CN116911317A

CN116911317A - Question-answering processing method, device, equipment, vehicle and storage medium

Info

Publication number: CN116911317A
Application number: CN202211678165.5A
Authority: CN
Inventors: 黄海涛; 王松怡
Original assignee: Beijing Co Wheels Technology Co Ltd
Current assignee: Beijing Co Wheels Technology Co Ltd
Priority date: 2022-12-26
Filing date: 2022-12-26
Publication date: 2023-10-20

Abstract

The disclosure relates to a question-answering processing method, a question-answering processing device, question-answering equipment and a storage medium, wherein the question-answering processing method comprises the following steps: receiving questioning information input by a user; processing the questioning information to obtain target vehicle information and target intention information; and inquiring in a preset map database based on the target vehicle information and the target intention information to obtain the answer information matched with the question information. Compared with the technical scheme of the embodiment of the disclosure that the whole question and answer is directly searched in the question and answer database in a semantic similarity mode, the target vehicle information and the target intention information are used as graph nodes, the graph database is searched and matched with the answer information, answer searching efficiency is improved, after the vehicle information and the intention information in the question information are respectively extracted, the answer accuracy is improved by considering the semantic level and the sentence front-back relation in the question information based on the vehicle information and the intention matching answer information.

Description

Question-answering processing method, device, equipment, vehicle and storage medium

Technical Field

The disclosure relates to the technical field of vehicle intellectualization, in particular to a question-answering processing method, a question-answering processing device, question-answering processing equipment, a vehicle and a storage medium.

Background

At present, with the development of computer internet technology, various intelligent technologies are widely applied. The user can accurately acquire the required internet of vehicles industry knowledge by using the question-answering system, and the method has become an important way for acquiring information.

The existing question and answer system is mostly characterized in that the answer information input by a user is obtained, answer information corresponding to the question information is queried from a preset question and answer database in a semantic similarity mode, and the answer information is fed back to the user.

In the existing question-answer system, the question information is taken as a whole to search the corresponding answer information from the question-answer database, so that the search efficiency is low, the relation between the semantic level and the sentence front-back in the question information is not considered, and the accuracy of the answer information is low.

Disclosure of Invention

The invention provides a question and answer processing method, a device, equipment and a storage medium, which are used for recalling answer information in a graph database according to target vehicle information and intention information by extracting vehicle type information and the intention information from question information, so that the retrieval efficiency is improved, the relation between semantic layers and sentences in the question information is considered, and the accuracy of the answer information is improved.

In a first aspect, an embodiment of the present disclosure provides a question-answering processing method, including:

Receiving questioning information input by a user;

processing the questioning information to obtain target vehicle information and target intention information;

and inquiring in a preset map database based on the target vehicle information and the target intention information to obtain the answer information matched with the question information.

In a second aspect, an embodiment of the present disclosure provides a question-answering processing apparatus, including:

the questioning information receiving module is used for receiving questioning information input by a user;

the questioning information processing module is used for processing the questioning information to obtain target vehicle information and target intention information;

and the reply information determining module is used for inquiring in a preset map database based on the target vehicle information and the target intention information to obtain reply information matched with the question information.

In a third aspect, an embodiment of the present disclosure provides an electronic device, including:

a memory;

a processor; and

a computer program;

wherein the computer program is stored in the memory and configured to be executed by the processor to implement the question-answering processing method according to the first aspect.

In a fourth aspect, an embodiment of the present disclosure provides a vehicle that performs the question-answering method according to the first aspect.

In a fifth aspect, embodiments of the present disclosure provide a computer-readable storage medium having stored thereon a computer program that is executed by a processor to implement the question-answering processing method according to the first aspect.

The method, device, equipment and storage medium for processing questions and answers provided by the embodiment of the disclosure, wherein the method comprises the following steps: receiving questioning information input by a user; processing the questioning information to obtain target vehicle information and target intention information; and inquiring in a preset map database based on the target vehicle information and the target intention information to obtain the answer information matched with the question information. Compared with the technical scheme of the embodiment of the disclosure that the whole question and answer is directly searched in the question and answer database in a semantic similarity mode, the target vehicle information and the target intention information are used as graph nodes, the graph database is searched, the answer information is matched, answer searching efficiency is improved, after the vehicle information and the intention information in the question information are respectively extracted, the answer accuracy is improved based on the fact that the vehicle information and the intention match the answer information, the semantic level and the sentence front-back relation in the question information are considered, and the answer accuracy is improved.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the disclosure and together with the description, serve to explain the principles of the disclosure.

In order to more clearly illustrate the embodiments of the present disclosure or the solutions in the prior art, the drawings that are required for the description of the embodiments or the prior art will be briefly described below, and it will be obvious to those skilled in the art that other drawings can be obtained from these drawings without inventive effort.

