Method and apparatus for processing semantic analysis result based on artificial intelligence

• US 10,191,900 B2
• Filed: 06/29/2017
• Issued: 01/29/2019
• Est. Priority Date: 11/30/2016
• Status: Active Grant

• Current Assignee: Baidu Online Network Technology (Beijing) Co., Ltd (Baidu Incorporated)
• Original Assignee: Baidu Online Network Technology (Beijing) Co., Ltd (Baidu Incorporated)
• Inventors: Hu, Zejin, Yao, Shujie
• Primary Examiner(s): McFadden, Susan I
• Application Number:
  US15/637,419
• Publication Number:
  US 20180150455A1
• Time in Patent Office: 579 Days
• Field of Search: 704 9
• CPC Class Codes: G06F 40/279 Recognition of textual entities
• 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 20/00 Machine learning
• CPC Class Codes: G06N 5/00 Computing arrangements using knowledge-based models
• CPC Class Codes: G06V 30/274 Syntactic or semantic context, e.g. balancing

Abstract

A method and an apparatus for processing a semantic analysis result based on AI are provided. With the method, the weight of the analysis text in the corresponding analysis result is determined according to the preset weight configuration information; the semantic confidence of the analysis result is detected via the pattern matching algorithm; the analysis type of the analysis texts is determined by the first classification model, and the field matching confidence of the analysis texts over the analysis result is determined by the second classification model; and then the analysis quality data of the analysis result is obtained according to the weights, the semantic confidence, the analysis type and the field matching confidence.

Claims

1. A method for processing a semantic analysis result based on artificial intelligence, comprising:

determining, by at least one computing device, a plurality of analysis results of input text received from a user;

determining, by the at least one computing device, weights of analysis texts in a corresponding analysis result of the analysis results according to preset weight configuration information;

detecting, by the at least one computing device, a semantic confidence of the analysis result via a pattern matching algorithm;

determining, by the at least one computing device, an analysis type of the analysis texts by a first classification model;

determining, by the at least one computing device, a field matching confidence of the analysis texts over the analysis result by a second classification model;

obtaining, by the at least one computing device, analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence;

comparing, by the at least one computing device, the analysis quality data of a plurality of analysis results to a preset quality threshold;

deleting, by the at least one computing device, at least one of the analysis results that does not reach to the quality threshold;

sorting, by the at least one computing device, analysis results that reach to the threshold; and

feeding back the sorted analysis results to the user.

2. The method according to claim 1, wherein, determining, by at least one computing device, weights of analysis texts in a corresponding analysis result according to preset weight configuration information comprises:

obtaining, by the at least one computing device, a field type of each of the analysis texts;

determine, by the at least one computing device, a weight of the each of the analysis texts in the corresponding analysis result according to preset weight configuration information corresponding to the field type.

3. The method according to claim 1, wherein, obtaining, by the at least one computing device, analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence comprises:

normalizing, by the at least one computing device, the weights, the semantic confidence, the analysis type, and the field matching confidence via a preset function model to obtain normalized results of the weights, the semantic confidence, the analysis type, and the field matching confidence;

summing up, by the at least one computing device, the normalized results of the weights, the semantic confidence, the analysis type, and the field matching confidence so as to obtain the analysis quality data of the analysis result.

4. The method according to claim 1, wherein, obtaining, by the at least one computing device, analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence comprises:

performing, by the at least one computing device, a weighted calculation to obtain the analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence via a preset weighted calculation model.

5. An apparatus for processing a semantic analysis result based on artificial intelligence, comprising:

a processor; and

a memory, configured to store instructions executable by the processor, wherein the processor is configured to;

determine a plurality of analysis results of input text received from a user;

determine weights of analysis texts in a corresponding analysis result according to preset weight configuration information;

detect a semantic confidence of the analysis result via a pattern matching algorithm;

determine an analysis type of the analysis texts by a first classification model;

determine a field matching confidence of the analysis texts over the analysis result by a second classification model;

obtain analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence,

compare the analysis quality data of a plurality of analysis results to a preset quality threshold;

delete at least one of the analysis results that does not reach to the quality threshold;

sort analysis results that reach to the threshold; and

,

feed back at least part of the analysis result based at least in part on the analysis quality data.

6. The apparatus according to claim 5, wherein the processor is configured to obtain analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence by acts of:

performing a weighted calculation to obtain the analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence via a preset weighted calculation model.

7. The apparatus according to claim 5, wherein the processor is configured to determine weights of analysis texts in a corresponding analysis result according to preset weight configuration information by acts of:

obtaining a field type of each of the analysis texts;

determine a weight of the each of the analysis texts in the corresponding analysis result according to preset weight configuration information corresponding to the field type.

8. The apparatus according to claim 5, wherein the processor is configured to obtain analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence by acts of:

normalizing, the weights, the semantic confidence, the analysis type, and the field matching confidence via a preset function model to obtain normalized results of the weights, the semantic confidence, the analysis type, and the field matching confidence;

summing up the normalized results of the weights, the semantic confidence, the analysis type, and the field matching confidence so as to obtain the analysis quality data of the analysis result.

9. A non-transitory computer readable storage medium, with instructions stored, wherein the instructions are executed by a processor to achieve a method for processing a semantic analysis result based on artificial intelligence, and the method comprises:

determining a plurality of analysis results of input text received from a user;

determining weights of analysis texts in a corresponding analysis result according to preset weight configuration information;

detecting a semantic confidence of the analysis result via a pattern matching algorithm;

determining an analysis type of the analysis texts by a first classification model;

determining a field matching confidence of the analysis texts over the analysis result by a second classification model;

obtaining analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence;

comparing the analysis quality data of a plurality of analysis results to a preset quality threshold;

deleting at least one of the analysis results that does not reach to the quality threshold;

sorting analysis results that reach to the threshold; and

feeding back the sorted analysis results to the user.

10. The non-transitory computer readable storage medium according to claim 9, wherein obtaining analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence comprises:

performing a weighted calculation to obtain the analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence via a preset weighted calculation model.

11. The non-transitory computer readable storage medium according to claim 9, wherein obtaining analysis quality data of the analysis result according to the weights, the semantic confidence, the analysis type, and the field matching confidence comprises:

normalizing the weights, the semantic confidence, the analysis type, and the field matching confidence via a preset function model to obtain normalized results of the weights, the semantic confidence, the analysis type, and the field matching confidence;

summing up the normalized results of the weights, the semantic confidence, the analysis type, and the field matching confidence so as to obtain the analysis quality data of the analysis result.

12. The non-transitory computer readable storage medium according to claim 9, wherein determining weights of analysis texts in a corresponding analysis result according to preset weight configuration information comprises:

obtaining a field type of each of the analysis texts;

determine a weight of the each of the analysis texts in the corresponding analysis result according to preset weight configuration information corresponding to the field type.