CN116541487A

CN116541487A - Cloud service searching method and system based on voice

Info

Publication number: CN116541487A
Application number: CN202310539958.7A
Authority: CN
Inventors: 叶朋
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2023-05-12
Filing date: 2023-05-12
Publication date: 2023-08-04

Abstract

The invention relates to the technical field of voice processing and the digital medical field, and discloses a cloud service searching method and a system based on voice, wherein the target application terminal is based on a preset voice receiving unit to acquire voice information of a user; the target application end extracts search word information in the voice information and transmits the search word information to the cloud service end; the cloud server judges whether keyword information related to the search word information exists in a historical database; if the keyword information related to the search word information exists, the cloud service end acquires service type information corresponding to the related keyword information, generates a related service display table and transmits the related service display table to the target application end so that a user can acquire required services, and the target application end can be an intelligent diagnosis and treatment application, an intelligent home application, an intelligent life service application and the like. The invention improves the speed of providing the demand service for the user by the application.

Description

Cloud service searching method and system based on voice

Technical Field

The invention relates to the technical field of voice processing and the digital medical field, in particular to a cloud service searching method and system based on voice.

Background

With the development of intelligent applications such as intelligent diagnosis and treatment applications, intelligent home applications and intelligent life service applications, application programs are gradually integrated, and various service micro-applications and service applets are integrated in the application programs, so that a user can acquire various services in one application program. The user can acquire the corresponding service by clicking the service icon in the application program, so that better use experience is brought to the user.

However, as applications are updated, new service icons often appear in the applications, and the layout of the service icons is continuously changed, so that many users have difficulty in quickly finding a required service.

Disclosure of Invention

The invention provides a cloud service searching method and system based on voice, which solve the technical problem that a user is difficult to quickly find a required service.

In a first aspect, a cloud service searching method based on voice is provided, and the method is applied to a target application end and a cloud service end, wherein the target application end and the cloud service end are connected through a network, and the method comprises the following steps:

the target application terminal obtains voice information of a user based on a preset voice receiving unit;

the target application end extracts search word information in the voice information and transmits the search word information to the cloud service end;

The cloud service end judges whether keyword information related to the search word information exists in a historical database, and the historical database comprises a plurality of keyword information and corresponding service type information;

if the keyword information related to the search word information exists, the cloud service end acquires service type information corresponding to the related keyword information, generates a related service display table and transmits the related service display table to the target application end so that a user can acquire required service;

if the keyword information related to the search word information does not exist, the cloud service end performs word splitting processing on the search word information and generates a plurality of sub-search word information;

the cloud server acquires keyword information related to the sub-search word information in the history database;

and the cloud service end acquires corresponding service type information based on the keyword information related to the sub-search word information, generates an associated service display table and transmits the associated service display table to the target application end so as to enable a user to acquire required service.

In a second aspect, a cloud service search system based on voice is provided, including a target application end and a cloud service end, where the target application end is connected with the cloud service end through a network, and the target application end includes:

The receiving module is used for acquiring voice information of a user based on a preset voice receiving unit;

the extraction module is used for extracting search word information in the voice information and transmitting the search word information to the cloud server;

the cloud service end comprises:

the judging module is used for judging whether keyword information related to the search word information exists in a historical database or not, and the historical database comprises a plurality of keyword information and corresponding service type information;

the processing module is used for acquiring service type information corresponding to the related keyword information if the keyword information related to the search word information exists, generating a related service display table and transmitting the related service display table to the target application end so as to enable a user to acquire required service;

the word splitting module is used for splitting the search word information and generating a plurality of sub-search word information if the keyword information related to the search word information does not exist;

the acquisition module acquires keyword information related to the sub-search word information in the history database;

and the generation module is used for acquiring corresponding service type information based on the keyword information related to the sub-search word information, generating an associated service display table and transmitting the associated service display table to the target application end so as to enable a user to acquire required service.

In the scheme implemented by the cloud service searching method and the cloud service searching system based on the voice, the voice information of the user can be obtained based on the voice receiving unit preset in the target application end, and the search word information in the voice information can be extracted. The user's desired service can be queried using the search term information. When the keyword information related to the search word information exists in the history database of the cloud service end, the related keyword information and the corresponding service type information can be acquired, and the service type information is the service required to be searched by the user. The associated service display list is generated by summarizing the service type information, and can be displayed on an application end so that a user can acquire required services. When the keyword information related to the search word information does not exist in the history database of the cloud service end, the search word information can be split into a plurality of sub-search word information. The range of the search word information is enlarged, and the related service type information is more easily queried. According to the method and the device for generating the service display list, the target application can quickly acquire the user demand information based on voice recognition and search related demand services of the user at the cloud service end, so that the service display list is generated for the user to select. The speed of providing the demand service for the user by the application is improved, so that the use feeling of the user is improved, and the application utilization rate of the user is improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments of the present invention will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic view of an application environment of a voice-based cloud service searching method according to an embodiment of the present invention;

