover, improved findability and especially discoverability translates to a greater awareness of the information that already exists within the enterprise, meaning users are less likely to waste time recreating information that already existed within the enterprise but of which they were unaware. An additional element of this is:

• Awareness and Alignment – When we’re consistently tagging not just our content, but also our people with a well-designed taxonomy, we’re creating a great view of the organization as a whole. This means users are more likely to discover content elsewhere in the organization similar to that which they’re working upon, as well as people within the organization that hold similar or sought after expertise. 

Improved awareness and alignment means that users within an organization are more likely to connect with other end users that can help them learn, complete their tasks, or develop new knowledge. This translates to improved collaboration and coordination, with traditional silos of knowledge breaking down and new enterprise communities of knowledge and learning developing. 

Over time, improved awareness and alignment results in greater upscaling of employees as they find and leverage people from whom they can learn more effectively, as well as improved innovation within the organization as more experts collaborate across geographic and organizational boundaries. This leads to:

• Standardization – Enterprise taxonomy can align disparate systems, people, and processes, helping the organization to better communicate, collaborate, and integrate.

Standardization can result in lower administrative burden and greater integration of different information stores and organizational groups. Different systems that leverage the same taxonomies can be more effectively integrated in search. In addition, a great value add to effective enterprise taxonomy is that these controlled vocabularies begin seeping into conversations and day to day language, meaning that the overall way that people describe what their needs are and what they’re doing becomes more consistent, again, enabling greater collaboration and clearer communication.

As an organization begins mastering their overall information management with taxonomy, a common outcome is:

• Understanding – As taxonomy is consistently applied to content as tags, an organization has a better understanding of their content. A well-designed taxonomy applied consistently to content will ensure an organization understands what their content is about, who its for, and ideally, how it is being used.

Greater understanding of an organization’s content means that the organization can be more strategic about the content they’re creating and maintaining. An organization that understand what their content is about and how it is being used can identify gaps in their own knowledge and proactively work to address those gaps. Moreover, understanding content can help an organization decide what is no longer of value and should be archived or dispositioned. This, in turn, reduces organizational overhead from maintaining content that shouldn’t be kept, and decreases organizational risk from keeping content that is old and outdated.

Though all of these taxonomy value propositions have held true over the decades, the most common conversation today is about:

• Artificial Intelligence Readiness – A well-designed enterprise taxonomy serves as a critical building block for an organization to design ontologies, a key element of Knowledge AI.

Organizations that are investing in taxonomy now will possess a distinct advantage in designing and establishing enterprise ontologies, opening the path to Knowledge AI and creating greater avenues to integrate their content, data, people, and everything else that matters to their business.

Still struggling to get started with taxonomies, unable to convince your leadership of their value, or ready to take the next steps in maturity to ontologies and AI? Give us a call and let’s get started.

The post The Value of Taxonomy: Why Taxonomy (Still) Matters appeared first on Enterprise Knowledge.

1. Does my organization have any documents that detail the role of product owners on search projects?
2. Have we previously done work for a client that utilizes this set of technologies or capabilities?
3. Given what I know of my customers, what solutions, services, or products should I offer them?
4. Who at my organization could help me complete this work or provide the input of a subject matter expert?

Knowledge Graphs allow organizations to answer all of these questions in one system, regardless of where the source information resides.

Use Case 1:  Search Application

The first three questions may be interpreted as search queries enhanced with some sweet facets. For example, an organization could set up a search portal that supports faceting on “Document” as a content type, “Product Owner” as a role, and “Search” as a topic. Almost every content management system has some flavor of search, so search isn’t limited to knowledge graphs. However, a knowledge graph enables an organization to perform searches across multiple platforms by:

• aggregating content and/or metadata into a graph database;
• enhancing the metadata through auto-tagging and classification; and
• providing links back to the source system.

By aggregating content from multiple systems, users can access information in one place and more easily synthesize information. Similar to Google’s results page, when a user searches a knowledge graph, the result is a link to the source system, making it easier to find and view organizational information.

The search experience can be enhanced even further with an organizational business taxonomy. A business taxonomy, applied consistently to all of an organization’s content, allows users to apply common filters to content and more efficiently refine search results from across previously siloed content sources. User questions are answered by providing one unified search system instead of many.

Use Case 2:  Expert Finder

The fourth question is a commonly requested solution – an expert finder. If we can pull content from multiple systems and store it in a graph database, then why not also store information about people? Every employee, author, and document reviewer in your organization is also represented in the knowledge graph. In knowledge graphs, we define relationships between people and content:

• Who was mentioned in the full-text?
• Who reviewed the content before it was published?
• Who is the author of the content?
• Who was on the team which produced the document?

