Intention acquisition method, electronic device and computer-readable storage medium

• US 10,423,725 B2
• Filed: 08/31/2017
• Issued: 09/24/2019
• Est. Priority Date: 08/04/2017
• Status: Active Grant

• Current Assignee: Ping An Technology Company Limited (Ping An Insurance (Group) Company of China, Ltd.)
• Original Assignee: Ping An Technology Company Limited (Ping An Insurance (Group) Company of China, Ltd.)
• Inventors: Zhou, Junkun, Guan, Min, Yan, Jianghao, Peng, Xiaoming
• Primary Examiner(s): Chawan, Vijay B
• Application Number:
  US16/337,929
• Publication Number:
  US 20190228069A1
• Time in Patent Office: 754 Days
• Field of Search: 704 9, 704235, 704257, 707999003, 707999004, 707999005, 707888007, 707711, 706 52, 709204, 709205, 715210, 715264
• CPC Class Codes: G06F 16/2455 Query execution
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06N 3/042 Knowledge-based neural networks; Logical representations of neural networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition

Abstract

An intention acquisition method, an electronic device and a computer-readable computer storage are provided. The method includes: segmenting a text sentence into multiple words; mapping the multiple words obtained by segmentation to preset keywords; determining an intention node most semantically approximate to the mapped words in a preset knowledge graph; performing semantic supplementing on the most approximate intention node according to a preset supplementing manner to obtain a supplemented intention node; if the supplemented intention node is an egress node, acquiring detailed intention information associated with the supplemented intention node in a specified range in a detailed intention layer of the knowledge graph according to intention information corresponding to the supplemented intention node, and determining an intention of a user according to the acquired detailed intention information; and if the supplemented intention node is a non-egress node, determining the intention of the user according to a preset query mode.

Claims

1. An electronic device, comprising a memory, a processor and an intention acquisition system stored on the memory and capable of running on the processor, the intention acquisition system being executed by the processor to implement the following steps of:

acquiring a text sentence of a present dialog of a user and segmenting the text sentence into multiple words through a preset word segmentation algorithm;

mapping the multiple words obtained by segmentation to preset keywords according to a synonym mode and parts of speech of the words to obtain mapped words of the text sentence;

determining an intention node most semantically approximate to the mapped words in a preset knowledge graph;

performing semantic supplementing on the most semantically approximate intention node according to a preset supplementing manner to obtain a supplemented intention node;

if the supplemented intention node is an egress node of an intention node layer in the knowledge graph, acquiring detailed intention information associated with the supplemented intention node in a specified range in a detailed intention layer of the knowledge graph according to intention information corresponding to the supplemented intention node, and determining an intention of the user according to the acquired detailed intention information; and

if the supplemented intention node is not the egress node of the intention node layer, determining the intention of the user according to a preset query mode.

2. The electronic device of claim 1, wherein the preset knowledge graph comprises the intention node layer, the detailed intention layer and a knowledge base layer, the intention node layer comprises nodes of multiple levels consisting of word elements of different types, the detailed intention layer stores detailed intention information associated with egress nodes of the intention node layer, and the knowledge base layer stores detailed intention information screened from the detailed intention layer and records the screened detailed intention information as specific knowledge.

3. The electronic device of claim 2, wherein the word elements of different types comprise a series element, a target element and an action element, and the intention node layer comprises nodes of three levels:

first-level nodes consisting of series elements, second-level nodes consisting of target elements and third-level nodes consisting of action elements.

4. The electronic device of claim 2, wherein the preset supplementing manner comprises a channel information supplementing manner, a user information supplementing manner and a knowledge graph information supplementing manner, channel information comprises credit card channel information, property and casualty channel information and telephone channel information, and user information comprises an account type of the user and an operation right of an account of each type.

5. The electronic device of claim 4, wherein performing semantic supplementing on the most approximate intention node according to the preset supplementing manner comprises:

if the supplementing manner is the channel information supplementing manner, supplementing missing channel information in the most approximate intention node;

if the supplementing manner is the user information supplementing manner, supplementing missing user information in the most approximate intention node; and

if the supplementing manner is the knowledge graph information supplementing manner, determining all element information corresponding to a node path where the most approximate intention node is located in the intention node layer of the knowledge graph, and supplementing missing element information in the most approximate intention node.

6. The electronic device of claim 4, wherein the preset query mode comprises an enumeration mode and an element mode, the enumeration mode refers to determining different enumeration strategies according to different channel types and outputting corresponding enumerated prompting information according to different enumeration strategies, and the element mode refers to outputting corresponding element prompting information according to missing element information in the supplemented intention node.

7. The electronic device of claim 6, wherein determining the intention of the user according to the preset query mode comprises:

if the preset query mode is the enumeration mode and the channel type is a credit card channel, outputting a first predetermined amount of enumerated prompting information;

if the preset query mode is the enumeration mode and the channel type is a property and casualty channel, outputting a second predetermined amount of enumerated prompting information; and

if the preset query mode is the enumeration mode and the channel type is a telephone channel, outputting a third predetermined amount of enumerated prompting information.

8. An intention acquisition method, applied to an electronic device and comprising:

acquiring a text sentence of a present dialog of a user and segmenting the text sentence into multiple words through a preset word segmentation algorithm;

mapping the multiple words obtained by segmentation to preset keywords according to a synonym mode and parts of speech of the words to obtain mapped words of the text sentence;

determining an intention node most semantically approximate to the mapped words in a preset knowledge graph;

performing semantic supplementing on the most semantically approximate intention node according to a preset supplementing manner to obtain a supplemented intention node;

if the supplemented intention node is an egress node of an intention node layer in the knowledge graph, acquiring detailed intention information associated with the supplemented intention node in a specified range in a detailed intention layer of the knowledge graph according to intention information corresponding to the supplemented intention node, and determining an intention of the user according to the acquired detailed intention information; and

if the supplemented intention node is not the egress node of the intention node layer, determining the intention of the user according to a preset query mode.

9. The intention acquisition method of claim 8, wherein the preset knowledge graph comprises the intention node layer, the detailed intention layer and a knowledge base layer, the intention node layer comprises nodes of multiple levels consisting of word elements of different types, the detailed intention layer stores detailed intention information associated with egress nodes of the intention node layer, and the knowledge base layer stores detailed intention information screened from the detailed intention layer and records the screened detailed intention information as specific knowledge.