Skip to content
STIMSMITH

Architectural Knowledge

Concept WIKI v1 · 7/9/2026

Architectural Knowledge is a concept denoting information about the structure, behavior, and constraints of a system's architecture. It is used in both software engineering, where it is distributed across heterogeneous artifacts such as issues, requirements, design diagrams, and code, and in hardware verification, where it constitutes a subset of design knowledge derivable from a Design Under Test (DUT) specification, including instruction formats, registers, exception behavior, and translation mechanisms.

Architectural Knowledge

Definition

Architectural Knowledge is information that describes the architecture of a system in terms of its structural elements, behavioral rules, and contextual constraints. The concept is used in two principal domains reflected in the evidence:

  • Software architecture: knowledge-centric information about a software system's architecture, distributed across heterogeneous artifacts such as requirements documents, design diagrams, source code, issue trackers, and documentation.
  • Hardware verification: design knowledge of a Design Under Test (DUT) that can be obtained by reviewing the DUT's specification, including instruction format and semantics, registers, exception behavior, and translation mechanisms.

Software Architecture Context

In software engineering research, Architectural Knowledge is treated as knowledge-centric by nature and is recognized to be distributed across heterogeneous software artifacts, which makes it difficult for developers to access and utilize effectively [arxiv:2601.19548].

Empirical studies of large Apache projects have investigated how Architectural Knowledge concepts are textually represented in issue trackers such as Jira. Such concepts include architectural component behavior and contextual constraints, and they may appear in issues in linguistic forms such as adjectives. The studies also examined which Architectural Knowledge concepts commonly occur in issues and which concepts tend to co-occur, with the goal of supporting empirically-grounded search tools for identifying these concepts in issues for future reuse [arxiv:2106.11140].

A broader research vision proposes an automated pipeline that systematically extracts Architectural Knowledge from diverse artifacts, links them, identifies and resolves inconsistencies between artifacts, and consolidates this knowledge into a structured knowledge base. The envisioned pipeline supports architecture conformance checking, change impact analysis, and natural language question-answering, with specialized extractors per artifact type, a unified knowledge representation schema, consistency checking mechanisms, and retrieval-augmented generation for conversational access [arxiv:2601.19548].

Hardware Verification Context

In the hardware verification domain, "design knowledge" is an umbrella term that includes both architectural knowledge and micro architectural knowledge. Architectural Knowledge of a DUT is design knowledge that can be manually obtained by reviewing the specification of the DUT [patent:US8683282].

Architectural Knowledge specifically includes elements such as:

  • Instruction format and semantics
  • Registers
  • Exception behavior
  • Translation mechanisms

It is contrasted with Micro Architectural Knowledge, which is implementation-specific design knowledge (for example, a model of execution time per instruction or conditions for pipeline flushes). For Micro Architectural Knowledge, reviewing the specification is generally not sufficient; the DUT itself must be reviewed [patent:US8683282].

Automated Identification

Automated techniques have been proposed to identify Architectural Knowledge without requiring a specification designer to manually intervene. In one approach, machine learning classification is applied to data collected while generating instructions for a DUT, producing classification output that is reviewable by a human user. The classified data is then expressed as design knowledge in a predetermined formal language usable by verification tools. As a technical effect, Architectural Knowledge can be determined from an early stage of verification—once an architectural reference model can execute instructions—even before a final formal definition of the DUT is available [patent:US8683282].

Relationship to Other Concepts

  • Design Knowledge — Architectural Knowledge is a constituent category of the broader notion of Design Knowledge; the latter is defined as any information considered as either micro architectural knowledge and/or Architectural Knowledge [patent:US8683282].
  • Register — Registers are among the explicit elements that Architectural Knowledge describes in the hardware verification sense [patent:US8683282].
  • Exception Behavior — Exception behavior is among the explicit elements that Architectural Knowledge describes in the hardware verification sense [patent:US8683282].

See Also

LINKED ENTITIES

3 links
Register MENTIONS
Exception Behavior MENTIONS
Design Knowledge PART_OF

CITATIONS

7 sources
7 citations
[1] Software architecture is inherently knowledge-centric, with Architectural Knowledge distributed across heterogeneous software artifacts such as requirements documents, design diagrams, code, and documentation. From Scattered to Structured: A Vision for Automating Architectural Knowledge Management
[2] Architectural Knowledge concepts such as architectural component behavior and contextual constraints are textually represented in issues (e.g. as adjectives), and empirical studies analyzed which concepts commonly occur and co-occur in Jira issues of large Apache projects. An Exploratory Study on Architectural Knowledge in Issue Tracking Systems
[3] Design Knowledge is defined as any information considered as either micro architectural knowledge and/or Architectural Knowledge, with Architectural Knowledge being specifiable from review of a DUT specification. Automatic identification of information useful for generation-based verification
[4] Architectural Knowledge includes instruction format and semantics, registers, exception behavior, and translation mechanisms. Automatic identification of information useful for generation-based verification
[5] Micro Architectural Knowledge is implementation-specific design knowledge (e.g., execution time per instruction, pipeline flush conditions) that typically requires reviewing the DUT itself rather than just the specification. Automatic identification of information useful for generation-based verification
[6] Architectural Knowledge can be automatically identified using machine learning classification applied to instruction-generation data, and expressed in a predetermined formal language usable by verification tools. Automatic identification of information useful for generation-based verification
[7] Learning Architectural Knowledge can commence at early verification stages once an architectural reference model can execute instructions, even before a final formal definition of the DUT is available. Automatic identification of information useful for generation-based verification