FIG. 2 is a flow chart of a method for searching for voice-based cloud services according to an embodiment of the present invention;

FIG. 3 is a flowchart illustrating the step S220 in FIG. 2;

FIG. 4 is a flowchart illustrating the step S230 in FIG. 2;

FIG. 5 is a flowchart illustrating step S240 in FIG. 2;

FIG. 6 is a flowchart illustrating the step S242 of FIG. 5;

FIG. 7 is a flowchart illustrating the step S260 in FIG. 2;

FIG. 8 is a flowchart illustrating the step S270 of FIG. 2;

fig. 9 is a schematic structural diagram of a voice-based cloud service search system according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The cloud service searching method based on the voice provided by the embodiment of the invention can be applied to the application environment as shown in fig. 1. The target application end can be an intelligent diagnosis and treatment application, an intelligent home application, an intelligent life service application and the like. It should be noted that, the target application 110 may communicate with the cloud service 120 through a network. The target application 110 may obtain the voice information of the user based on a preset voice receiving unit, and extract the search term information in the voice information. The search term information may include features of a service desired by the user, and the service desired by the user may be queried using the search term information. The target application terminal 110 may transmit the search term information to the cloud service terminal 120, and when the keyword information related to the search term information exists in the history database of the cloud service terminal 120, the cloud service terminal 120 may obtain the related keyword information and the corresponding service type information, where the service type information is the service that the user needs to search. By aggregating such service type information to generate an associated service display list, it can be displayed on the application 110 for the user to obtain the desired service. It should be further noted that, when no keyword information related to the search word information exists in the history database of the cloud server 120, the search word information may be split into a plurality of sub-search word information. At this time, the information amount of the sub-search word information is larger, and the related service type information is easier to inquire. Therefore, in the present invention, a voice receiving unit is introduced into the target application 110 to obtain the voice information of the user, and the target application 110 extracts the search word information, so that the cloud service 120 queries the relevant service types in the history database based on the search word information. And the search term information is subjected to word splitting processing, so that the search range of the search term information is improved, and related service type information is more easily queried. The generated service presentation list is then available for selection by the user. Therefore, the speed of providing the required service for the user by the application is improved, the use feeling of the user is improved, and the application utilization rate of the user is improved.

It should be mentioned that the target application end may be, but not limited to, applications in various personal computers, notebook computers, smart phones, tablet computers and portable wearable devices. The cloud service end can be implemented by an independent server or a server cluster formed by a plurality of servers. The present invention will be described in detail with reference to specific examples.

Referring to fig. 2, fig. 2 is a schematic flow chart of a voice-based cloud service searching method according to an embodiment of the present invention, and is described in detail below.

Step S210, the target application terminal obtains voice information of a user based on a preset voice receiving unit.

Step S220, the target application end extracts the search word information in the voice information and transmits the search word information to the cloud service end.

Step S230, the cloud service end judges whether keyword information related to the search word information exists in a history database, wherein the history database comprises a plurality of keyword information and corresponding service type information.

Step 240, if there is keyword information related to the search word information, the cloud server acquires service type information corresponding to the related keyword information, generates an associated service display table, and transmits the associated service display table to the target application end, so that the user can acquire the required service.

Step S250, if no keyword information related to the search word information exists, the cloud service end performs word splitting processing on the search word information and generates a plurality of sub-search word information.

Step S260, the cloud server acquires keyword information related to the sub-search word information in the history database.

Step S270, the cloud service end obtains corresponding service type information based on the keyword information related to the sub-search word information, generates an associated service display table and transmits the associated service display table to the target application end so that a user can obtain required service.

When step S210 is performed, the user may open the target application terminal 110, and the target application terminal 110 may be a health service application, an entertainment application, a home application, or other application programs. The user speaks a wake-up instruction to wake up the preset voice receiving unit in the target application 110. The wake-up instruction can be set according to actual requirements, for example, "love", "cloud", "fly", etc. The voice receiving unit in the target application terminal 110 can be evoked by the wake-up instruction, so that the voice receiving unit receives voice information of the user. The voice information may serve information for the needs of the user. Specifically, the user may speak the voice information, and the voice information may be received by the voice receiving unit of the target application end 110, so that the cloud service end 120 queries the corresponding required service according to the voice information.

Specifically, as an example, when the present invention is used in a smart home application scenario, a user may open a smart home application in a mobile phone or other devices and speak a wake-up instruction to wake up a voice receiving unit of the application, and the voice information of the user may be "open a sound box", "close a television", etc. The target application end 110 is preset with a voice receiving unit and a voice recognition unit, the voice receiving unit can receive voice information, and the voice recognition unit can further recognize the voice information to obtain service information required by the user so as to query services required by the user.

It should be noted that, the cloud service searching method based on voice provided by the invention is not limited to the intelligent home application, but also can be applied to intelligent applications in other application scenes, such as applications on an intelligent customer service robot, intelligent health service applications, entertainment applications and the like, and can be specifically set according to actual requirements.

