Trie Tree Node Merging for Memory and Retrieval Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional trie trees face issues with increased memory usage and retrieval time due to a large number of nodes when dealing with long keys, and the Patricia tree approach does not efficiently manage memory for keys with many characters.
Innovation Solution
A trie tree structure is implemented where nodes corresponding to predetermined characters are connected in a tree structure, allowing new character strings to be registered by tracing nodes based on their characters and adding new nodes only as necessary, with tag keys arranged in depth-first order to reduce memory usage and improve retrieval efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a conventional trie tree is used to store long keys, then the retrieval function is provided, but the number of nodes increases and memory usage increases
Solution Approach 1:
The patent merges multiple nodes that share common character sequences into single nodes. Specifically, when multiple keys share a common prefix, the patent creates one node for the common prefix rather than separate nodes for each key, thereby reducing the total number of nodes while preserving the retrieval function.
Solution Approach 2:
Each node in the patent is designed to serve multiple purposes: it stores character information, maintains child node pointers, and can represent shared prefixes for multiple different keys. This multi-functionality allows the tree to efficiently handle long keys without proportionally increasing node count.
2Reliability
If a conventional trie tree is used for long keys, then the retrieval function is provided, but the number of comparisons during retrieval increases and retrieval time increases
Solution Approach 1:
By merging nodes that represent common character sequences, the patent reduces the total number of node comparisons required during retrieval. When multiple keys share a prefix, the retrieval process traverses the shared prefix once rather than re-comparing it for each key.
Solution Approach 2:
The patent segments the key space into hierarchical levels based on character positions. Each node represents a specific character position and stores only the relevant child pointers, allowing the retrieval process to skip over irrelevant branches and reduce comparison operations.
3Quantity of substance
If a Patricia tree is used to reduce memory usage, then memory efficiency is improved, but a node is created for each predetermined character string and memory usage increases for keys with many characters
Solution Approach 1:
The patent applies different node creation strategies based on local characteristics of the key data. Instead of uniformly creating nodes for every character string, the patent creates nodes selectively based on whether characters form shared prefixes across multiple keys, adapting the structure to the local density of key patterns.
Solution Approach 2:
The patent changes the parameter for node creation from 'each character string' to 'shared character sequences'. This parameter change allows the tree to differentiate between unique and shared prefixes, creating nodes only where necessary to represent divergent key paths, thereby reducing overall node count for keys with many characters.
Data Source
AI summary
A retrieval device according to the present embodiment associates one tag key with one node and removes a node not having a tag key when a trie tree generating unit generates a trie tree. Moreover, because a tag key having a low priority is registered in a node close to a root node when the trie tree generating unit registers tag keys in the nodes of the trie tree, the retrieval device can narrow down a node region that is a comparison target when a trie tree searching unit performs a retrieval process.