FIG. 1 is a flow chart of a question-answering method in an embodiment of the present disclosure;

FIG. 2 is a schematic diagram of an identification bit setting provided in an embodiment of the present disclosure;

fig. 3 is a flowchart of a question-answering processing method provided in an embodiment of the present disclosure;

fig. 4 is a flowchart of a question-answering processing method provided in an embodiment of the present disclosure;

FIG. 5 is a flowchart of an intent recognition model training method provided by an embodiment of the present disclosure;

fig. 6 is a flowchart of a question-answering processing method provided in an embodiment of the present disclosure;

fig. 7 is a block diagram of a question-answering processing method according to an embodiment of the present disclosure;

Fig. 8 is a schematic structural diagram of a question-answering processing apparatus according to an embodiment of the present disclosure;

fig. 9 is a schematic structural diagram of an electronic device according to an embodiment of the disclosure.

Detailed Description

In order that the above objects, features and advantages of the present disclosure may be more clearly understood, a further description of aspects of the present disclosure will be provided below. It should be noted that, without conflict, the embodiments of the present disclosure and features in the embodiments may be combined with each other.

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure have been shown in the accompanying drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.

It should be understood that the various steps recited in the method embodiments of the present disclosure may be performed in a different order and/or performed in parallel. Furthermore, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the present disclosure is not limited in this respect.

The term "including" and variations thereof as used herein are intended to be open-ended, i.e., including, but not limited to. The term "based on" is based at least in part on. The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments. Related definitions of other terms will be given in the description below.

It should be noted that the terms "first," "second," and the like in this disclosure are merely used to distinguish between different devices, modules, or units and are not used to define an order or interdependence of functions performed by the devices, modules, or units.

It should be noted that references to "one", "a plurality" and "a plurality" in this disclosure are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be understood as "one or more" unless the context clearly indicates otherwise.

The names of messages or information exchanged between the various devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

The following is a brief description of the terminology involved in the embodiments of the present disclosure.

The existing question-answering system mostly obtains question information by acquiring audio information input by a user, identifies the audio information, inquires answer information corresponding to the question information from a preset question-answering database in a semantic similarity mode, and feeds the answer information back to the user. For example: after the identification by the audio information, the question information obtained is "how much is the vehicle insurance of the XX brand XX model? ". In the prior art, in a preset question-answer database, question information with highest similarity with the question information is selected, answer information corresponding to the question information is obtained, and the answer information is fed back to a user.

In the above scheme, the similarity calculation using the whole question information belongs to a fuzzy matching scheme, because a question is a sentence and may include many nonsensical words, for example: "do you return to me, the AA brand BB size now sell little money? "how much of" is "how much", "you can answer me", etc. belong to nonsensical words, but in the question-answer matching process, it is still to be used as a part of content in a question information, and answer information is retrieved in a question-answer database, which results in a decrease in retrieval efficiency. Furthermore, in the prior art, the question information is taken as a whole, the corresponding answer information is searched in the database, and the relation before and after the semantic level and the sentence in the question information cannot be considered comprehensively, so that the answer accuracy is low.

For this reason, the embodiment of the disclosure provides a question-answering processing method, which mainly includes: receiving questioning information input by a user; processing the questioning information to obtain target vehicle information and target intention information; the method comprises the steps of carrying out a first treatment on the surface of the And inquiring in a preset map database based on the target vehicle information and the target intention information to obtain the answer information matched with the question information.

Compared with the technical scheme of the embodiment of the disclosure that the whole question and answer is directly searched in the question and answer database in a semantic similarity mode, the target vehicle information and the target intention information are used as graph nodes, the graph database is searched, the answer information is matched, no meaningless words in the question information are required to be processed, answer searching efficiency is improved, after the vehicle information and the intention information in the question information are respectively extracted, the answer accuracy is improved by considering the semantic layer and the sentence front-back relation in the question information based on the vehicle information and the intention matching answer information.

The method for processing questions and answers provided in the present disclosure will be described in detail with reference to specific embodiments and drawings.

Example 1

Fig. 1 is a flowchart of a question-answering processing method in an embodiment of the disclosure, where the embodiment is applicable to a case of intelligent voice questions and answers, the method may be performed by a question-answering processing device, the question-answering processing device may be implemented in a software and/or hardware manner, and the question-answering processing device may be configured in an electronic device. The electronic device may be configured in a vehicle terminal. Further, the electronic device may be any device having a voice acquisition function. The electronic device can be deployed in the places such as automobile sales shops, automobile exhibitions and the like to help users to inquire about the relevant information of the vehicle.

Specifically, as shown in fig. 1, the question-answering processing method provided in the embodiment of the present disclosure mainly includes steps S101 to S104.

S101, receiving question information input by a user.

The question information refers to a question to be asked, which is input by a user. Further, receiving a question audio input by a user, and performing voice recognition on the question audio to obtain question information. The questioning audio is audio information acquired by the audio acquisition device of the electronic equipment.

In the embodiment of the disclosure, an audio acquisition device of the electronic device acquires the questioning audio input by a user and sends the questioning audio to a questioning and answering processing device, and the questioning audio is processed after the questioning audio is received by the questioning and answering processing device to obtain questioning information.

In one embodiment of the present disclosure, processing the question audio to obtain question information includes: automatic speech recognition technology (Automatic Speech Recognition, ASR) is used to convert the question audio into its corresponding question information.