When step S220 is performed, the search term information refers to a search term regarding a desired service in the user' S voice information. For example, the user's voice message may be "how open the box can be," which is a search term for the desired service. For another example, the user's voice message may be "i am watching a movie," where "watching a movie" is a search term for a desired service.

By extracting the search word related to the required service in the voice information, the cloud service end 120 can better utilize the search word to query the related service type information in the historical database, so as to avoid the influence of irrelevant vocabulary in the voice information on the query effect.

It is noted that when the user is an elderly crowd, dialects are often used to influence the extraction of the search term information when the elderly crowd speaks the voice information. When the user is the low-age group, the problem of unclear voice and voice error is easy to occur when the low-age group speaks the voice message. When the user is middle-aged people, popular words are often mixed when the middle-aged people speak the voice message, and extraction of search word information is affected. Therefore, different voice recognition models can be set in the voice recognition unit of the application terminal 110, so as to extract the voice messages of the crowd at different ages in a targeted manner, and improve the extraction speed of the search word information. The age range of the above-mentioned age group may be set based on actual conditions, for example, the elderly population may be set to 50 to 100 years old, the middle-aged population may be set to 11 to 50 years old, and the low-aged population may be set to 1 to 10 years old.

When step S230 is performed, the history database may refer to a database in the cloud server 120, where the history database may store common keywords and service types corresponding to each keyword.

As an example, when the present invention is used in a smart diagnosis application, a user may speak his own health problems in the form of a voice message. The cloud service 120 may query the historical database for related keywords according to the search terms in the voice message, and obtain a related service department list for the user to use. For the historical database, for example, keywords "dyspnea", "shortness of breath", "skin rash", "fracture", etc. may be stored in the historical database. The service type corresponding to the keyword 'dyspnea' and 'shortness of breath' can be respiratory medicine, the service type corresponding to the keyword 'skin erythema' can be dermatological, and the service type corresponding to the keyword 'fracture' can be orthopedics.

It should be further explained that the history database may store all available service type information and corresponding keywords. The cloud service end 120 may query the history database based on the search word information of the user, and when the keyword information related to the search word information exists in the history database, the service type information corresponding to the keyword information is the service type required by the user.

In the above-described determination of the correlation, it is to be noted that, when the same or semantically similar vocabulary exists in the search word information and the keyword information, the search word information is correlated with the keyword information. As an example, a statistical method may be employed to calculate the relevance between the search term and the keyword. The search word information a of the user may be set to "dyspnea", and the certain keyword information B of the history database may be set to "dyspnea". For search term information a, the frequency of all the terms may correspond to dyspnea (1) difficulty (1) not (0) smooth (0). For search term information B, the frequency of all terms may correspond to dyspnea (1) difficulty (0) not (1) smoothness (1). From the above, the feature word vector W1 of the search word information a is (1, 0), and the feature word vector W2 of the search word information B is (1, 0, 1). Next to this, the process is carried out, calculating cosine of the feature word vector W1 and the feature word vector W2 similarity cos θ= (w1.w2)/(|w1||w1||) and a method for determining the similarity, wherein w1×w1 is the modulo-long product of two feature word vectors. Cosine similarity can be expressed as correlation. For example, when the correlation threshold is set to 0.1, it may be determined that keyword information related to the search term information exists in the history database when the cosine similarity, that is, the correlation is 0.1 or more.

As another example, the user's search term information may be "dyspnea," the keywords in the history database may be "dyspnea," "dyspnea," and "skin erythema," and the cloud server 120 queries the keywords in the history database for "dyspnea," "dyspnea," and "skin erythema" as to whether "dyspnea" is relevant. If the relevance of the keywords "dyspnea", "dyspnea" and the search word "dyspnea" in the historical database is larger than the relevance threshold, the keywords "dyspnea", "dyspnea" and the search word "dyspnea" are related, the keywords "skin erythema" and "dyspnea" are not related, and the service types corresponding to the keywords "dyspnea" and "dyspnea" can be related service types.

It should be noted that, the number of service types related to a certain keyword may be plural, and the number of keywords corresponding to a certain service type may be plural.

When step S240 is performed, if keywords related to the search word of the user exist in the history database, such related keyword information and corresponding service types may be obtained. By summarizing the service types corresponding to such related keyword information, an associated service display table may be formed. The associated service display list can be transmitted to the target application terminal 110 and displayed on the page of the target application terminal 110, so that the user can select the required service.

As an example, the present invention is described as applied to the intelligent diagnosis and treatment application, when the search word information of the user is "dyspnea", and related keywords such as "dyspnea", "shortness of breath" exist in the history database, keywords related to the search word of the user exist in the history database. At this time, the service type corresponding to the related keyword may be respiratory department, cardiology department, gastroenterology department, etc. The service types are summarized and sequentially arranged to form an associated service display list, and the arrangement order can be set according to actual requirements.

