Symbolic priors for recurrent neural network based semantic parsing
AssigneeXerox Corporation
InventorsChunyang Xiao, Marc Dymetman
Life of this patent
Estimating the term…
Abstract
Methods, systems, and devices for semantic parsing. In an example embodiment, a method for semantic parsing can include steps, operations, or instructions such as obtaining a data pair for learning, the data pair comprising logical form data and natural utterance data; acquiring grammar for targeted logical forms among the logical form data of the data pair; modeling data comprising other available prior knowledge utilizing WFSA (Weighted Finite State Automata); combining with the targeted logical forms with the data modeled comprising the other available prior knowledge to form a background; and exploiting the background on the data pair. Note that we do not “learn” the background, but “learn” the background-RNN (Recurrent Neural Network).
Claims20
A method for semantic parsing, said method comprising:
- obtaining a data pair for learning, said data pair comprising logical form data and natural utterance data;
- acquiring grammar for targeted logical forms among said logical form data of said data pair;
- modeling data comprising other available prior knowledge using WFSA (Weighted Finite State Automata);
- combining said targeted logical forms with said data to form a background after said modeling of said data comprising said other available prior knowledge;
- and
- automatically learning a background-RNN (Recurrent Neural Network) based on said data pair, wherein said background-RNN comprises a combination of an RNN with a background.
A system for semantic parsing, said system comprising:
- at least one processor;
- and
- a non-transitory computer-usable medium embodying computer program code, said computer-usable medium capable of communicating with said at least one processor, said computer program code comprising instructions executable by said at least one processor and configured for:
- obtaining a data pair for learning, said data pair comprising logical form data and natural utterance data;
- acquiring grammar for targeted logical forms among said logical form data of said data pair;
- modeling data comprising other available prior knowledge;
- combining said targeted logical forms with said data to form a background after said modeling of said data comprising said other available prior knowledge using WFSA (Weighted Finite State Automata);
- and
- automatically learning a background-RNN (Recurrent Neural Network) based on said data pair, wherein said background-RNN comprises a combination of an RNN with a background.
A non-transitory processor-readable medium storing computer code representing instructions to cause a process for semantic parsing, said computer code comprising code to:
- obtain a data pair for learning, said data pair comprising logical form data and natural utterance data;
- acquire grammar for targeted logical forms among said logical form data of said data pair;
- model data comprising other available prior knowledge using WFSA (Weighted Finite State Automata);
- combine said targeted logical forms with said data to form a background after said modeling of said data comprising said other available prior knowledge;
- and
- automatically learn a background-RNN (Recurrent Neural Network) based on said data pair, wherein said background-RNN comprises a combination of an RNN with a background.
Dependencies are inferred from the claim text.