S102, processing the question information to obtain target vehicle information and target intention information;

in the embodiment of the disclosure, the question information is processed respectively to obtain the target vehicle information and the target intention information. For example: processing the questioning information by adopting an entity naming recognition model to obtain target vehicle information; and processing the questioning information by adopting a pre-trained intention recognition model to obtain target intention information.

The vehicle information comprises a vehicle brand and a vehicle model, and further, the target vehicle information refers to vehicle information extracted from the questioning information.

In the embodiment of the disclosure, the target vehicle information is extracted from the question information through a trained target vehicle information extraction model, wherein the trained target vehicle information extraction model can be an entity naming identification model.

In the embodiment of the disclosure, the target vehicle information base may be pre-constructed, the target vehicle information in the target vehicle information base is matched with the question information, and the matched target vehicle information is used as the target vehicle information extracted from the question information.

In the embodiment of the disclosure, the target vehicle information in the target vehicle information base exists in the form of a vehicle brand and a vehicle model, and the vehicle brand and the question information can be matched to obtain the target vehicle brand. And respectively matching the vehicle signals corresponding to the vehicle brands with the questioning information after the vehicle brands included in the questioning information to obtain the target vehicle model. And taking the brand of the target vehicle and the model of the target vehicle as target vehicle information extracted from the questioning information.

Wherein the target intention information refers to an intention that the user wants to inquire, for example: the intention information in the vehicle field may include: sales price, purchase loans, vehicle maintenance, vehicle cruising, quality assurance limits, and the like. The target intention information refers to intention information related to the vehicle extracted from the question information.

In the embodiment of the disclosure, semantic understanding processing can be performed on the question information to obtain target intention information. Wherein, the semantic understanding refers to the question information being converted into machine readable information. For example: the question information is "what money is paid for a bare car of the AA brand CC size", and the target intention obtained after processing is "price".

In the embodiment of the disclosure, the question information is input into a pre-trained intention recognition model, and the intention recognition model processes the question information to obtain target intention information.

S103, inquiring in a preset map database based on the target vehicle information and the target intention information to obtain answer information matched with the question information.

In one embodiment of the present disclosure, the preset map database includes correspondence relations among vehicle information, intention information, and reply information. Inquiring in the corresponding relation based on the target vehicle information and the target intention information to obtain reply information corresponding to the target vehicle information and the target intention information, and taking the reply information as reply information matched with the questioning information.

Specifically, the correspondence of the target vehicle information, the intention information, and the reply information is stored in the form of a data triplet, for example: { AA brand BB size, price, specific price value }.

In the embodiment of the disclosure, the corresponding relation stored in the form of the data triples is stored in a preset graph database in the form of key graph nodes, so that the retrieval speed is improved.

In one embodiment of the present disclosure, the query is performed in preset map data based on the target vehicle information and the target intention information, to obtain reply information matched with the question information, including: inquiring in a preset problem map based on the target vehicle information and the target intention information to obtain target problem information matched with the question information; inquiring in a preset answer map based on the target question information to obtain answer information matched with the target question information, wherein the preset question map and the preset answer map are stored in the preset map database.

In one embodiment of the disclosure, user recording data of different scenes is collected (on the premise of obtaining user consent), the user recording data is converted into problem data, the problem data is stored in a Hive database, vehicle types and intentions in the problem data are obtained, and an abstract mathematical triplet structure is innovatively created, and the abstract mathematical triplet structure represents (vehicle types, problems and intentions). For example: the user asks what the price is for the AA brand BB size, (AA brand BB size, price). And finally, forming a preset problem map shown in figure 2 based on the node commonality relation. As shown in fig. 2, if the target vehicle information is a vehicle type 1 and the target intention information is an intention 1, it may be determined that the corresponding target problem information is a problem 1. The target problem information is determined according to a preset problem map stored in the map database in advance.

In the embodiment of the disclosure, the preset answer spectrum is a knowledge spectrum constructed according to question-answer logic, and the preset answer spectrum is stored in a graph database to determine the logic relationship between the questions and the answers, so that the retrieval efficiency is improved.

The graph database can be any one or more of a Neo4j graph database, a FlockDB graph database, an Allegrograph database, a GraphDB graph database and an InfinieGraph graph database. Optionally, the graph database in the embodiments of the present disclosure is a Neo4j graph database.

The Neo4j graph database is a popular graph database and is open-source. The community version of Neo4j has been shifted from following the AGPL license agreement to following the GPL license agreement. Neo4j is based on Java implementation, compatible with ACID features, and also supports other programming languages such as Ruby and Python.

In one embodiment of the disclosure, after determining target question information matched with question information, similarity between the target question information and the question information is calculated, and after determining that the similarity is greater than or equal to a preset similarity threshold, query is performed in a preset answer map based on the target question information to obtain answer information matched with the target question information. If the similarity is smaller than a preset similarity threshold, acquiring the spam from the spam database, and feeding the spam back to the user as reply information.

In one embodiment of the present disclosure, after obtaining the reply information, the method further includes: and feeding back the reply information to the user.