How content is related to people is unique to each organization, but the result is a dataset that powers an expert finder.

We add a classification model on top of the graph database that analyzes each individual, determining which metadata values are most related to the individual. As an example exercise, go to the EK Knowledge Base and pick any topic from the facets on the left-hand side. You will notice that some authors appear more frequently depending on which topic you choose.

A classification model performs a similar analysis, but on a larger scale, allowing the system to automatically infer who is an expert based on metadata. Once the system identifies experts, it can quickly recommend who someone should talk to when they need help with a certain subject matter.

Summary

Knowledge Graphs are powerful tools that encourage search and discovery within an organization. Whether developing a search application or an organizational expert finder, contact Enterprise Knowledge and let’s discuss how we can build the right experience for you and your users.

The post Knowledge Graphs for Search and Discovery appeared first on Enterprise Knowledge.

• The differences between findability and discoverability;
• Why they’re equally important; and
• How they manifest in successful KM initiatives.

Findability v. Discoverability: What’s the Difference?

Findability is a term for the ease with which information can be found. It means that users can easily find content or information they assume is present on a website. As an example, if I’m looking on EK’s website for information about Agile Transformations, I can search for a specific term (agile transformation), and find information about the topic.

However, users rarely know all of the content on your website. A good knowledge management strategy also promotes discoverability, which involves making sure that new content or information can be found, even if the user(s) don’t know that it exists yet.

Let’s say that after clicking on and reading through the first search result (“Why Agile Fails When Organizations Try to ‘Go Agile’”), I’ve realized that, while this content addresses my initial search, I’d like to find out more.

In the image above, the right hand side of the page has lists of “Related Content” and “Related News”, which enable me as a user to further explore and discover other content that’s, for instance, about the same core topics, by the same author, or of the same type.

Findability and Discoverability for Improved User Experience

Findability and discoverability are important because they address two distinct needs of end users searching for the content they need. A user’s informational needs are not satisfied solely through finding content that they know exists, or exclusively through discovering information that they hadn’t known about previously. A user’s goals constantly change between finding specific information and discovering new information, and any well-thought-out KM strategy needs to accommodate both.

The findability/discoverability challenge is even more critical when considering intranets, knowledge bases, and other internal content/document management systems, where users may need to wade through thousands of documents to find what they’re seeking. Below are three ways that findability and discoverability manifest in KM initiatives.

Well-Designed Main Navigation

Helping your users navigate efficiently should always be a high priority. Simply relying on keyword search will not be enough. A well-designed and prominent main navigation menu acts as a map that directs your end users to the information they’re seeking. Here are several tips that will help ensure that your navigation menu is easily findable for end users:

• Keep it simple. Strip down your navigation menu to the bare minimum and keep only the menu items which direct users to the most sought-after content on your site.

• Give your main navigation menu the visual weight it deserves. Consider font and icon size, and the color contrast (especially as it relates to links within your navigation).
• Consider responsive design. Users increasingly access web content on mobile devices rather than on a desktop monitor. It’s essential to consider how your menu will look on a tablet, and a smartphone. Less screen space means less content, so be purposeful in deciding what information people really need to see in your navigation menu.

Facted Navigation

Faceted navigation is something that sounds complex, but is easy to recognize in practice. Faceted navigation means picking the facets (or filters) that a user will utilize to choose their own search path. For example, if I wanted to buy a pair of shoes, some facets that I would like to use to filter my search results could include shoe size, color, and heel height.

Determining what values to use as filters for your organization’s content can be tricky, because every organization has unique needs and situations which shape how their information should be organized. Through our business taxonomy design workshops, we help organizations to identify and prioritize these facets.

Related Content

Too often, websites miss opportunities to further engage their end users beyond their first click online. However, this isn’t necessarily due to the quality of your content, but rather the discoverability of your content.

One way to improve your content’s discoverability is by offering relevant links to content that is related to the original inquiry. While this process may seem straightforward, there are several tips to consider:

• Keep it simple. Simplicity reigns once again. It’s impossible to be completely accurate when deciding which links to share as related content, so it can be tempting to provide more to address all possibilities. This can easily overwhelm the end user, so cap the related links at five to seven.
• The level of specificity of related links depends on where the user is within the site. Offer a broader range of suggestions for posts on main pages, because they act as a bridge to content that end users may not have considered yet. On a topic-specific landing page, offer only related links which are relevant to that topic.
• Don’t disregard visual design. Typically, websites reserve the right side of a webpage for less important information (e.g., advertisements). As a result, it’s critical to avoid any visual elements similar to advertisements when designing your related content, because content is more likely to be ignored in that format.

