- Number
- 11164087
- Published
- 2021-11-02
- Filed
- 2016-05-20
- Assignee
- Disney Enterprises, Inc.
- Inventors
- Li; Boyang, Luan; Yi
- CPC
- G06F16/355; G06F40/30; G06F40/35; G06N20/00; G06N5/02
- Verdict
- Set aside NLP semantic-role research, generic
- Source
- Google Patents · FreePatentsOnline
Abstract
There is provided a system including a non-transitory memory storing an executable code and a hardware processor executing the executable code to receive an input sentence including a first predicate and at least a first argument depending from the first predicate, identify the first predicate, identify the first argument based on the first predicate, apply a dependency multiplication to determine a semantic role of the first argument based on the first predicate, and assign the first argument to an argument cluster including one or more similar arguments based on the semantic role of the first argument.
Background
BACKGROUND (1) On the quest to create artificial intelligence, understanding natural language is a challenge. Natural language is complex, and statements describing the same situation can be formulated in more than one way. For example, a sentence may be stated in a passive form or may be stated in an active form, but still communicate the same information. Conventional systems for understanding natural language rely on training data and manual annotation of the training data to interpret natural language. SUMMARY (2) The present disclosure is directed to systems and methods for determining semantic roles of arguments in sentences, substantially as shown in and/or described in connection with at least one of the figures, as set forth more completely in the claims.
Claims
1. A system comprising: a non-transitory memory storing an executable code; a hardware processor configured to execute the executable code to: receive an input sentence including a first predicate and a first argument depending from the first predicate, the input sentence further including a second argument depending from the first predicate; identify the first predicate; identify the first argument based on the first predicate; represent the first argument as a first multi-dimensional vector having a plurality of components each corresponding to a respective one of a plurality of letters included in the first argument; apply a dependency multiplication by multiplying the first multi-dimensional vector representing the first argument with a representation of a syntactic relation between the first argument and the first predicate to determine a semantic role of the first argument in the input sentence; assign the first argument to a first argument cluster based on the semantic role of the first argument; assign the second argument to the first argument cluster; and in response to assigning both the first argument and the second argument to the first argument cluster, determine a similarity between the first argument cluster and a second argument cluster by deducting a value as a penalty from a cosine similarity between a first centroid of the first argument cluster and a second centroid of the second argument cluster. ||
7. A method for use with a system including a non-transitory memory and a hardware processor, the method comprising: receiving, using the hardware processor, an input sentence including a first predicate and a first argument depending from the first predicate, the input sentence further including a second argument depending from the first predicate; identifying, using the hardware processor, the first predicate; identifying, using the hardware processor, the first argument based on the first predicate; representing, using the hardware processor, the first argument as a first multi-dimensional vector having a plurality of components each corresponding to a respective one of a plurality of letters included in the first argument; applying, using the hardware processor, a dependency multiplication by multiplying the first multi-dimensional vector representing the first argument with a representation of a syntactic relation between the first argument and the first predicate to determine a semantic role of the first argument in the input sentence; assigning, using the hardware processor, the first argument to a first argument cluster based on the semantic role of the first argument; assigning, using the hardware processor, the second argument to the first argument cluster; and in response to assigning both the first argument and the second argument to the first argument cluster, determining, using the hardware processor, a similarity between the first argument cluster and a second argument cluster by deducting a value as a penalty from a cosine similarity between a first centroid of the first argument cluster and a second centroid of the second argument cluster. ||
15. A method for use with a system including a non-transitory memory and a hardware processor, the method comprising: receiving, using the hardware processor, an input sentence including a first predicate, a first argument depending from the first predicate, and a second argument depending from the first predicate; identifying, using the hardware processor, the first predicate; identifying, using the hardware processor, the first argument based on the first predicate; representing, using the hardware processor, the first argument as a first multi-dimensional vector having a plurality of components each corresponding to a respective one of a plurality of letters included in the first argument; applying, using the hardware processor, a dependency multiplication to the first multi-dimensional vector representing the first argument to determine a semantic role of the first argument based on the first predicate; assigning, using the hardware processor, the first argument to a first argument cluster based on the semantic role of the first argument; assigning, using the hardware processor, the second argument to the first argument cluster; and in response to assigning both the first argument and the second argument to the first argument cluster, determining, using the hardware processor, a similarity between the first argument cluster and a second argument cluster by deducting a value as a penalty from a cosine similarity between a first centroid of the first argument cluster and a second centroid of the second argument cluster.