Specifically, the reply information is converted into an audio signal by using Text To Speech (TTS) technology, and the audio signal is played by using an audio playing device in the electronic device, so that the user can hear the corresponding answer.

Specifically, the reply information is displayed in a text form in a display screen of the electronic equipment, so that a user can watch the reply text corresponding to the question audio.

The question-answering processing method provided by the embodiment of the disclosure comprises the following steps: receiving questioning information input by a user; processing the questioning information to obtain target vehicle information and target intention information; and inquiring in a preset map database based on the target vehicle information and the target intention information to obtain the answer information matched with the question information. Compared with the technical scheme of the embodiment of the disclosure that the whole question and answer is directly searched in the question and answer database in a semantic similarity mode, the target vehicle information and the target intention information are used as graph nodes, the graph database is searched, the answer information is matched, answer searching efficiency is improved, after the vehicle information and the intention information in the question information are respectively extracted, the answer accuracy is improved based on the fact that the vehicle information and the intention match the answer information, the semantic level and the sentence front-back relation in the question information are considered, and the answer accuracy is improved.

Example two

On the basis of the above embodiment, the question-answer processing method is further optimized in the embodiment of the present disclosure, and as shown in fig. 3, the optimized question-answer processing method mainly includes steps S201 to S205.

S201, receiving question information input by a user.

The step S201 provided in the embodiment of the present disclosure is the same as the specific execution flow of the step S101 provided in the above embodiment, and specifically reference may be made to the description in the above embodiment, which is not specifically limited.

S202, extracting target vehicle information from the questioning information.

The specific execution flow of "extracting the target vehicle information from the question information" 2 provided in the embodiment of the present disclosure is the same as that provided in the above embodiment, and specifically reference may be made to the description in the above embodiment, and the embodiment of the present disclosure is not limited thereto.

S203, processing the questioning information by utilizing a pre-trained intention recognition model to obtain a plurality of initial intention information and probability values corresponding to the initial intention information.

In one embodiment of the present disclosure, after feature extraction is performed on the question information, a sentence vector is obtained, the sentence vector is input into a pre-trained intent recognition model, and after the intent recognition model processes the sentence vector, a plurality of initial intent information and probability values corresponding to the respective intent information are obtained.

In the embodiment of the present disclosure, the sum of probability values corresponding to the plurality of initial intention information is equal to 1.

S204, determining initial intention information corresponding to the maximum probability value as target intention information.

In the embodiment of the present disclosure, the plurality of probability values obtained in step S203 are compared, and initial intention information corresponding to the maximum probability value is taken as target intention information.

In one embodiment of the present disclosure, after feature extraction is performed on question information, a sentence vector is obtained, the sentence vector is input into a pre-trained intent recognition model, the intent recognition model processes the sentence vector to obtain one piece of intent information, and the intent information output by the intent recognition model is directly used as target intent information.

In one embodiment of the present disclosure, the determining initial intention information corresponding to the maximum probability value as target intention information includes: comparing the maximum probability value with a preset probability threshold value; and if the maximum probability value is larger than the probability threshold value, determining initial intention information corresponding to the maximum probability value as target intention information.

The preset probability threshold may be set according to actual situations, and optionally, the preset probability threshold may be any value between 0.5 and 1, which is not specifically limited in the embodiments of the present disclosure.

In the embodiment of the present disclosure, the multiple probability values obtained in step S203 are compared, after the maximum probability value is determined, the maximum probability value is compared with a preset probability threshold, and if the maximum probability value is greater than the preset probability threshold, the initial intention information corresponding to the maximum probability value is used as the target intention information.

In the embodiment of the disclosure, under the condition that the maximum probability value is larger than the preset probability threshold value, the initial intention information corresponding to the maximum probability value is used as the target intention information, so that the accuracy of the intention information is further improved.

And if the maximum probability value is smaller than or equal to the preset probability threshold value, sequencing the probability values in order from large to small, and acquiring initial intention information corresponding to the probability values of the preset quantity as target intention information. In other words, if the maximum probability value is less than or equal to the probability threshold, steps S304-S309 in the third embodiment described below may be performed, and specific information may refer to descriptions in the following embodiments, and in the embodiments of the present disclosure, details are not repeated.

S205, inquiring in a preset map database based on the target vehicle information and the target intention information to obtain answer information matched with the question information.

Step S205 provided in the embodiment of the present disclosure is the same as the specific execution flow of step S104 provided in the above embodiment, and specific reference may be made to the description in the above embodiment, which is not specifically limited.

In the embodiment of the disclosure, the probability values corresponding to the plurality of initial intention information are obtained through the intention recognition model, and the initial intention information corresponding to the maximum probability value is used as the target intention information, so that the accuracy of the intention information can be improved, and the accuracy of the reply information is further improved.

Example III

On the basis of the above embodiment, the question-answer processing method is further optimized in the embodiment of the present disclosure, and as shown in fig. 4, the optimized question-answer processing method mainly includes steps S301 to S309.

S301, receiving question information input by a user.

S302, extracting target vehicle information from the questioning information.