As another example, the associated service exposure table may be,

1. the respiratory department of the respiratory medicine,

2. the utility model relates to a heart internal medicine,

3. digestive system department.

The user may click or voice select the presentation sequence number of the service type to select the desired service. Specifically, the user clicks "1" or the user speaks "1", i.e. the user can jump to the details service page of the respiratory department. Furthermore, the corresponding voice service function can be accessed into the service page of the respiratory department, so that the user can continuously utilize voice to realize man-machine interaction, and the speed of acquiring the required service by the user is improved.

It should be noted that, during the process of using the voice query service, the user may speak the voice information including the exit instruction to exit the voice service. The exit instruction can be set according to actual requirements, for example, the exit instruction can be a "love, exit", "cloud, please exit", "fly, bar exit" or other exit instructions. Further, when the voice service needs to be awakened, the voice receiving unit can be awakened through an awakening instruction, for example, "love", "cloud", "fly", and the like, so as to reenter the voice service.

When step S250 is performed, if no keyword information related to the search word information exists in the history database, the search word information may be split into a plurality of sub-search words, so as to expand the search range.

As an example, when the present invention is used in the intelligent diagnosis and treatment application, the search word information of the user may be "chest distress short", the keyword information in the history database may be "dyspnea", "shortness of breath", etc., and if the keyword information related to "chest distress short" does not exist in the history database after the correlation calculation, the cloud service end 120 may split the search word "chest distress short", so as to expand the search range. Specifically, the keyword "chest distress and shortness of breath" may be split into a plurality of sub-search words, "chest distress", "shortness of breath" and the like. At this time, if the cloud service end 120 determines that the sub-search word "shortness of breath" is related to the keyword information "dyspnea" and "unsmooth breathing" in the history database after the correlation calculation, the service department corresponding to "dyspnea" and "unsmooth breathing" may be the service type required by the user. Therefore, by splitting the search term information into a plurality of sub-search terms, the required service types can be more effectively inquired after the search range is enlarged.

When step S260 is executed, it should be noted that, after the search word information is split into a plurality of sub-search words, the scope of the search words can be enlarged, so that related keywords can be more easily queried in the history database. When keywords related to sub-search words are queried in the history database, such keyword information may be obtained in order to query the corresponding service type.

When step S270 is performed, specifically, if keywords related to the sub-search terms exist in the history database, the service type corresponding to the related keywords may be obtained. By summarizing such related service types, an associated service presentation table may be formed. The associated service display list may be transmitted to the target application end and displayed on the page of the application end 110 for the user to select the desired service.

It should be noted that, when the keyword related to the sub-search term still cannot be queried after the search term information is split, other popular service types may be summarized and displayed to the application end 110, so as to provide other alternative service types for the user.

It should be noted that, the cloud server 120 may also store the age group ratio of the user of each service type in advance, so as to recommend the service type preferred by the age group based on the age group of the user. As an example, if a certain service type a is pre-stored in the cloud service end 120, and the history users of the service type a have a senior group population ratio of 10%, a senior group population ratio of 70% and a senior group population ratio of 20%. Because the history users of the service type A have higher occupation of middle-aged people, the service type A can be summarized into the popular service types preferred by the middle-aged people. When the user searching based on voice is the middle-aged layer and the related service type is not found, the cloud service end 120 may transmit the service type a as a popular service type to the target application end 110 for the user to select.

In an exemplary embodiment, as shown in fig. 3, the process of extracting the search term information in the voice information by the target application side may include,

step S221, the target application end identifies the voice message and acquires age layer information corresponding to the voice message, wherein the age layer information comprises an advanced age layer, a middle age layer and a low age layer.

Step S222, based on the age group information, the target application end carries out correction processing on the voice message to generate correction information.

Step S223, the target application side extracts the search word information in the correction information.

Because the characteristics of the voice information of the crowd at different ages are different, the extraction of search words in the voice information can be influenced. Therefore, the age group information corresponding to the voice message can be recognized first, and voice correction processing is carried out on crowds with different age groups to obtain correct voice information, so that search words are extracted.

Specifically, an age recognition module may be first set in the voice recognition unit of the application terminal 110, and the age layer corresponding to the voice information may be recognized by extracting the feature information in the voice information through the age recognition module. The range of age groups may be preset, for example, the elderly population may be set to 50 to 100 years old, the middle-aged population may be set to 11 to 50 years old, and the low-aged population may be set to 1 to 10 years old. When the age identification module identifies that the age corresponding to the voice information is 13 to 15 years old, the age layer of the voice information can be judged to be the middle age layer.

As an example, when the present invention is used in an intelligent health service application, if the user is an advanced crowd, when the advanced crowd speaks the voice information of "i feel dyspnea", sometimes, the voice information of "i catch up with the foot and inhale difficult" is formed due to dialect, which affects the extraction of the real search word. Therefore, a speech recognition model for the elderly population may be set in the speech recognition unit of the application terminal 110.