Closing

Findability and discoverability are key outcomes for any comprehensive KM strategy, but improving both begins with keeping your users and their needs at the center of your efforts. In turn, you’ll ensure that your organization’s information is findable, manageable, and reusable for those who need it. Need help improving the findability and discoverability of your content? Contact Enterprise Knowledge to learn more.

The post Findability v. Discoverability appeared first on Enterprise Knowledge.

One of the most powerful ways we can explain the value of taxonomies is by discussing their business value and outcomes. We frequently invoke our concepts of findability and discoverability.

We leverage real world examples, often from eCommerce, to convey the value and outcomes from a well-designed and actualized taxonomy. Sites like Amazon, Zappos, and Home Depot have all designed extremely powerful faceted navigations, powered by taxonomies, to maximize the findability and discoverability of their products. Improved findability and discoverability improve business value by improving sales and customer satisfaction.

The concept of faceted navigation (also known as faceted search or browse) means that you:

1. Define a set of metadata fields (or attributes) that best describe your “products.”
2. Create a finite set of values (taxonomy terms, or controlled vocabulary) to discretely populate each of those metadata fields.
3. Document a set of rules for how each of those values will be applied to each of those fields for each “product.” For instance, which fields are mandatory, which fields can receive only one value from a taxonomy versus multiple fields, etc.
4. Create a simple user interface to surface your facets.

If done well, these concepts will allow an individual user to choose their own path by selecting the various values that fit their specific needs, finding everything that matches all of their criteria, but excluding that which doesn’t.

If, instead of being an award-winning, globally recognized consultancy in Agile Knowledge Management services, EK, were, for instance, an online candy store, I would:

1. Wear a Willy Wonka style suit to work every day; and
2. Want to ensure any visitor to our store could find exactly what matched all of their wants and needs (findability), and
3. Ensure any visitor would get to discover the broader range of products that met some or most of their criteria, helping them find something they wanted but didn’t know they wanted (discoverability).

We wouldn’t offer a single bucket of mixed candy. We’d offer our end users the ability to be the proverbial kid in a candy store and select any candy in the store which looks tempting. This, very basically, is the idea of faceting.

Instead of a random mix of candy, we would begin asking what our customers care the most about. In the case of our candy, some of the most important criteria might be:

• Type
• Color
• Flavor
• Wrapping
• Brand

Next, we would define a taxonomy of values for each of the fields we wanted to employ in our faceting:

• Type: Chocolate, Hard Candy, Gummy
• Color: Purple, Green, Blue
• Flavor: Lime, Apple, Blueberry, Raspberry, Grape, Chocolate
• Wrapping: Yes, No
• Brand: M&M, Jelly Belly, Wonka

Translating this into a clean and simple navigation would mean we’d prioritize the most critical taxonomy values. Do customers care more about flavor or color? If the answer is “flavor is more important,” then flavor gets more visual weight on the page. As a result, our end user, in one click, would be able to move from the single bucket of candy…

To seeing only candy of a particular color…

Or candy of a particular type…

What’s even more exciting about faceting, however, is that in just a couple clicks, our user can choose the candy of a particular color and a particular type…

In this way, a user can move from an unwieldy bucket of everything (which a bunch of candy they don’t want), to a bucket of their favorite candy. They can also discover some new sweets that meet the same criteria as their favorite candies which they didn’t even know existed..

Now, in truth, no matter how many of you wish you ran a candy store, you most likely do not. In fact, many of you aren’t selling anything at all, or at least aren’t responsible for the selling of your company’s products or services. Instead, many of you are purveyors of knowledge and information.

These same concepts of faceting can be applied to your organization’s information, either internally or externally, in order to drive the findability, discoverability,and overall usability of the content with your organization’s intranet, knowledge base, help desk, learning management system, or website.

The key to leveraging facets for content is similar to our candy example. Just as a shopper would want to filter candy by type, color, and flavor, a seeker of information in your organization’s content systems would want to filter content based on common facets such as  topic/subject, type/format, and source. EK’s own Knowledge Base provides a great example of this. Each organization has their own perspectives and priorities for how their own information should be faceted. Our business taxonomy design workshops help organizations to identify and prioritize these facets.

Faceting can be one of the best ways for you to ensure your people find and discover the content they need to do their job, complete their mission, and enjoy a user-friendly experience. As a result, your organization will better use and reuse the content you have, serving your employees, customers, and partners more effectively and efficiently.

If you need help ensuring this sweet experience for your organization, let us know.

The post Faceting is Sweet appeared first on Enterprise Knowledge.