TLDR: The paper introduces “boldsea,” an architecture that uses “executable ontologies” to model complex dynamic systems. It integrates event semantics with a dataflow architecture to overcome limitations of traditional Business Process Management (BPM) and object-oriented semantic technologies. The system uses a formal language (BSL) to define semantic models that directly execute business logic, allowing for runtime modification, temporal transparency, and a unified framework for data and logic.
In the realm of enterprise systems, a persistent challenge has been the disconnect between how knowledge is represented and how dynamic processes are executed. Traditional Business Process Management (BPM) systems, while efficient for predefined workflows, often struggle with rigidity and are detached from underlying data structures. Similarly, event-driven architectures (EDA) capture changes effectively but often treat events as unstructured data, hindering validation. Object-oriented semantic technologies like RDF/OWL excel at depicting domain knowledge but lack the inherent ability to model dynamic activities or drive process execution.
Addressing these limitations, Aleksandr Boldachev introduces a novel architecture called boldsea, which implements the paradigm of “executable ontologies.” This approach synthesizes event semantics with a dataflow architecture to create semantic models that are not merely static descriptions but dynamic structures directly controlling process execution. The core idea is to bridge the ‘semantic-process gap’ (where knowledge systems can’t manage processes, and process systems use semantically poor data) and the ‘structural–execution gap’ (asynchronous execution without semantics, or semantics without execution).
The Foundation: Event Semantics
At the heart of boldsea is a fundamental rethinking of activity modeling, moving away from the traditional separation of static objects and dynamic processes. Instead, all activity, including resources and actors, is viewed as a stream of semantically typed events. These events, linked by causal relationships, form a temporal graph that naturally models both states and transitions within a single executable ontology.
Each event is atomic and immutable, following a unified structure: Id, Base:Type:Value, Actor, Cause, Model, and Timestamp. The inclusion of the ‘Actor’ field ensures a subject-centric description, tying every fact to its source and enabling data provenance. The ‘Cause’ field links to preceding events that logically conditioned its generation, forming a directed acyclic graph (DAG). This structure inherently supports native parallelism, complete historical reconstruction, and causal analysis.
To ensure validity and executability, a two-level data structure is employed: Model Events act as templates or schemas, defining the semantics of individuals, restrictions on values, and conditions for execution. Reification Events are specific instances or ‘facts’ created according to these model event templates.
Dataflow Architecture: Executing Semantic Models
The key innovation is the application of dataflow architecture for executing these semantic models. Unlike imperative control-flow BPM systems that follow rigid, sequential steps, boldsea’s event models implement an asynchronous principle. Model events act as ‘operators,’ and reification events as ‘data.’ A model event is activated—generating a new event—not by an external command, but when its logical ‘Condition’ expression evaluates to true as necessary reification events appear in the graph. This creates a reactive system where changes cascade through the graph, ensuring native parallelism and eliminating the need for centralized coordination.
Business logic is expressed declaratively through ‘restricting properties’ (Restrictions) assigned to model events. Key properties include:
- Condition: A logical expression that must be true for a new reification event to be created.
- SetValue: Used for automatic value assignment, generating reification events with values based on calculations or graph queries.
- Permission: Defines access rights based on roles or dynamic conditions, enforcing security at the semantic level.
- SetDo: Initiates system acts like creating or editing individuals, managing the full lifecycle of objects.
The entire approach is formalized through the boldsea Semantic Language (BSL), a domain-specific language with a formal BNF grammar. BSL ensures predictable, verifiable, and deterministic execution, allowing models to run without compilation. It supports variables for accessing the event graph’s current state and queries for extracting semantically connected data.
The boldsea-engine and its Advantages
The boldsea-engine is the tool for semantic activity modeling, interpreting and validating BSL models, processing queries, and executing business logic via a subscription mechanism. Its architecture ensures reactivity, validation, and asynchronous business logic execution. The engine, along with an Integrated Development Environment (IDE), allows business analysts to design, create, and debug executable business applications without coding.
The advantages of this approach are significant: architectural flexibility and adaptability (runtime modification of models), complete temporal transparency (auditing, analytics, machine learning), and the unification of knowledge representation, data, and business logic within a single semantic framework. This eliminates the need for complex integration layers between disparate systems.
Also Read:
- Charting the Path to Self-Driving Science with AI Agents
- State Algebra: An Algebraic Framework for Propositional Logic
Challenges and Future Outlook
Despite its promise, the boldsea architecture faces challenges, primarily a cognitive barrier and learning curve for transitioning from object-oriented paradigms to event-semantic ones. Performance optimization for large temporal graphs is an ongoing area of research, though initial tests have shown no issues for standard processes. The approach may also be less effective for simple CRUD operations or high-frequency transactional systems where the event model could be redundant.
Future research directions include deep integration with Large Language Models (LLMs) for generating and validating semantic models and leveraging the temporal graph as external, long-term semantic memory. The technology also holds promise for building decentralized P2P networks and flexible multi-agent systems, as well as developing automatic verification tools to prove the correctness of business processes. For more details, you can refer to the original research paper: Executable Ontologies: Synthesizing Event Semantics with Dataflow Architecture.
In conclusion, boldsea represents a significant step towards a future digital environment where autonomous AI agents can interact through a shared, semantically rich, and cryptographically secure knowledge graph, moving beyond isolated programs to truly unified and dynamic systems.