S303, processing the questioning information by utilizing a pre-trained intention recognition model to obtain a plurality of initial intention information and probability values corresponding to the initial intention information.

Steps S301 to S303 provided in the embodiments of the present disclosure are the same as the specific execution flow of steps S201 to S203 provided in the above embodiments, and specific reference may be made to the description in the above embodiments, which is not specifically limited in the embodiments of the present disclosure.

S304, sorting the probability values in the order from big to small.

In the embodiment of the present disclosure, the plurality of probability values obtained in step S303 are sequentially ordered in order from large to small. It should be noted that, any numerical ranking manner may be used to rank the probability values, and the ranking manner of the probability values is not specifically limited in the embodiments of the present disclosure.

S305, acquiring initial intention information corresponding to the probability values of the preset quantity before sorting as target intention information.

The probability value ranking is lower the later, the probability that the initial intention is the target intention information is lower, so that initial intention information corresponding to a plurality of probability values ranked at the front can be selected as the target intention information.

The preset number can be set according to actual conditions, and specifically, the preset number can be any one of 3,4 and 5. In order to reduce the amount of computation of the subsequent text similarity, the preset number is set to 3 in the embodiment of the present disclosure.

Specifically, initial intention information corresponding to the first 3 probability values is used as target intention information. The target intention information obtained in step S305 is plural.

S306, inquiring in a preset problem map based on the target vehicle information and the target intention information aiming at each target intention information to obtain a plurality of problem information matched with the question information.

In the embodiment of the present disclosure, for the plurality of target intention information obtained in step S305, query is performed in a preset problem map based on the target vehicle information and the target intention information, so as to obtain problem information corresponding to each target intention information, i.e., obtain a plurality of problem information, where the number of problem information is the same as the number of target intention information.

It should be noted that, based on the target vehicle rationality information and the target intention information, the process of obtaining the problem information and the description in the above embodiments are queried in the preset problem map, and specific reference may be made to the description in the above embodiments, and the embodiments of the disclosure are not limited in any way.

S307, calculating the similarity of the question information and the question information according to each question information.

In the embodiment of the disclosure, text similarity calculation is performed on each question information and the question information, so as to obtain the similarity corresponding to each text question information.

The text similarity calculation method comprises one or more of the following steps: cosine similarity, edit distance, TF-IDF, etc. The similarity calculation mode is not limited in the embodiments of the present disclosure.

And S308, determining the problem information corresponding to the maximum similarity as target problem information.

In the embodiment of the disclosure, the problem information corresponding to the maximum similarity is taken as the target problem information, the similarity of the problem information and the question information is the maximum, that is, the probability that the problem information is the same as the question information is the highest, that is, the probability that the problem information is the problem that the user wants to ask is the highest, so that the accuracy of the reply information can be further improved.

S309, inquiring in a preset answer map based on the target question information to obtain answer information matched with the target question information, wherein the preset question map and the preset answer map are stored in the preset map database.

In the embodiment of the disclosure, a plurality of target intention information is obtained through an intention recognition model, corresponding problem information is calculated for each target intention information, and in the process of calculating the similarity between the problem information and the question information, the problem information with the largest similarity is determined as the target problem information.

Example IV

On the basis of the foregoing embodiments, the present disclosure provides a training method for an intent recognition model, and specifically, as shown in fig. 5, the training method for an intent recognition model provided in the embodiments of the present disclosure mainly includes steps S401 to S403.

S401, acquiring sample question information.

In the embodiment of the disclosure, under the condition of obtaining user consent, question information of users in different scenes is obtained.

In one embodiment of the present disclosure, a user's question audio is obtained, and the audio is converted into question information, resulting in sample question information. Related questioning information about the vehicle issued by the user on the network is acquired from the network through a crawler technology.

S402, carrying out intention labeling on the sample question information to obtain labeling information corresponding to the sample question information.

In the embodiment of the present disclosure, the intention labeling is performed on the sample question information, which may be a manual labeling mode or a machine labeling mode, and the embodiment of the present disclosure is not limited in detail.

In the practice of the present disclosure, text data is intended to be labeled, for example: marking the "help me inquire AA brand BB model to quality assurance period" as "quality assurance period"; what will be "what is insurance of the AA brand BB size? "marked" insurance "; the "what the AA brand BB size is for cruising" is noted as "cruising".

S403, training a preset network model by using the sample questioning information and the labeling information to obtain a pre-trained intention recognition model.

The preset network model may be any one of deep learning classification models such as bilstm+ Attention, textCNN, textRNN, BERT, and in the embodiment of the present disclosure, the specific type of the preset network model is not specific.

In the embodiment of the disclosure, the sample questioning information is input into a preset network model, the preset network model outputs intention information, the similarity between the output intention information and the labeling information is calculated, the model training is successful when the similarity is greater than or equal to a preset threshold value, and the preset network model is used as a pre-trained intention recognition model; and when the similarity is smaller than a preset threshold, indicating that model training is unsuccessful, and continuing training the preset network model.

