CN110874533A

CN110874533A - Intention identification method based on keyword matching and response path after voice identification

Info

Publication number: CN110874533A
Application number: CN201810923724.1A
Authority: CN
Inventors: 吕韶
Original assignee: Hangzhou Zhiyu Network Technology Co Ltd
Current assignee: Hangzhou Zhiyu Network Technology Co Ltd
Priority date: 2018-08-14
Filing date: 2018-08-14
Publication date: 2020-03-10
Anticipated expiration: 2038-08-14
Also published as: CN110874533B

Abstract

The invention discloses an intention identification method based on keyword matching and response path after voice identification, which uses keyword matching to promote the progress of a conversation, forms a client response path through sub-scenes of question and answer of clients in multiple rounds of conversations, and comprehensively evaluates the intention of the clients after the conversation is finished according to the intention value of each round of answer of the clients. The method of the invention realizes more accurate identification of the client intention, reduces the labor cost and the time cost of service promotion, has simple and convenient realization method and flexible means, and obviously improves the conversation quality and the evaluation capability.

Description

Intention identification method based on keyword matching and response path after voice identification

Technical Field

The invention relates to the field of computer artificial intelligent natural language processing, in particular to an intention identification method based on keyword matching and response path after voice identification.

Background

In the field of human-computer interaction, the recognition of intent for each sentence of a user has been regarded as an important consideration, while the recognition of the final intent of the user has not received much attention.

The existing methods are too complex, the accuracy and the rationality are not guaranteed too much, the pertinence to the fields and the industries is too strong, but various industries and fields are endless at present, the trend of continuous growth is presented under the drive of the rapid development of computers, the demand of intention identification is also increasing, and a method which is strong in generalization and integrates the rationality, the accuracy and the simplicity into a whole is urgently needed to be found.

Disclosure of Invention

The invention aims to provide an intention identification method based on keyword matching and a response path after voice identification aiming at the defects of the prior art. The purpose of the invention is realized by the following technical scheme:

an intention recognition method based on keyword matching and a response path after voice recognition is characterized by comprising the following steps of:

s1: segmenting the Chinese text information after the voice recognition through a Chinese word segmentation algorithm or a Chinese word segmentation tool to obtain a plurality of words;

s2: searching keywords related to the words obtained in the S1 in a database corresponding to the application scene of the voice, wherein the database comprises a plurality of sub scenes, each sub scene comprises a plurality of keywords and a plurality of responses corresponding to the sub scenes, each sub scene is provided with a preset score, matching the words obtained in the S1 with the keywords in the sub scenes, and when the matching of the words obtained in the S1 and at least one of all the keywords in the sub scenes is successful, entering S3, otherwise entering unidentified processing;

s3: randomly recording the name of a sub-scene corresponding to a successfully matched keyword and the preset score Di of the sub-scene, and randomly switching to a certain response corresponding to the sub-scene;

s4: repeating S1 to S3 until the conversation is ended, traversing the recorded response path, and calculating the total intention value D of the user according to the following formula: and D is equal to Σ Di, the response path is recorded, and the total intention value D of the user and the corresponding response path are used as the intention identification result of the user together.

Further, the Chinese word segmentation algorithm is preferably a word segmentation algorithm based on character string matching.

Furthermore, the Chinese word segmentation tool is an NLPIR Chinese word segmentation system.

Further, in S4, after traversing the recorded response path, if the ith sub-scene appears N times in the response path, where N is greater than or equal to 2, when the intention value D of the user is calculated, the score of the sub-scene is k times of the original score, and k is greater than 1.

The invention has the beneficial effects that: the method simplifies the intention identification of the customers, screens out various types of customers, provides the required customer intention values for the users and enterprises, and adapts the customer labels set by the users or the enterprises according to the intention values so as to be convenient for subsequent decision making. The method has high generalization, can be suitable for various industries, does not have too high technical barrier, has strong applicability and can be used quickly.

Drawings

FIG. 1 is a flow chart of an intent recognition method based on keyword matching and response paths after speech recognition.

Detailed Description

The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments, and the objects and effects of the present invention will become more apparent, and the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

As shown in fig. 1, an intention recognition method based on keyword matching and a response path after speech recognition is characterized by comprising the following steps:

Preferably, the chinese word segmentation algorithm is a word segmentation algorithm based on character string matching.

Preferably, the Chinese word segmentation tool is an NLPIR Chinese word segmentation system.

