[Paper Review] Using rational numbers to key nested sets

Dan Hazel|ArXiv.org|Jun 19, 2008

History and Theory of Mathematics3 references3 citations

TL;DR

This paper proposes a novel method for encoding hierarchical tree structures in relational databases using rational numbers as node keys, enabling efficient insertion without rebalancing. By deriving child keys from parent keys via finite continued fractions, it ensures that every node sorts before its descendants and after siblings with lower indices, with reversible parent-child key relationships and provable ordering guarantees.

ABSTRACT

This report details the generation and use of tree node ordering keys in a single relational database table. The keys for each node are calculated from the keys of its parent, in such a way that the sort order places every node in the tree before all of its descendants and after all siblings having a lower index. The calculation from parent keys to child keys is simple, and reversible in the sense that the keys of every ancestor of a node can be calculated from that node's keys without having to consult the database. Proofs of the above properties of the key encoding process and of its correspondence to a finite continued fraction form are provided.

Motivation & Objective

• To address the scalability issue of traditional nested set models that require costly rebalancing after insertions.
• To design a key encoding system using rational numbers that preserves hierarchical ordering without modifying existing node keys.
• To ensure that node keys derived from finite continued fractions maintain correct ancestor-descendant and sibling ordering.
• To provide a reversible key system where parent keys can be reconstructed from any node’s key without database lookup.
• To formally prove the correctness and ordering properties of the proposed rational key encoding.

Proposed method

• The method uses finite continued fractions to generate unique rational keys for each node based on its path in the tree, with each node's key derived from its parent's key.
• Child node keys are computed using a recurrence relation involving numerators and denominators that maintain strict ordering between parent, child, and next sibling.
• The key generation process is reversible: ancestor keys can be reconstructed from a node’s key using the same recurrence, eliminating the need for database lookup.
• The method leverages properties of continued fractions to ensure that every child’s key lies strictly between its parent’s key and its parent’s next sibling’s key.
• The encoding is formalized using matrix multiplication of transition matrices representing each level’s child count, ensuring mathematical consistency.
• Theoretical proofs are provided to verify that the key ordering satisfies the required nested set properties, including ancestor-descendant and sibling ordering.

Experimental results

Research questions

• RQ1 Can rational number keys be used to represent hierarchical tree structures in a single relational table while avoiding the need for rebalancing after insertions?
• RQ2 Does the proposed key encoding ensure that every node appears before its descendants and after lower-indexed siblings in sort order?
• RQ3 Can the parent keys of any node be reconstructed from its key without accessing the database?
• RQ4 Is the rational key encoding based on finite continued fractions mathematically consistent and provably correct for hierarchical ordering?
• RQ5 How does this method compare to prior approaches like Tropashko’s encoding in terms of ordering behavior and uniqueness?

Key findings

• The proposed method successfully encodes tree nodes using rational numbers derived from finite continued fractions, ensuring correct hierarchical ordering.
• The key generation process is reversible: parent keys can be computed from any node’s key using only the node’s key and the path information, without querying the database.
• Every child node’s key lies strictly between its parent’s key and its parent’s next sibling’s key, guaranteeing correct sort order.
• The method proves that the rational key of any node is less than that of its next sibling, ensuring correct sibling ordering.
• Theoretical proofs confirm that the encoding maintains the nested set property: for any node, all ancestors have keys that bound the node’s key, and all descendants have keys that lie strictly within the bounds of their ancestors.
• The approach avoids the need for rebalancing during insertions, as there are infinitely many rational numbers between any two keys, providing ample space for new nodes.