In an embodiment of the present disclosure, training a preset network model by using the sample question information and the labeling information to obtain a pre-trained intention recognition model, including: converting the sample question information into a sample vector; extracting and processing the sample vector by using the preset network model, and outputting intention information; performing similarity calculation on the output intention information and the labeling information to obtain similarity of the output intention information and the labeling information; when the similarity of the two is larger than or equal to a preset threshold value, the preset network model is used as a pre-trained intention recognition model; and when the similarity is smaller than a preset threshold value, continuing training the preset network model.

In one embodiment of the present disclosure, the sample problem information is used with the sequence q= (Q) ₁ ，Q ₂ ，……，Q _n ) Expressed, the intended sequence is represented by i= (I ₁ ，I ₂ ，……，I _n ) To represent. I.e. each sample question information establishes its corresponding identification means.

In one embodiment of the present disclosure, special characters, stop words, etc. in sample question information are removed, a vocabulary is built, a representation of a character variable (char) is used, and the vocabulary is converted into a representation of { word: id }, an intent classification catalog is fixed, and a representation of a category { category: id }. The obtained sample problem information is divided into a training set and a verification set which are respectively used for training the preset network model in different stages.

In one embodiment of the disclosure, sample problem information is converted into a sample vector, that is, features in the sample problem information are extracted, the sample vector is constructed, feature extraction and model training are performed on the sample vector by adopting a BiLSTM+attribute network, and an intention class with the highest classification probability of a user input sentence q, that is, a first intention expressed in the embodiment, is output.

In one embodiment of the present disclosure, a question-answering processing method is provided in an embodiment of the present disclosure, as shown in fig. 6, where the question-answering processing method provided in the embodiment of the present disclosure mainly includes the following steps:

Specifically, S501 receives a question audio input by a user, S502 performs ASR processing on the question audio, and obtains question information. S503, inputting the questioning information into the NER model to obtain target vehicle information, and S504, inputting the questioning information into the intention recognition model to obtain a plurality of initial intention information and probability values corresponding to the initial intention information. S505, judging whether the maximum probability value is larger than a preset probability threshold, and executing step S506 if the maximum probability value is larger than the preset probability threshold; s506, taking initial intention information corresponding to the maximum probability threshold value as target intention information. If the maximum probability value is less than or equal to the preset probability threshold, steps S507 and S507 are executed, the probability values are sorted in the order from the big to the small, and the initial intention information corresponding to the probability value sorted in the previous 3 is used as the target intention information. S508, inquiring in a preset problem map based on the target vehicle information and the target intention information to obtain a plurality of problem information matched with the question information. S509, calculating the similarity between the question information and the question information. S510, comparing the maximum similarity with a preset similarity threshold, and if the maximum similarity is greater than or equal to the similarity threshold, executing steps S511 and S511, and taking the problem information corresponding to the maximum similarity as target problem information. S512, inquiring in a preset answer map based on the target question information to obtain answer information matched with the target question information, and S513, converting the answer information into answer audio by using a TTS technology to play. If the maximum similarity is smaller than the similarity threshold, step S514 is executed to obtain spam, and the spam is played.

In the embodiment of the disclosure, compared with the method for identifying the intention by adopting a machine learning probability statistical method, the method for identifying the object by adopting a deep learning language model naming entity and identifying the intention to determine the question-answer key points in the automobile field respectively, and then based on the retrieval of a graph database, solves the problems of lower model generalization capability of answer matching and lower precision of the machine learning model from the semantic level. Compared with the method of searching by adopting the semantic similarity direct full question and answer, the method and the device have the advantages that the vehicle type and the intention are determined through semantic analysis, then semantic calculation is performed after the graph database recalls the result, the number of times of semantic similarity calculation is greatly reduced, and the service performance is improved. The retrieval efficiency and the accuracy are higher. The interactive mode is adopted on the vehicle type positioning, the user stands at the angle, meaningless answers are avoided, and the optimal answers are determined through recall and sequencing, so that the question-answer logic is more accurate. And the user experience is improved.

In the embodiment of the present disclosure, a block diagram of a question-answer processing method is provided, and as shown in fig. 7, the method mainly includes an intelligent device 71, a voice recognition module 72, a semantic analysis module 73, a dialogue management module 74, and a voice synthesis module 75. The intelligent device 71 is configured to obtain a question audio input by a user, transmit the question audio to the voice recognition module 72, the voice recognition module 72 is configured to receive the question audio, convert the question audio into question information, and transmit the question information to the semantic analysis module 73, the semantic analysis module 73 is configured to perform semantic analysis on the received question information to obtain question information, and transmit the question information to the dialogue management module 74, the dialogue management module 74 is configured to receive the question information, and match answer information matched with the question information from the graph database, or obtain an spam, transmit the answer information or the spam to the voice synthesis module 75, and after the voice synthesis module 75 receives the answer information or the spam, convert the answer audio into answer audio by using TTS technology, transmit the answer audio to the intelligent device 71, and play the answer audio by the intelligent device 71. Furthermore, the method includes: the man-machine interaction module 76 is used for the interaction between the voice recognition module and the user.

