The Philosophy of 1.5hop+: Meaning-Oriented Concept Neighborhoods
In BoK (Body of Knowledge), knowledge from development projects is accumulated as knowledge graphs and knowledge vectors. When transferring knowledge to generative AI via RAG (Retrieval-Augmented Generation), searching the knowledge graph becomes necessary. In this context, we examine 1.5hop+ as the search approach.
1.5hop+ is a philosophy for constructing concept neighborhoods based not on fixed hop counts, but on semantic structures essential for understanding a concept.
While grounded in the expressive power of RDF and OWL, this approach goes beyond mere graph traversal to enable model-based semantic understanding.
In SimpleModeling, an object model derived from the UML (Unified Modeling Language) metamodel is adopted, providing a structured framework for models. By leveraging this structure, 1.5hop+ can be made practical.
Limitations of Fixed-Hop Traversal
In RDF / OWL representations, the meaning of a concept is not confined to a single node, but emerges from a structure composed of multiple nodes and relations.
For example, understanding a class requires not only its properties, but also the data types of those properties and the constraints applied to their value ranges.
These structures naturally extend beyond two hops in RDF graphs, and fixed hop limits often omit information essential for semantic understanding.
Basic Principles of 1.5hop+
1.5hop+ determines exploration scope not by traversal distance, but by whether an element is essential for understanding the concept.
The basic principles are as follows:
-
Always include directly related elements (1-hop)
-
Include type definitions, value ranges, and constraints when they are semantically inseparable
-
Avoid unbounded traversal by enforcing a safe upper limit
The “+” indicates that exploration depth is adaptive and that meaning takes precedence over distance.
Concept Structures Targeted by 1.5hop+
A typical concept structure expanded by 1.5hop+ looks like this:
Types and constraints under properties are treated not as peripheral information, but as defining components of the concept itself.
Significance of Being CML-Based
The primary reason 1.5hop+ works in practice lies in the use of CML for model definition.
CML is a language derived from the UML metamodel, where elements such as classes, properties, types, and generalizations are defined with explicit semantic roles.
This makes it possible to mechanically determine which nodes are essential for concept understanding and which relations are indispensable for type comprehension.
Affinity with Generative AI
For generative AI, what matters is not node distance, but the semantic roles played by each element.
A concept neighborhood constructed with 1.5hop+ provides sufficient semantic structure for AI to reason about why constraints exist and what types represent.
Positioning in SimpleModeling / SIE
In SimpleModeling and the Semantic Integration Engine (SIE), 1.5hop+ is adopted as a fundamental exploration unit for concept understanding.
The explainConcept and graph search APIs do not require callers to specify hop counts; instead, they internally perform exploration based on concept neighborhoods defined by 1.5hop+.
Summary
Exploring knowledge graphs often sacrifices performance when generality is prioritized.
If the nature of the knowledge being handled can be identified, it becomes possible to balance accuracy and performance by leveraging its characteristics in exploration.
1.5hop+ was developed based on this perspective as a method to enable smooth operation of BoK within SimpleModeling.
References
Glossary
- BoK (Body of Knowledge)
-
At SimpleModeling, the core knowledge system for contextual sharing is called the BoK (Body of Knowledge). The goal of building a BoK is to enable knowledge sharing, education, AI support, automation, and decision-making assistance.
- Retrieval-Augmented Generation (RAG)
-
A generation technique that supplements a language model’s internal (parametric) knowledge by retrieving relevant external information before generation. RAG systems first search knowledge sources such as databases or knowledge graphs and then use the retrieved context as input for text generation.
- knowledge graph
-
A semantic graph-based knowledge base where nodes represent entities or concepts and edges represent their relationships.
- RDF
-
A W3C-standardized data model that represents information as subject–predicate–object triples.
- UML (Unified Modeling Language)
-
A standardized modeling language for object-oriented analysis and design. It represents system structures and behaviors through diagrams such as class, sequence, and use case diagrams. Serves as the foundational language for UP and CBD.
- CML (Cozy Modeling Language)
-
CML is a literate modeling language for describing Cozy models. It is designed as a domain-specific language (DSL) that forms the core of analysis modeling in SimpleModeling. CML allows model elements and their relationships to be described in a narrative style close to natural language, ensuring strong compatibility with AI support and automated generation. Literate models written in CML function as intermediate representations that can be transformed into design models, program code, or technical documentation.
- Semantic Integration Engine (SIE)
-
An integration engine that unifies structured knowledge (RDF) and document knowledge (SmartDox) derived from the BoK, making them directly accessible to AI.