Contextual filtering
First Claim
1. A contextual filter for use within a computer-based speech recognition system comprising:
- a shallow parser input for receiving a representation of an utterance into a shallow parser component in a form of a first word lattice, the first word lattice including a set of potential textual word sequence matches for the utterance, where the textual word sequence matches include at least one word;
the shallow parser component operative for receiving and analyzing the first word lattice to produce a second word lattice, and including a filter selected from a group consisting of a sub-categorization licensing filter, a predication licensing filter, an adverbial licensing filter, and a discourse licensing filter, where the filter assigns a probability match for a potential word match in relation to another potential word match in the utterance based on a particular filter-specific criteria; and
a shallow parser output for outputting the probability matches from the shallow parser component as a portion of a second word lattice for further processing to determine most likely combinations of words present in the utterance.
1 Assignment
0 Petitions
Accused Products
Abstract
The present invention relates to techniques for contextual filtering for improving an output of a speech recognizer. The techniques comprise receiving a representation of a speech utterance into a shallow parser component in the form of a first word lattice including a set of potential language matches for the utterance, where the language matches include at least one word; the parser component operative for receiving and analyzing the first word lattice to produce a second word lattice, and including a filter that assigns a probability match for a potential word match in relation to another potential word match in the utterance based on a particular filter-specific criteria; and outputting the probability matches from the shallow parser component as a portion of a second word lattice for further processing to determine most likely combinations of words present in the speech utterance.
-
Citations
42 Claims
-
1. A contextual filter for use within a computer-based speech recognition system comprising:
-
a shallow parser input for receiving a representation of an utterance into a shallow parser component in a form of a first word lattice, the first word lattice including a set of potential textual word sequence matches for the utterance, where the textual word sequence matches include at least one word; the shallow parser component operative for receiving and analyzing the first word lattice to produce a second word lattice, and including a filter selected from a group consisting of a sub-categorization licensing filter, a predication licensing filter, an adverbial licensing filter, and a discourse licensing filter, where the filter assigns a probability match for a potential word match in relation to another potential word match in the utterance based on a particular filter-specific criteria; and a shallow parser output for outputting the probability matches from the shallow parser component as a portion of a second word lattice for further processing to determine most likely combinations of words present in the utterance. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14)
-
-
15. A method for contextual filtering, the method operating on a computer system including a processor, an input coupled with the processor for receiving an input, and an output coupled with the processor for producing an output, the method comprising acts of:
-
receiving a representation of an utterance into a shallow parser component in the form of a first word lattice including a set of potential textual word sequence matches for the utterance, where the textual word sequence matches include at least one word; receiving and analyzing the first word lattice to produce a second word lattice utilizing a filter selected from a group consisting of a sub-categorization licensing filter, a predication licensing filter, an adverbial licensing filter, and a discourse licensing filter, where the filter assigns a probability match for a potential word match in relation to another potential word match in the utterance based on a particular filter-specific criteria; and outputting the probability matches from the shallow parser component as a portion of a second word lattice for further processing to determine most likely combinations of words present in the utterance. - View Dependent Claims (16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28)
-
-
29. A computer program product for contextual filtering, comprising computer instruction means stored on a computer readable medium, for:
-
receiving a representation of an utterance into a shallow parser component in a form of a first word lattice including a set of potential textual word sequence matches for the utterance, where the textual word sequence matches include at least one word; receiving and analyzing the first word lattice to produce a second word lattice utilizing a filter selected from a group consisting of a sub-categorization licensing filter, a predication licensing filter, an adverbial licensing filter, and a discourse licensing filter, where the filter assigns a probability match for a potential word match in relation to another potential word match in the utterance based on a particular filter-specific criteria; and outputting the probability matches from the shallow parser component as a portion of a second word lattice for further processing to determine most likely combinations of words present in the utterance. - View Dependent Claims (30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42)
-
Specification