Example five

Fig. 8 is a schematic structural diagram of a question-answering processing device in an embodiment of the disclosure, where the embodiment is applicable to an intelligent voice question-answering situation, the question-answering processing device may be implemented in a software and/or hardware manner, and the question-answering processing device may be configured in an electronic device. The electronic device may be configured in a vehicle terminal. Further, the electronic device may be any device having a voice acquisition function. The electronic device can be deployed in the places such as automobile sales shops, automobile exhibitions and the like to help users to inquire about the relevant information of the vehicle.

Specifically, as shown in fig. 8, the question-answering processing apparatus 80 provided in the embodiment of the present disclosure mainly includes: a question information receiving module 81, a question information processing module 82, and a answer information determining module 83.

The questioning information receiving module 81 is configured to receive questioning information input by a user; a question information processing module 82, configured to process the question information to obtain target vehicle information and target intention information; and a reply information determining module 83, configured to query in a preset map database based on the target vehicle information and the target intention information, and obtain reply information matched with the question information.

In one embodiment of the present disclosure, the questioning information processing module 82 comprises: the probability value determining unit is used for processing the questioning information by utilizing a pre-trained intention recognition model to obtain a plurality of initial intention information and probability values corresponding to the initial intention information; and a target intention information determining unit for determining initial intention information corresponding to the maximum probability value as target intention information.

In one embodiment of the present disclosure, a target intention information determination unit includes: the probability value sorting subunit is used for comparing the maximum probability value with a preset probability threshold value; and the target intention information determining subunit is used for determining initial intention information corresponding to the maximum probability value as target intention information if the maximum probability value is larger than the probability threshold value.

In one embodiment of the present disclosure, the questioning information processing module 82 comprises: the probability value determining unit is used for processing the questioning information by utilizing a pre-trained intention recognition model to obtain a plurality of initial intention information and probability values corresponding to the initial intention information; a probability value sorting unit for sorting the probability values in order from big to small; the target intention information determining unit is used for acquiring initial intention information corresponding to the probability values which are ranked in the preset quantity before as target intention information.

In one embodiment of the present disclosure, the reply information determination module 834 includes: the target problem information determining unit is used for inquiring in a preset problem map based on the target vehicle information and the target intention information to obtain target problem information matched with the question information; and the answer information determining unit is used for inquiring in a preset answer map based on the target question information to obtain answer information matched with the target question information, wherein the preset question map and the preset answer map are stored in the preset map database.

In one embodiment of the present disclosure, when the target intention information is at least two or more intention information, a target question information determination unit includes: the problem information determining subunit is used for inquiring in a preset problem map based on the target vehicle information and the target intention information aiming at each target intention information to obtain a plurality of problem information matched with the audio; a similarity calculating subunit, configured to calculate, for each question information, a similarity between the question information and the question information; and the target problem information determining subunit is used for determining the problem information corresponding to the maximum similarity as target problem information.

In one embodiment of the present disclosure, the apparatus further comprises: the model training module is used for constructing an intention recognition model and comprises the following components: the sample questioning information acquisition module is used for acquiring sample questioning information; the intention labeling module is used for labeling the intention of the sample question information to obtain labeling information corresponding to the sample question information; and the model training module is used for training a preset network model by utilizing the sample questioning information and the labeling information to obtain a pre-trained intention recognition model.

In one embodiment of the present disclosure, a model training module includes: a sample vector conversion unit for converting the sample question information into a sample vector; the first intention output unit is used for extracting and processing the sample vector by utilizing the preset network model to obtain output intention information; the similarity calculation unit is used for calculating the similarity between the output intention information and the labeling information to obtain the similarity of the output intention information and the labeling information; the model determining unit is used for taking the preset network model as a pre-trained intention recognition model when the similarity of the model determining unit and the model determining unit is larger than or equal to a preset threshold value; and the model training unit is used for continuing training the preset network model when the similarity is smaller than a preset threshold value.

The question-answering processing device in the embodiment shown in fig. 8 may be used to implement the technical solution of the above method embodiment, and its implementation principle and technical effects are similar, and are not described herein again.

In one embodiment of the present disclosure, a vehicle is provided, where the vehicle is used to implement the question-answering processing method described in any one of the foregoing embodiments, and the implementation principle and technical effects are similar, and are not repeated herein.

Example six

Fig. 9 is a schematic structural diagram of an electronic device in an embodiment of the disclosure. Referring now in particular to fig. 9, a schematic diagram of an electronic device 900 suitable for use in implementing embodiments of the present disclosure is shown. The electronic device 900 in the embodiments of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), wearable terminal devices, and the like, and fixed terminals such as digital TVs, desktop computers, smart home devices, and the like. The electronic device shown in fig. 9 is merely an example, and should not impose any limitations on the functionality and scope of use of embodiments of the present disclosure.

As shown in fig. 9, the electronic device 900 may include a processing means (e.g., a central processor, a graphic processor, etc.) 901, which may perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM) 902 or a program loaded from a storage means 908 into a Random Access Memory (RAM) 903 to implement a question-answering processing method according to embodiments as described in the present disclosure. In the RAM 903, various programs and data necessary for the operation of the terminal device 900 are also stored. The processing apparatus 901, the ROM 902, and the RAM 903 are connected to each other through a bus 909. An input/output (I/O) interface 905 is also connected to the bus 909.