Specifically, a large amount of voice information of the senior citizen population can be collected first, each voice information of the senior citizen population is used as an input layer of a voice recognition model, a voice text actually corresponding to each voice information of the senior citizen population is used as an output layer of the voice recognition model, and a large amount of sample data is utilized to train the voice recognition model of the senior citizen population so as to establish the voice recognition model of the senior citizen population. When the user is an advanced crowd, the voice information which is uttered by the user, such as 'i catch up with the foot and inhale difficultly', can be input into a voice recognition model of the advanced crowd so as to correct the voice information and output a voice text 'i feel dyspnea' which corresponds to the voice information. Similarly, when the user is a low-age group or a middle-age group, the problem of unclear voice and voice error is easy to occur when the low-age group speaks the voice message. When the user is middle-aged people, popular words are often mixed when the middle-aged people speak the voice message, and extraction of search word information is affected. At this time, a large amount of sample data can be used to train the voice recognition model of the corresponding age group so as to perform targeted correction on the voice information. The specific training manner of the speech recognition model in this embodiment is not described herein, and may be obtained from the existing literature.

After the voice information is corrected, the search term information therein may be extracted. For example, the voice text corresponding to the voice information is "i feel dyspnea", and in order to extract the search word in the voice text, a text extraction model may be set in the voice recognition unit of the target application terminal 110. Specifically, the text extraction model may recognize unnecessary words in the speech text, delete unnecessary words, and retain necessary search words. For example, "i feel" in "i feel dyspnea" is a non-essential word, and "dyspnea" is a necessary search word that needs to be kept. For another example, "i want to go" in "i want to sing" is a non-essential word, and "singing" is an essential search word that needs to be kept.

In an exemplary embodiment, as shown in fig. 4, the process of determining whether keyword information related to the search term information exists in the history database by the cloud service end may include,

step S231, the cloud server calculates the relativity of each keyword information and the search word information in the history database.

Step 232, the cloud server judges whether the correlation is greater than a correlation threshold.

Step S233, if the correlation threshold is greater than or equal to the correlation threshold, determining that keyword information related to the search word information exists in the history database, and if the correlation threshold is less than the correlation threshold, determining that keyword information related to the search word information does not exist in the history database.

It should be noted that, the relevance between each keyword information and the search term information in the history database may be calculated through a relevance model. The correlation model may be a pre-trained learning model, and the correlation of a plurality of samples may be calculated by a statistical calculation method when training the learning model. Specifically, one search word information a may be set, and one keyword information B may be set. The search term information a may correspond to one feature term vector W1, and the search term information B may correspond to one feature term vector W2. Then, cosine similarity between the feature word vector W1 and the feature word vector W2 may be calculated. Cosine similarity can be expressed as correlation.

As an example, when the present invention is used in a scenario of a life service application, the search word information a of the user may be set to "go to western restaurant to drink", and the certain keyword information B of the history database may be set to "go to chinese dining hall to eat". For search word information a, the frequency of all words corresponds to going (1) to dining (0) in dining room (0) in western dining room (1) drinking (1). For search word information B, the frequency of all words corresponds to going (1) to western restaurant (0) drinking (0) for dining (1) in chinese dining hall (1). From the above, the feature word vector W1 of the search word information a is (1, 0), and the feature word vector W2 of the search word information B is (1, 0, 1). Next to this, the process is carried out, calculating cosine of the feature word vector W1 and the feature word vector W2 similarity cos θ= (w1.w2)/(|w1||w1||) and a method for determining the similarity, wherein w1×w1 is the modulo-long product of two feature word vectors. If the correlation threshold is set to 0.1, when the cosine similarity is greater than or equal to 0.1, it can be determined that keyword information related to the search term information exists in the history database.

In an exemplary embodiment, as shown in fig. 5, if there is keyword information related to the search term information, the cloud service side obtains service type information corresponding to the related keyword information, and the process of generating the associated service display table may include,

step S241, if there is keyword information related to the search word information, the cloud server acquires related relevance and service type information corresponding to the related keyword information.

Step S242, based on the correlation degree, the cloud service end sorts the service type information to generate an associated service display table.

It should be noted that, the relevant service display list may be formed by summarizing the service type information corresponding to the related keyword information. Specifically, the plurality of service type information may be ranked based on the relevance of the corresponding keyword information, so that the user may obtain the service type closest to the requirement according to the ranked display table. For example, the service type information a corresponds to a degree of correlation of the keyword information with the user search term of 80%. And the service type information B, wherein the correlation degree between the corresponding keyword information and the user search word is 90%. And the service type information C, wherein the correlation degree between the corresponding keyword information and the user search word is 70%. And the service type information D, wherein the correlation degree between the corresponding keyword information and the user search word is 85%. The ranking basis of the plurality of service type information may be ranking based on relevance on the basis of the popularity value. When the heat values of the plurality of service type information are the same, the higher the correlation corresponding to a certain service type information, the closer the service type information is to the required service of the user. Wherein, the heat value is the historical click times of a certain service type information, and the more the historical click times, the higher the heat value. The associated service presentation table arranged in ascending order based on the correlation may be,

The service type information C is used to determine,

the service type information a is used to determine,

the service type information D is used to determine,

service type information B.

