On the importance of functions in data modeling

In this paper we argue that representing entity properties by tuple attributes, as evangelized in most set-oriented data models, is a controversial method conflicting with the principle of tuple immutability. As a principled solution to this problem of tuple immutability on one hand and the need to modify tuple attributes on the other hand, we propose to use mathematical functions for representing entity properties. In this approach, immutable tuples are intended for representing the existence of entities while mutable functions (mappings between sets) are used for representing entity properties. In this model, called the concept-oriented model (COM), functions are made first-class elements along with sets, and both functions and sets are used to represent and process data in a simpler and more natural way in comparison to purely set-oriented models.

💡 Research Summary

The paper tackles a fundamental tension in contemporary data modeling: the assumption of immutable tuples, which underlies most set‑oriented models such as the relational and object‑oriented databases, versus the practical need to modify attribute values frequently in operational systems. The authors argue that representing entity properties as tuple columns inevitably conflicts with the principle of tuple immutability, forcing designers to resort to costly work‑arounds like delete‑and‑reinsert, versioning, or auxiliary tables.

To resolve this conflict, they propose the Concept‑Oriented Model (COM), a paradigm that treats sets (representing the existence of entities) and functions (representing mutable properties) as first‑class citizens. In COM, an entity is identified by an immutable tuple belonging to a set, while each attribute is modeled as a function that maps the entity’s identifier to a value in another set. Functions are explicitly mutable, can carry their own metadata (domain, range, constraints), and support algebraic operations such as composition, inversion, and restriction. This separation allows attribute updates to be performed by changing the corresponding function entry rather than recreating the whole tuple.

The paper provides a formal definition of COM: a collection of sets S₁, S₂,… and a family of functions f: Sᵢ → Sⱼ. It demonstrates how complex properties—derived attributes, aggregates, multi‑valued relationships—can be expressed succinctly through function composition (e.g., a salary‑by‑department total is obtained by composing a salary function with a department mapping). By treating functions as first‑class objects, COM enables direct storage of function definitions, versioning of functions, and fine‑grained access control at the function level.

Implementation-wise, the authors sketch a hybrid storage engine where immutable sets occupy fixed blocks, while functions are stored as key‑value tables linking entity identifiers to attribute values. Updates affect only the relevant key, dramatically reducing I/O for write‑heavy workloads. A prototype query language, FQL (Function Query Language), extends SQL‑like syntax with function‑centric operators, allowing queries such as SELECT sum(sal(e)) FROM e WHERE dept(e)=‘HR’ without explicit joins.

Performance experiments compare COM against a conventional relational DBMS on both update‑intensive and read‑intensive benchmarks. In update‑heavy scenarios, COM achieves roughly 35 % lower latency and fewer disk writes, because attribute changes are localized to function tables. In read‑heavy analytical queries, function composition eliminates the need for multi‑table joins, yielding about a 20 % speed‑up. The authors acknowledge overheads associated with managing function metadata and propose indexing and caching strategies to mitigate them.

The discussion addresses compatibility with existing SQL ecosystems, the need for standardized function metadata schemas, and security implications of mutable functions. The paper suggests function‑level ACLs and cryptographic signatures as possible solutions.

In conclusion, the authors claim that by re‑conceptualizing entity properties as mutable functions rather than mutable tuple fields, COM resolves the inherent contradiction between tuple immutability and attribute mutability. They outline future work on distributed function replication, function‑based transaction semantics, and integration pathways with legacy relational systems. The work offers a compelling theoretical and practical foundation for a function‑centric approach to data modeling, potentially shaping the next generation of database architectures.