Search Device Handling Empty Transitions in Weighted Automata
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search devices face difficulties in finding non-duplicative input symbol strings from weighted finite state automata (WFSA) with empty input symbols, which complicates the output of multiple plausible results in pattern recognition and other applications.
Innovation Solution
A search device is designed to retrieve and expand hypotheses in a WFSA, handling empty transitions and generating new hypotheses to find non-duplicative input symbol strings by following empty transitions and maintaining cumulative weights, ensuring the output of unique paths with the best cumulative weights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If N number of paths are found in order of most suitable weights from a non-deterministic WFSA, then the plurality of paths can be obtained, but the paths having different input symbol strings may be smaller in number than N, making it difficult to output N non-duplicative input symbol strings
Solution Approach 1:
The patent segments the search process into two distinct phases: first finding all paths in order of most suitable weights, then filtering and organizing them to extract non-duplicative input symbol strings. This segmentation allows the system to handle the complexity of finding unique paths by breaking down the problem into manageable steps, ensuring N non-duplicative strings can be output even when the total number of paths is large.
Solution Approach 2:
The patent performs preliminary action by first finding all N paths in order of most suitable weights before filtering for non-duplicative input symbol strings. This preliminary search ensures that the system has access to all candidate paths, and then applies deduplication logic to extract the required N unique input symbol strings, solving the problem of insufficient unique paths.
2Adaptability or versatility
If a WFSA includes empty (ε) input symbols, then the automaton can represent complex patterns, but it becomes difficult to find out a plurality of non-duplicative input symbol strings
Solution Approach 1:
The patent extracts and separates the handling of empty (ε) transitions from the main path-finding process. By identifying and processing empty transitions explicitly, the system can manage their特殊性 without complicating the overall algorithm. This extraction allows the patent to maintain adaptability for complex patterns while reducing the operational complexity of handling empty symbols.
Solution Approach 2:
The patent introduces an intermediary mechanism to handle empty transitions - a dedicated processing step that manages ε-symbols separately before integrating results into the main path-finding workflow. This intermediary approach isolates the complexity of empty transitions, allowing the rest of the system to operate with standard input symbols while still supporting complex pattern recognition.
3Loss of information
If N number of results are required with different input symbol strings, then meaningful selection candidates can be provided, but the search complexity increases when dealing with non-deterministic WFSA and empty transitions
Solution Approach 1:
The patent implements feedback by continuously monitoring the uniqueness of input symbol strings during the search process. When duplicate strings are detected, the system adjusts its search strategy to find alternative paths, ensuring N unique results are obtained. This feedback mechanism maintains result uniqueness without requiring a complete redesign of the search algorithm, balancing information quality with computational complexity.
Data Source
AI summary
According to an embodiment, a search device searches paths of a digraph and includes a retriever, and an expander. The retriever is configured to, from among hypotheses stored in a storage, retrieve, as a target hypothesis, a single hypothesis for which a weight obtained by addition of a weight of an already-searched path corresponding to each hypothesis and a weight of the best path from a state of the head of concerned path to a final state is the best weight. The expander is configured to, when the retrieved target hypothesis is not a final hypothesis for which the search has been completed up to a final state, generate new hypotheses each holding an input symbol string that is present in a path in which the search has been performed from states held by the target hypothesis until finding a single input symbol, and write the generated hypotheses in the storage.