Using communicative discourse trees to detect a request for an explanation
AssigneeOracle International Corporation
InventorBoris Galitsky
Life of this patent
Estimating the term…
Abstract
Systems, devices, and methods of the present invention relate to detecting a request for explanation in text. In an example, a method creates a discourse tree from a subset of text. The discourse tree includes nodes, each nonterminal node representing a rhetorical relationship between two of the fragments and each terminal node of the nodes of the discourse tree is associated with one of the fragments. The method forms a communicative discourse tree from the discourse tree by matching each fragment that has a verb to a verb signature. The method further identifies that the subset of text comprises a request for an explanation by applying a classification model trained to detect a request for an explanation to the communicative discourse tree.
Claims18
A method for detecting a request for explanation in text, the method comprising:
- receiving, from a user device, text comprising fragments;
- creating a discourse tree from a subset of the text, wherein the discourse tree comprises a plurality of nodes, each nonterminal node representing a rhetorical relationship between two of the fragments and each terminal node of the nodes of the discourse tree associated with one of the fragments;
- forming a communicative discourse tree that represents the subset of the text by matching each fragment that has a verb to a verb signature;
- identifying that the subset of the text comprises a request for an explanation by applying a classification model trained to detect a request for an explanation to the communicative discourse tree;
- and
- adjusting a response based on the determined request for explanation and providing the adjusted response to the user device.
A method of building a training dataset comprising:
- accessing text comprising fragments;
- creating a discourse tree from the text, wherein the discourse tree comprises a plurality of nodes, each nonterminal node representing a rhetorical relationship between two of the fragments and each terminal node of the nodes of the discourse tree is associated with one of the fragments;
- matching each fragment that has a verb to a verb signature, thereby forming a communicative discourse tree;
- accessing a positive set of communicative discourse trees and a negative set of communicative discourse trees, wherein the positive set of communicative discourse trees comprises of communicative discourse trees representing represents text containing a request for an explanation and the negative set of communicative discourse trees comprises of communicative discourse trees representing represents text without a request for an explanation;
- identifying whether the communicative discourse tree is from the positive set of communicative discourse trees or the negative set of communicative discourse trees by applying a classification model to the formed communicative discourse tree;
- and
- adding the formed communicative discourse tree to either the positive set of communicative discourse trees or the negative set of communicative discourse trees based on the identifying.
A system comprising:
- a non-transitory computer-readable medium storing computer-executable program instructions;
- and
- a processing device communicatively coupled to the non-transitory computer-readable medium for executing the computer-executable program instructions and executing the computer-executable program instructions comprising:
- accessing text comprising fragments;
- creating a discourse tree from a subset of the text, wherein the discourse tree comprises a plurality of nodes, each nonterminal node representing a rhetorical relationship between two of the fragments and each terminal node of the nodes of the discourse tree associated with one of the fragments;
- forming a communicative discourse tree that represents the subset of the text by matching each fragment that has a verb to a verb signature;
- identifying that the subset of the text comprises a request for an explanation by applying a classification model trained to detect a request for an explanation to the communicative discourse tree;
- adjusting a response based on the determined request for explanation;
- and
- providing the adjusted response to a user device.
Dependencies are inferred from the claim text.