Implementing Semantic Autoclassification at a Top-10 Accounting Firm.

About the Client

Company AE is among the top accounting firms in the United States. In the mid-2020s, Company AE engaged Iknow to develop its first enterprise-wide knowledge management strategy and roadmap, including the taxonomy foundation to be implemented in this project (see Iknow Case Study #218, Catching Up to the Giants: Building a Knowledge Management Foundation for a Top-10 Accounting Firm).

Industry Context

A taxonomy only creates value once it moves from a strategic artifact to working, automated classification that tags content consistently as people create and upload it — otherwise, even the best-designed taxonomy remains an unused reference document. Professional services firms generate enormous volumes of client deliverables, templates, and internal guidance daily, far faster than any manual tagging process could keep up. That’s why semantic autoclassification platforms like Progress Semaphore exist: they apply a taxonomy’s rules automatically as content is created or ingested, rather than relying on individual employees to tag content correctly and consistently by hand. Semaphore itself reflects this industry consolidation — originally built by Smartlogic, the platform was folded into MarkLogic and then Progress Software, becoming Progress Semaphore, a pattern common across the enterprise content intelligence market as larger technology vendors acquire specialized semantic tools.

Current Situation

This project was a direct continuation of Iknow’s earlier KM strategy and roadmap engagement with Company AE, which had identified the lack of a working auto-classification capability and enterprise search as core gaps and produced the underlying taxonomy the firm would need. With that strategy and taxonomy in place, Company AE engaged Iknow again to implement and test Progress Semaphore for auto-classification across its main knowledge repositories, moving the firm from a documented recommendation to a functioning system.

• A taxonomy that existed on paper but not yet in production. Company AE had an 11-facet, roughly 3,200-term taxonomy from the earlier KM strategy engagement, but it wasn’t yet operationalized in any classification or search system.
• Terminology overlap and ambiguity across business domains. Consulting, audit, tax, and wealth advisory terminology overlapped and duplicated across Company AE’s service lines, requiring reconciliation before automated rules could be applied consistently.
• No proven, tested classification rules. Semaphore’s out-of-the-box rule templates existed but hadn’t been customized or validated against Company AE’s actual documents and business requirements.
• Deployment into a live, in-use SharePoint Online environment. Any new autoclassification capability had to integrate with Company AE’s existing SharePoint term store and site structure without disrupting day-to-day use.

• Further develop Company AE’s KM taxonomy to enable automated classification in Semaphore.
• Customize and test Semaphore’s classification rules with real Company AE content until achieving satisfactory tagging results.
• Support the initial deployment of Semaphore in Company AE’s SharePoint Online environment.
• Train Company AE’s KM team to independently maintain and extend the taxonomy and classification rules.

How Iknow Structured the Work

Iknow organized the six-month engagement around three major tasks — taxonomy development, classification rule customization and testing, and SharePoint deployment support — with hands-on training for Company AE’s KM team integrated throughout.

Key Activities & Decisions

• Taxonomy development. Iknow built on earlier categorizations of Company AE’s internal departments, business processes, industries, service lines, and locations, adding new facets for document types, certification types, IT applications, legal terminology, and cross-cutting consulting, accounting, tax, and wealth advisory topics. Key steps included deduplicating terminology across business domains, adding synonyms, acronyms, and other alternative labels (including negative evidence for concepts expressed in many different ways), minimizing pre-coordination, using the minimal number of facets consistent with likely search use cases, and disambiguating overly generic terms such as “Assets” with more specific modifiers or child terms.
• Classification rule customization and testing. Iknow modified Semaphore’s rule templates to fit Company AE’s business requirements, then iteratively tested the rules against more than 100 documents from across the business, repeating rounds of refinement until classification results were satisfactory.
• SharePoint deployment support. Iknow provided scripts to set up SharePoint site columns corresponding to the Semaphore facets and performed the initial synchronization of the Semaphore taxonomy with the SharePoint term store, enabling autoclassification within SharePoint Online.
• KM team training. Iknow delivered several hands-on online sessions covering taxonomy editing and maintenance, use of the classification rule templates and publishing, and governance topics such as using tasks to manage ongoing development.

Stakeholders & Collaboration

Iknow served as prime contractor, continuing its relationship with Company AE’s KM team established during the earlier strategy engagement and embedding hands-on training throughout to enable the firm’s IT staff to take over maintenance.

Reconciling Overlapping Terminology Across Very Different Business Lines

Audit, tax, consulting, and wealth advisory terminology overlapped and duplicated in ways that would have undermined automated classification if left unresolved. Iknow addressed this by treating taxonomy quality as a prerequisite for classification quality: deliberately deduplicating terms across domains, adding synonyms and negative evidence, minimizing pre-coordination, and disambiguating overly generic terms with modifiers or child terms before any classification rule was built.

Getting Classification Rules to Work on Real Content, Not Just in Theory

Rule templates that looked correct on paper often produced inconsistent results when applied to Company AE’s actual, varied document base spanning multiple service lines. Iknow addressed this by iteratively testing the customized rules against more than 100 real documents from across the business, repeating rounds of refinement until results were genuinely satisfactory rather than accepting a single test pass.

Quantitative Outcomes

• Taxonomy expansion: New facets for document types, certification types, IT applications, legal terminology, and cross-cutting topics were added, building on the existing 11-facet, roughly 3,200-term foundation.
• Testing: 100+ documents from across the business were tested repeatedly to validate classification rules.
• Deployment: SharePoint site columns and term store synchronization were delivered to enable autoclassification in SharePoint Online.
• Training: Multiple hands-on online sessions were delivered to Company AE’s KM team.

Qualitative Outcomes

Company AE moved from a KM strategy and taxonomy on paper to a working, tested auto-classification capability deployed across selected repositories. Company AE’s KM team gained hands-on skills to maintain and extend the taxonomy and classification rules themselves, rather than depending indefinitely on outside consultants. Company AE continued refining the rules and taxonomy after the engagement ended to ensure consistent classification across the rest of the firm’s subject domains. The engagement represents the direct continuation of Iknow’s initial KM strategy and roadmap project into working technology — concrete evidence that Company AE’s KM investment moved from strategic recommendation to practical execution.

Timeline to Impact

Iknow delivered the engagement within the six-month contract period, with Company AE’s trained team continuing the rollout and refinement across additional repositories after the engagement’s close, extending the value of the earlier strategy investment into ongoing operational practice.

Core Skills

• Semantic autoclassification
• Taxonomy refinement & disambiguation
• SharePoint enterprise search deployment
• KM training & enablement

Methods & Frameworks

• Iterative classification rule testing
• Taxonomy de-duplication and disambiguation methodology
• SharePoint term store synchronization

Technologies & Tools

• Progress Semaphore
• SharePoint Online (site columns, term store)
• Classification rule templates