Finite-State Transducer Compression for Redundant Data Elimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for compressing finite-state transducer (FST) data result in redundant data, leading to inefficient use of memory resources due to the inclusion of unnecessary data elements, which increases the waste of data space.
Innovation Solution
A method and device for compressing FST data by decomposing state transition and state data into fine-grained categories, sequentially arranging the decomposed data, and performing classification statistics to reduce redundancy, thereby optimizing the data structure and minimizing space usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the final weight of the final state is wrapped as a weight of the state transition to generate compressed data in a unified format, then the data format consistency is improved, but the data space waste increases due to packaging non-existent next state identifier and weight data
Solution Approach 1:
The patent segments the FST data into distinct components: state transition data (including signal label, weight, and next state identifier) and state data (including final weight). By decomposing the unified format into separate segments, the patent eliminates the need to package non-existent data elements, thereby reducing data space waste while maintaining operational ease through structured organization.
Solution Approach 2:
The patent extracts and removes redundant data elements (non-existent next state identifier and weight) from the compressed data structure. By taking out only the essential components needed for FST representation, the patent achieves space efficiency without sacrificing data format consistency, as the extracted essential data is organized in a structured manner.
2Device complexity
If a unified compressed data format is used by wrapping final weight into state transition weight, then the compression process is simplified, but the memory resource efficiency deteriorates due to redundant data packaging
Solution Approach 1:
The patent divides the compression process into segmented operations: decomposing state transition data into signal label, weight, and next state identifier components; decomposing state data into final weight components; and separately organizing these segments. This segmentation simplifies the compression process by providing clear operational steps while improving memory resource efficiency by eliminating redundant data packaging.
Solution Approach 2:
The patent performs preliminary decomposition of state transition data and state data into their respective components before compression. By preparing the data in advance with clear structural separation, the patent simplifies the subsequent compression process and ensures optimal memory resource efficiency, as the preliminary organization prevents redundant data inclusion.
3Stability of the object's composition
If redundant data elements are included to maintain unified data format, then the data structure consistency is improved, but the space occupation increases
Solution Approach 1:
The patent segments the data structure into distinct state transition data and state data components, each with their own organized elements. This segmentation maintains data structure consistency through clear structural boundaries and relationships while eliminating the need for redundant data elements, thereby reducing space occupation.
Solution Approach 2:
The patent applies local quality by organizing different data components (signal label, weight, next state identifier, final weight) with their specific structures and relationships appropriate to their function. This localized organization maintains overall data structure consistency while avoiding universal redundant packaging, thus reducing space occupation.
Data Source
AI summary
A method and device for compressing FST data are provided. The method includes: acquiring to-be-compressed FST data, where the FST data includes state transition data and state data; decomposing the state transition data based on first data categories to acquire first decomposition data; decomposing the state data based on second data categories to acquire second decomposition data; sequentially arranging, for each of the first data categories, the first decomposition data of the first data category, to acquire first arrangement data of the first data category; alternately arranging the first arrangement data and the second decomposition data according to a sequential order used in the first arrangement data, to acquire second arrangement data; performing classification statistics on the first arrangement data and the second arrangement data to acquire index data; and combining the first arrangement data, the second arrangement data, and the index data, to obtain the compressed FST data.