The associated service presentation table arranged in descending order based on the correlation degree may be,

the service type information B is used to determine,

the service type information D is used to determine,

the service type information a is used to determine,

service type information C.

In an exemplary embodiment, as shown in fig. 6, based on the correlation, the cloud service side orders the service type information, and the process of generating the associated service presentation table includes,

step S2421, the cloud service side obtains the historical click times of the service type information;

step S2422, based on the historical click times, the cloud service end calculates a heat value of the service type information;

step S2423, based on the heat value and the correlation, the cloud service side calculates a ranking index of the service type information;

step S2424, based on the ranking index, the cloud service end ranks the service type information to generate a related service display table.

It should be noted that, the ranking basis of the service type information may be ranking based on ranking indexes, where the higher the ranking index of a certain service type information is, the earlier the ranking position is. The ranking index may be calculated based on the popularity value of the service type information and the relevance of the corresponding keywords.

Specifically, the ranking index of a service type information may be expressed as p=s×q1+c×q2, and q1+q2=100%. Wherein, S may represent a heat value of certain service type information, C may represent a correlation degree of a keyword corresponding to certain service type information, Q1 may represent a weight given to S, and a value of Q1 may be set based on the ranking requirement. Q2 may represent a weight given to C, and the value of Q2 may be set based on ordering requirements. For example, the value of Q1 is set to 30%, the value of Q2 is set to 70%, the value of S is set to 60%, and the value of C is set to 90%. Its ranking index may be expressed as p=60% +30% +90% +70% =81%.

In an exemplary embodiment, as shown in fig. 7, the process of obtaining keyword information related to sub-search word information in the history database by the cloud service side includes,

in step S261, the cloud service end determines whether keyword information related to the sub-search word information exists in the history database.

In step S262, if there is keyword information related to the sub-search term information, the cloud service end obtains the keyword information related to the sub-search term information.

Step S263, if no keyword information related to the sub-search term information exists, the cloud service end generates a hot service display table based on the hot value of the service type information in the history database.

When the keyword information related to the search word information does not exist in the history database, the search word information can be subjected to word splitting processing to form a plurality of sub-search words, so that the range of the search word information is enlarged, and related keywords are more easily queried in the history database.

As an example, when the present invention is used in a scenario of a life service application, the search word information a of the user may be set to "go to western restaurant to drink", and the certain keyword information B of the history database may be set to "go to chinese restaurant to eat", "go to western dining", "go to chinese dining". When the correlation degree calculation is carried out by utilizing the whole search word information A and the related keyword information B cannot be queried in the history database, the search word information A can be split into sub-search words of 'go', 'western style', 'hall', 'drink', and 'wine'. At this time, the range of the search word information is expanded, and the relevance between each sub-search word and the keywords in the history database can be calculated again to inquire the related keywords and the corresponding service types.

It should be noted that, if it is queried that keywords related to sub-search words exist in the history database, service type information corresponding to the related keywords can be summarized. Based on the correlation degree of the related keywords and the sub-search word information and the heat degree of the service type information, the service type information can be ranked, so that a user can acquire the service type closest to the requirement of the user according to the ranked display table.

Specifically, when ordering the plurality of service type information, the ordering may be based on an ordering index. The higher the ranking index of a certain service type information, the earlier the ranking position. The ranking index may be calculated based on the popularity value of the service type information and the relevance of the corresponding keyword to the sub-search term. The ranking index of a certain service type information may be expressed as p1=s1×q3+c1×q4, and q3+q4=100%. Wherein, S1 may represent a heat value of a certain service type information, C1 may represent a correlation degree between a keyword corresponding to a certain service type information and a sub-search word, Q3 may represent a weight given to S1, and a value of Q3 may be set based on a ranking requirement. Q4 may represent a weight given to C1, and the value of Q4 may be set based on the ordering requirements.

It should be noted that, if keywords related to the sub-search terms still do not exist in the history database, other popular service types may be summarized and displayed to the application end 110, so as to provide other alternative service types for the user. It should be further explained that the hot service type may be determined based on the hot value. Specifically, there are multiple service types in the history database, and the more the number of history clicks of each service type, the higher the hotness value. The historical number of clicks may be the number of times the user has previously clicked or confirmed the service type. Accordingly, when setting the hot service type, a hot value threshold, for example, 40%, may be set in advance. When the popularity value of a certain service type is higher than 40%, the service type can be set as a popular service type. A hotness service display table may be generated to summarize the hotness service types in the history database to provide the user with alternative service types.

It should be further explained that the popularity value of a certain service type may be expressed as s=p1/P2, where P1 may be expressed as the number of historical clicks of the service type and P2 may be expressed as the sum of the number of historical clicks of all service types in the historical database.