Preferably, in S4, after traversing the recorded response path, if the ith sub-scene appears N times in the response path, where N is greater than or equal to 2, when the intention value D of the user is calculated, the score of the sub-scene is k times of the original score, and k is greater than 1.

The present invention will be described in detail below with reference to a representative embodiment selected in the context of real estate marketing, where the database includes several sub-scenarios "initial scenario", "price", "location", "rejection", where,

the 'price' sub-scene corresponds to keywords of 'average price', 'price' and 'price', and a preset scene score d is 20;

the sub-scene of the position corresponds to the keywords of the position, the position and the section, and the preset score d of the scene is 20;

the 'refusal' sub-scene corresponds to keywords of 'don't use 'and' don't use', and the preset score d of the scene is-100;

for simplifying the description, the scenes are all set to be matched successfully as long as one keyword can be matched successfully.

Example (b):

the robot comprises: "hello, your home in Shanghai's essence, will be interested in knowing about? "

The user: "is the location? "

The robot comprises: "in the west subway entrance, locate at XX way and XX way intersection of XX district, only the distance of about 500 meters from Wanda square, just in the north of XX overhead, the position is very good. "

The user: "how much is the price? "

The robot comprises: "Thanks, our house is one side with one price, all around thirty thousand, with elaborate decorations. The parking space is about one in twenty thousand. "

The user: "too expensive, not! "

The session is ended.

S1: segmenting the Chinese text information after the voice recognition through a Chinese word segmentation algorithm or a Chinese word segmentation tool to obtain the segmentation of the position, the position and the where;

s2: searching keywords related to the words obtained in the S1 in a database corresponding to the application scene of the voice, wherein the database comprises a plurality of sub scenes, each sub scene comprises a plurality of keywords and a plurality of responses corresponding to the sub scenes, each sub scene is provided with a preset score, matching the words obtained in the S1 with the keywords in the sub scenes, and when the matching of the words obtained in the S1 and at least one of all the keywords in the sub scenes is successful, entering S3, otherwise entering unidentified processing; here the keyword "location" is successfully matched;

s3: randomly recording the name of a sub-scene corresponding to a successfully matched keyword and the preset score Di of the sub-scene, and randomly switching to a certain response corresponding to the sub-scene; here, "location" and the round of scene scores D1 are recorded as 20;

s4: repeating S1 to S3 until the conversation is ended, traversing the recorded response path, and calculating the total intention value D of the user according to the following formula: and D is equal to Σ Di, the response path is recorded, and the total intention value D of the user and the corresponding response path are used as the intention identification result of the user together. The dialog of this embodiment forms a response path consisting of sub-scenarios "location", "price", "reject". Traversing the recorded response path, and calculating the total intention value D of the user, wherein the calculation formula is as follows: d ═ Di ═ D1+ D2+ D3 ═ 20+20+ (-100) — -60, and the intention value and the response path are used as the intention recognition result of the user.

It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and although the invention has been described in detail with reference to the foregoing examples, it will be apparent to those skilled in the art that various changes in the form and details of the embodiments may be made and equivalents may be substituted for elements thereof. All modifications, equivalents and the like which come within the spirit and principle of the invention are intended to be included within the scope of the invention.

Claims

1. An intention recognition method based on keyword matching and a response path after voice recognition is characterized by comprising the following steps of:

s1: and segmenting the Chinese text information after the voice recognition through a Chinese word segmentation algorithm or a Chinese word segmentation tool to obtain a plurality of words.

S2: searching keywords related to the words obtained in the S1 in a database corresponding to the application scene of the voice, wherein the database comprises a plurality of sub scenes, each sub scene comprises a plurality of keywords and a plurality of responses corresponding to the sub scenes, each sub scene is provided with a preset score, matching the words obtained in the S1 with the keywords in the sub scenes, and when the matching of the words obtained in the S1 and at least one of all the keywords in the sub scenes is successful, entering S3, otherwise entering unidentified processing.

2. The method for identifying an intention based on keyword matching and response path according to claim 1, wherein the Chinese word segmentation algorithm is preferably a word segmentation algorithm based on character string matching.

3. The intent recognition method based on keyword matching and response path according to claim 1, wherein the chinese word segmentation tool is NLPIR chinese word segmentation system.

4. The method for identifying an intention based on keyword matching and a response path according to claim 1, wherein in S4, after traversing the recorded response path, if the ith sub-scene appears N times in the response path, where N is greater than or equal to 2, then when calculating the intention value D of the user, the score of the sub-scene becomes k times of the original, where k is greater than 1.