In general, the following devices may be connected to the I/O interface 905: input devices 906 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, and the like; an output device 907 including, for example, a Liquid Crystal Display (LCD), a speaker, a vibrator, and the like; storage 908 including, for example, magnetic tape, hard disk, etc.; and a communication device 904. The communication means 904 may allow the terminal device 900 to communicate wirelessly or by wire with other devices to exchange data. While fig. 9 shows a terminal device 900 having various means, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead.

In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a non-transitory computer readable medium, the computer program containing program code for performing the method shown in the flowcharts, thereby implementing the question-answering processing method as described above. In such an embodiment, the computer program may be downloaded and installed from a network via the communications device 904, or from the storage device 908, or from the ROM 902. When executed by the processing device 901, performs the above-described functions defined in the methods of the embodiments of the present disclosure.

It should be noted that the computer readable medium described in the present disclosure may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present disclosure, however, the computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.

In some implementations, the clients, servers may communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol ), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the internet (e.g., the internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.

The computer readable medium may be contained in the electronic device; or may exist alone without being incorporated into the electronic device.

The computer readable medium carries one or more programs which, when executed by the terminal device, cause the terminal device to implement a method as described in any of the embodiments.

Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages, including, but not limited to, an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The units involved in the embodiments of the present disclosure may be implemented by means of software, or may be implemented by means of hardware. Wherein the names of the units do not constitute a limitation of the units themselves in some cases.

The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Application Specific Standard Product (ASSP), a system on a chip (SOC), a Complex Programmable Logic Device (CPLD), and the like.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by persons skilled in the art that the scope of the disclosure referred to in this disclosure is not limited to the specific combinations of features described above, but also covers other embodiments which may be formed by any combination of features described above or equivalents thereof without departing from the spirit of the disclosure. Such as those described above, are mutually substituted with the technical features having similar functions disclosed in the present disclosure (but not limited thereto).

Moreover, although operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the present disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are example forms of implementing the claims.

Claims

1. A question-answering processing method, characterized by comprising:

receiving questioning information input by a user;

2. The method of claim 1, wherein the processing the question information to obtain target intention information comprises:

processing the questioning information by utilizing a pre-trained intention recognition model to obtain a plurality of initial intention information and probability values corresponding to the initial intention information;

and determining initial intention information corresponding to the maximum probability value as target intention information.

3. The method according to claim 2, wherein the determining initial intention information corresponding to the maximum probability value as target intention information includes:

Comparing the maximum probability value with a preset probability threshold value;

and if the maximum probability value is larger than the probability threshold value, determining initial intention information corresponding to the maximum probability value as target intention information.

4. The method of claim 1, wherein processing the question information to obtain target intent information comprises:

sorting the probability values in order from big to small;

initial intention information corresponding to the probability values of the preset quantity before the sorting is obtained as target intention information.

5. The method according to claim 1, wherein the query in a preset map database based on the target vehicle information and the target intention information, to obtain the answer information matched with the question information, includes:

inquiring in a preset problem map based on the target vehicle information and the target intention information to obtain target problem information matched with the question information;

inquiring in a preset answer map based on the target question information to obtain answer information matched with the target question information, wherein the preset question map and the preset answer map are stored in the preset map database.

6. The method of claim 5, wherein, when the target intention information is at least two or more target intention information,

inquiring in a preset problem map based on the target vehicle information and the target intention information to obtain target problem information matched with the audio, wherein the method comprises the following steps:

inquiring in a preset problem map based on the target vehicle information and the target intention information aiming at each target intention information to obtain a plurality of problem information matched with the audio;

calculating the similarity of the question information and the question information aiming at each piece of question information;

and determining the problem information corresponding to the maximum similarity as target problem information.

7. The method according to claim 2, wherein the method for constructing the intention recognition model comprises:

acquiring sample questioning information;

performing intention labeling on the sample question information to obtain labeling information corresponding to the sample question information;

training a preset network model by using the sample questioning information and the labeling information to obtain a pre-trained intention recognition model.

8. The method of claim 7, wherein training a pre-set network model using the sample challenge information and the annotation information to obtain a pre-trained intent recognition model comprises:

Converting the sample question information into a sample vector;

carrying out vector extraction and processing on the sample vector by utilizing the preset network model to obtain output intention information;

performing similarity calculation on the output intention information and the labeling information to obtain similarity of the output intention information and the labeling information;

when the similarity of the two is larger than or equal to a preset threshold value, the preset network model is used as a pre-trained intention recognition model;

and when the similarity is smaller than a preset threshold value, continuing training the preset network model.

9. A question-answering processing apparatus, comprising:

10. An electronic device, comprising:

a memory;

a processor; and

a computer program;

wherein the computer program is stored in the memory and configured to be executed by the processor to implement the method of any one of claims 1-8.

11. A vehicle, characterized in that it performs the method according to any one of claims 1-8 or comprises the electronic device according to claim 10.

12. A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of any of claims 1-8.