In an exemplary embodiment, as shown in fig. 8, after the step of generating the associated service display table and transmitting the associated service display table to the target application end, the cloud service end obtains corresponding service type information based on the keyword information related to the sub-search word,

and S271, the cloud service end sets the search word information which is different from all the keyword information in the history database as a word to be updated.

S272, the cloud server acquires the historical searching times of the word to be updated and the service type information corresponding to the word to be updated.

S273, the cloud server judges whether the historical search times are larger than a time threshold.

And S274, if the number of times threshold is larger than or equal to the number of times threshold, the cloud service terminal aggregates the word to be updated and the corresponding service type information thereof into the keyword information of the historical database, and updates the keyword information.

And S275, if the number of times is smaller than the threshold value, deleting the word to be updated by the cloud server.

It should be noted that, in order to improve the perfection of the keyword information in the history database, after the user searches, the keyword information in the history database may be updated based on the search word of the user.

As an example, when the present invention is used in a scenario of a life service application, the keyword information in the history database may be "go to western restaurant drink", "go to chinese dining hall drink", "go to restaurant drink", "eat western dining", or the like. If the search word information in the voice information of the user is 'go restaurant to drink beer', the search word information is different from all keyword information in the history database, and whether the search word information is related to the keywords or not is judged through relevance calculation or word splitting processing, and the user can search related service types. If the search word and the corresponding service type information are aggregated into the keyword information in the history database, when the search word "go to restaurant and drink beer" exists in the voice information of the next time, the cloud server 120 can quickly query the same search word and the corresponding service type information in the history database. Thereby improving the acquisition speed of the service type information.

It should be noted that, the search word information different from all the keyword information in the history database may be set as the word to be updated, and when the history search frequency of the word to be updated is greater than or equal to the frequency threshold, the "go to restaurant and drink beer" may be described as the search word information commonly used by the user. At this time, the search word of 'go to restaurant drink beer' and the corresponding service type information can be aggregated into the keyword information of the history database so as to perfect the keyword information. The number of times threshold may be set based on actual demand, and may be 10 times, 12 times, or other times. When the historical search times of the search word information is smaller than the times threshold value, the fact that the user drinks beer from restaurants is the unusual search word information of the user is indicated, the search word is not required to be aggregated into the keyword information of the historical database, the search word can be deleted, and the storage space is released.

Therefore, in the above scheme, for various applications, a voice receiving unit is introduced into the application to obtain voice information of the user, and then the relevant service types in the cloud service end are queried by extracting the search term information. And the search term information is subjected to word splitting processing, so that the search range of the search term information is improved, and related service type information is more easily queried. The generated service presentation list is then available for selection by the user. Therefore, the speed of providing the required service for the user by the application is improved, the use feeling of the user is improved, and the application utilization rate of the user is improved.

It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present invention.

In an embodiment, a voice-based cloud service search system is provided, which corresponds to the voice-based cloud service search method in the above embodiment one by one. As shown in fig. 9, the voice-based cloud service search system includes a target application end 110 and a cloud service end 120, where the target application end 110 is connected with the cloud service end 120 through a network, the target application end 110 includes a receiving module 201 and an extracting module 202, and the cloud service end 120 includes a judging module 203, a processing module 204, a word splitting module 205, an obtaining module 206 and a generating module 207.

In one embodiment, the extraction module 202 is configured, in particular,

and identifying the voice message, and acquiring age layer information corresponding to the voice message, wherein the age layer information comprises an advanced age layer, a middle age layer and a low age layer.

And carrying out correction processing on the voice message based on the age group information to generate correction information.

And extracting search word information in the correction information.

In one embodiment, the extraction module 202 is further configured to,

based on the age group information, inputting the voice message into a corresponding preset voice recognition model, and outputting correction information, wherein the voice recognition model comprises an advanced voice recognition model, a middle-aged voice recognition model and a low-aged voice recognition model.

In one embodiment, the determining module 203 is specifically configured to,

and calculating the relevance between each keyword information and the search word information in the historical database.

And judging whether the correlation degree is larger than a correlation degree threshold value or not.

And if the correlation threshold is larger than or equal to the correlation threshold, keyword information related to the search word information exists in the history database, and if the correlation threshold is smaller than the correlation threshold, keyword information related to the search word information does not exist in the history database.

In one embodiment, the processing module 204 is configured, in particular,

and if the keyword information related to the search word information exists, acquiring the related relevance and service type information corresponding to the related keyword information.

And sorting the service type information based on the correlation degree to generate a related service display table.

In one embodiment, the processing module 204 is further configured to,

and acquiring the historical click times of the service type information.

And calculating the heat value of the service type information based on the historical click times.

And calculating a ranking index of the service type information based on the heat value and the correlation.

And ordering the service type information based on the ordering index to generate an associated service display table.

In one embodiment, the acquisition module 206 is configured, in particular,

and judging whether keyword information related to the sub-search word information exists in the historical database.

And if the keyword information related to the sub-search word information exists, acquiring the keyword information related to the sub-search word information.

And if the keyword information related to the sub-search word information does not exist, generating a hot service display table based on the hot value of the service type information in the history database.

The invention provides a cloud service search system based on voice, which aims at various applications, introduces a voice receiving unit into the applications, acquires voice information of a user, and queries related service types in a cloud service end by extracting search word information. And the search term information is subjected to word splitting processing, so that the search range of the search term information is improved, and related service type information is more easily queried. The generated service presentation list is then available for selection by the user. Therefore, the speed of providing the required service for the user by the application is improved, the use feeling of the user is improved, and the application utilization rate of the user is improved.

Specific limitations regarding the service search system for speech recognition may be found in the above limitations regarding the service search method for speech recognition, and will not be described in detail herein. The various modules in the speech recognition service search system described above may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-described division of the functional units and modules is illustrated, and in practical application, the above-described functional distribution may be performed by different functional units and modules according to needs, i.e. the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-described functions.

The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.

Claims

1. The cloud service searching method based on the voice is characterized by being applied to a target application end and a cloud service end, wherein the target application end is connected with the cloud service end through a network, and comprises the following steps:

2. The voice-based cloud service search method according to claim 1, wherein the target application extracting search term information in the voice information includes:

the target application end identifies the voice message and acquires age layer information corresponding to the voice message, wherein the age layer information comprises an advanced age layer, a middle age layer and a low age layer;

based on the age group information, the target application end carries out correction processing on the voice message to generate correction information;

and the target application end extracts search word information in the correction information.

3. The method of claim 2, wherein the target application performs correction processing on the voice message based on the age group information, and generating correction information includes:

based on the age group information, the target application end inputs the voice message into a corresponding preset voice recognition model and outputs correction information, wherein the voice recognition model comprises an advanced age group voice recognition model, a middle age group voice recognition model and a low age group voice recognition model.

4. The voice-based cloud service search method according to claim 1, wherein the cloud service end judging whether keyword information related to the search word information exists in a history database comprises:

The cloud server calculates the relativity of each keyword information and the search word information in the historical database;

the cloud server judges whether the correlation is larger than a correlation threshold;

and if the correlation threshold value is larger than or equal to the correlation threshold value, determining that the keyword information related to the search word information exists in the historical database, and if the correlation threshold value is smaller than the correlation threshold value, determining that the keyword information related to the search word information does not exist in the historical database.

5. The method for searching for a cloud service based on voice as claimed in claim 1, wherein if there is keyword information related to the search word information, the cloud server obtains service type information corresponding to the related keyword information, and generating an associated service display table includes:

if the keyword information related to the search word information exists, the cloud server acquires the related degree of relativity and service type information corresponding to the related keyword information;

and based on the correlation, the cloud service end sorts the service type information to generate a related service display table.

6. The voice-based cloud service search method of claim 5, wherein the cloud service end orders the service type information based on the relevance, and generating an associated service presentation table comprises:

The cloud server acquires the historical click times of the service type information;

based on the historical click times, the cloud server calculates a heat value of the service type information;

and based on the heat value and the correlation, the cloud service end calculates a ranking index of the service type information.

And based on the ordering index, the cloud service end orders the service type information and generates an associated service display table.

7. The voice-based cloud service search method according to claim 6, wherein the ranking index is represented by p=sq1+cq2, wherein S represents a heat value of certain service type information, C represents a correlation degree between a keyword corresponding to certain service type information and a search word, Q1 represents a weight given to S, and Q2 represents a weight given to C.

8. The method of claim 1, wherein the obtaining, by the cloud server, keyword information related to the sub-search word information in the history database includes:

the cloud server judges whether keyword information related to the sub-search word information exists in the history database;

If the keyword information related to the sub-search word information exists, the cloud server acquires the keyword information related to the sub-search word information;

and if the keyword information related to the sub-search word information does not exist, the cloud service end generates a hot service display table based on the hot value of the service type information in the history database.

9. The method for searching for cloud service based on voice as claimed in claim 1, wherein the cloud service terminal obtains corresponding service type information based on the keyword information related to the sub-search word, and further comprises, after generating the associated service display table:

the cloud server sets the search word information which is different from all the keyword information in the historical database as a word to be updated;

and the cloud server acquires the historical searching times of the word to be updated and the service type information corresponding to the word to be updated.

The cloud server judges whether the historical searching times are larger than a time threshold;

and if the number of times threshold is greater than or equal to the number of times threshold, the cloud server aggregates the word to be updated and the corresponding service type information thereof into the keyword information of the historical database, and updates the keyword information.

And if the number of times is smaller than the number of times threshold, deleting the word to be updated by the cloud server.

10. The cloud service search system based on the voice is characterized by comprising a target application end and a cloud service end, wherein the target application end is connected with the cloud service end through a network, and the target application end comprises:

the cloud service end comprises: