TWI674517B - Information interaction method and device - Google Patents

Abstract

A method and device for information interaction. The method includes: determining a first target answer parameter; obtaining a target question set from a question library, wherein the target question set includes at least one request parameter, and the at least one request parameter is used for Requesting a user to send one answer parameter corresponding to the at least one request parameter; obtaining at least one target request parameter corresponding to the first target response parameter in the target question set; and transferring the target request parameter Send to a terminal device.

Description

   Method and device for information interaction   

This disclosure relates to the field of Internet technology, and more particularly to a method and device for information interaction.

In the field of service dialogue, the server needs to automatically ask the user some questions before providing some services to the user. The service robot can extract the answer parameters corresponding to the question from the user's answer, and finally provide the user with the user Related Services. For example, when providing a service for ordering air tickets to users, they first need to ask the user questions about the user ’s departure place, destination, and departure time, and extract the departure place, destination, and departure time from the user ’s answers to the questions. Wait for the response parameters, and complete the service of ordering air tickets for users based on these response parameters.

In the conventional technology, a fixed question is set in advance for each answer parameter, and if a certain answer parameter needs to be obtained during a conversation, a question corresponding to the answer parameter is automatically sent to the user. However, pre-defined questions are blunt, have a large gap with natural language, and lack a sense of wisdom.

In view of this, the present disclosure provides a method and device for information interaction to solve the technical problem that the method of generating questions in conventional technology is far from natural language and lacks a sense of wisdom.

To solve the above problems, the technical solution provided by the present disclosure is as follows.

A method for information interaction includes: determining a first target answer parameter, wherein the first target answer parameter is a parameter that does not obtain a request parameter corresponding to the first target answer parameter; obtaining a A target question set, wherein the target question set is a question set corresponding to the first target answer parameter, the target question set includes at least one request parameter, and the at least one request parameter is used to request a user to send The answer parameter corresponding to the at least one request parameter is described; at least one target request parameter corresponding to the first target answer parameter is obtained in the target question set; and the target request parameter is sent to a terminal device.

An information interaction device includes: a determining unit for determining a first target answer parameter, wherein the first target answer parameter is a parameter that does not obtain a request parameter corresponding to the first target answer parameter; a A first obtaining unit is configured to obtain a target question set from a question library, wherein the target question set is a question set corresponding to the first target answer parameter, and the target question set includes at least one request parameter, so The at least one request parameter is used to request a user to send an answer parameter corresponding to the at least one request parameter; and a second acquisition unit is configured to acquire a response corresponding to the first target answer parameter in the target question set. At least one target request parameter; and a sending unit, configured to send the target request parameter to a terminal device.

In the embodiment of the present disclosure, when a first target answer parameter is determined, a target question set is obtained from a question library, and at least one target request parameter corresponding to the first target answer parameter is obtained from the target question set, and The target request parameter is sent to a terminal device. The disclosure can make the target request parameters received by the user as a question more natural, closer to natural language, and improve the wisdom of asking questions.

401‧‧‧ confirm unit

402‧‧‧First Acquisition Unit

403‧‧‧Second Acquisition Unit

404‧‧‧ sending unit

501‧‧‧ processor

502‧‧‧Memory

503‧‧‧ input device

504‧‧‧Output device

505‧‧‧Bus

S101-S104, S201-S208, S301-S304‧‧‧steps

FIG. 1 is a flowchart of a method for information interaction provided in an embodiment of the disclosure.

FIG. 2 is a flowchart of a method for information interaction provided in another embodiment of the disclosure.

FIG. 3 is a flowchart of a method for information interaction provided in another embodiment of the disclosure.

FIG. 4 is a schematic diagram of an information interaction device provided in an embodiment of the disclosure.

FIG. 5 is a schematic diagram of a server provided in an embodiment of the disclosure.

In order to make the foregoing objectives, features, and advantages of the present disclosure more comprehensible, the embodiments of the present disclosure are described in further detail below with reference to the accompanying drawings and specific implementations.

In the field of service dialogues, in order to provide certain services to users, the server needs to know some parameters. In order to know these parameters, it is necessary to send the request parameters (that is, questions) corresponding to these parameters to the user, so that the user can Request parameters to answer, and send the answer information corresponding to the request parameters to the server, then extract the answer parameters corresponding to the question from the answer information sent by the user, and finally provide the user with related services. For example, when a user is provided with a service for ordering a flight ticket, the required response parameters include destination information, departure place information, and departure time information, and the request parameters corresponding to these response parameters need to be sent to the user. In the conventional technology, a fixed questioning method is set in advance for each answer parameter. If a certain answer parameter needs to be obtained during a conversation, a request parameter corresponding to the answer parameter is automatically sent to the user. However, this is relatively stiff for users, lacking the sense of wisdom and humanity. In addition, only one answer parameter is asked at a time, and when multiple answer parameters need to be obtained, the questioning efficiency is low. In order to solve the technical problems in the conventional technology, the embodiments of the present disclosure provide a method and device for information interaction, from a large number of real corpora (a large structured text combination, usually stored and operated in electronic form). You can learn the questioning methods of various answer parameters in training to obtain a question library. The question library includes a set of questions corresponding to various answer parameters. When you ask a question about one or some answer parameters, you obtain the answer from the question library. The request parameter corresponding to the answer parameter is sent to the user to improve the wisdom of the question, make the way of each question different, and conform more to human expression habits.

The embodiment of the present disclosure will be described from the perspective of a device for information interaction, and the device for information interaction may be specifically loaded in a server. Referring to FIG. 1, which illustrates a method for information interaction provided in an embodiment of the present disclosure, including the following steps:

Step S101: Determine a first target answer parameter.

When providing services to users, some request parameters need to be sent to the users to obtain the required response parameters. For example, when providing a user with a ticket booking service, answer parameters such as destination information, departure place information, and departure time information need to be obtained by asking the user questions. In practical applications, a first target answer parameter may be determined according to a required answer parameter and a second target answer parameter, and the first target answer parameter is a parameter for which no request parameter corresponding to the first target answer parameter is obtained. .

If it is the first time asking a user a question, the required answer parameters are all the first target answer parameters. If the user has already answered, the answer parameters answered by the user are recorded as the second target answer parameters. The second target answer parameter is a parameter that has obtained a request parameter corresponding to the second target answer parameter.

The first target answer parameter is determined according to the required answer parameter and the second target answer parameter. There can be one or more of the first target answer parameters.

Step S102: Obtain a target question set from the question base.

The target question set shown in this embodiment is a question set corresponding to the first target answer parameter.

In some possible implementation manners, the method for information interaction provided in the embodiments of the present disclosure may further include: training and generating the question database according to the input real corpus data, the question database includes answer parameters, and the answer parameters may be Answer parameters for the first target and / or answer parameters for the second target. That is, the question library includes a set of questions corresponding to various answer parameters.

For example, the question library includes a question set corresponding to the answer parameter A, a question set corresponding to the answer parameter B, a question set corresponding to the answer parameter C, a question set corresponding to the answer parameters A and B, and answer parameters A and B. And the problem set corresponding to C and so on. Each question set includes multiple questions, for example, the answer parameter A is the destination information, and the question set corresponding to the answer parameter A may include "where are you going?", "Where are you going?", "Where is your destination?" And so on. For example, the answer parameter B is the starting point information. The question set corresponding to the answer parameters A and B may include questions such as "Where do you go from here?", "Where is your starting point and destination?" That is to say, the question library includes multiple question sets, and a question set may be a question set corresponding to one answer parameter, or a question set corresponding to multiple answer parameters. How to train and generate a question database based on the input real corpus data will be explained in detail in the subsequent embodiments.

In this embodiment, when there is only one first target answer parameter, a target question set corresponding to the first target answer parameter may be directly obtained from the question library.

In some possible implementation manners, when there are n first target answer parameters, the step of obtaining the target question set from the question library includes: obtaining the target question set from the question library, the target question set and There are m corresponding first target answer parameters, where n is a positive integer greater than or equal to 2 and m is a positive integer less than or equal to n.

For example, suppose n is 3, which means that there are 3 first target answer parameters, namely answer parameter A, answer parameter B, and answer parameter C. Then the question library has a question set corresponding to one answer parameter, that is, the answer parameter. A question set corresponding to A, a question set corresponding to the answer parameter B, and a question set corresponding to the answer parameter C.

There are also question sets corresponding to two answer parameters in the question library, that is, question sets corresponding to answer parameters A and B, question sets corresponding to answer parameters B and C, and question sets corresponding to answer parameters A and C. At the same time, the question library also has a question set corresponding to the three answer parameters, that is, a question set corresponding to the answer parameters A, B, and C. Then, one of these question sets is obtained as the first target answer parameter. Question collection. That is, when there are multiple first target answer parameters, a question may be asked for one of the answer parameters, or a question may be asked for a plurality of the answer parameters.

Step S103: Acquire at least one target request parameter corresponding to the first target answer parameter in the target question set.

In some possible implementation manners, the specific implementation of obtaining at least one target request parameter corresponding to the first target answer parameter in the target question set may include: randomly selecting from the question set corresponding to the first target answer parameter A request parameter is selected as the target request parameter.

In an implementation manner, a request parameter may be randomly selected as a target request parameter from a target problem set corresponding to the first target answer parameter obtained in the foregoing steps, and the target request parameter is selected from the target problem set and the target problem. The set includes various request parameters that may appear in the real context that correspond to the first target answer parameter, and the frequency of the various request parameters in the target question set is basically the same as in the real context, so the selected target request The parameters are also closer to the real context.

In some possible implementation manners, the specific implementation of obtaining at least one target request parameter corresponding to the first target answer parameter in the target question set may also include: inputting the target request parameter to a neural network (neural network) network) model, and obtain the target request parameters output by the neural network model.

In another implementation, when there are enough request parameters in the question set, the neural network model can be trained to generate a target request parameter corresponding to the first target answer parameter, so that the way of asking questions will be richer, not only It is limited to the existing questions in the question bank, but generates questions more in line with human habits.

Step S104: Send the target request parameter to the terminal device.

Send the target request parameters obtained in step S103 to the terminal device, so that the user using the terminal device can answer the first target answer parameter.

In the embodiment of the present disclosure, when a question is asked about the first target answer parameter, a target question set corresponding to the first target answer parameter is obtained from the question library, and the target question set includes a plurality of ANDs obtained from real corpus data. The corresponding target request parameter of the first target answer parameter is to obtain a target question set corresponding to the first target answer parameter from the question library, and obtain a target request parameter corresponding to the first target answer parameter from the target question set to send. To the user, the target request parameters received by the user can be more natural and close to natural language, which improves the wisdom of asking questions.

Referring to FIG. 2, which illustrates a method for information interaction provided in another embodiment of the present disclosure, including the following steps:

Step S201: Determine a first target answer parameter.

Step S202: Obtain a target question set from the question base.

Step S203: Acquire at least one target request parameter corresponding to the first target answer parameter in the target question set.

Step S204: Send the target request parameter to the terminal device.

Steps S201-S204 are the same as steps S101-S104. For related description, refer to the foregoing embodiments, and details are not described herein again.

Step S205: Receive the response information sent by the user.

After sending the target request parameter to the terminal device, so that the user using the terminal device obtains the target request parameter, the user will answer the target request parameter to generate answer information, and after the user answers, the user can receive The answer information sent by the terminal device includes a second target answer parameter, and the second target answer parameter is a parameter that has obtained a request parameter corresponding to the second target answer parameter.

Step S206: Extract a second target answer parameter from the answer information.

After receiving the answer information, the second target answer parameter is extracted. It can be understood that if multiple questions and answers have been performed, the second target answer parameter is not limited to the answer parameter extracted in this answer.

Step S207: Determine whether the first target answer parameter exists according to the second target answer parameter and the required answer parameter, and if it exists, return to step S201; if it does not exist, proceed to step S208.

According to the second target answer parameter and the required answer parameter, it can be determined whether the first target answer parameter exists. For example, the required answer parameter includes destination information, departure place information, and departure time. The second target answer parameter includes the purpose. Location information and departure location information, there is still a first target answer parameter that is not determined (departure time), and it is necessary to return to step S201 to re-determine the first target answer parameter. Questions need to be asked again during departure time.

Step S208: Provide a corresponding service according to the second target answer parameter.

If all required second-target answer parameters have been obtained, the user can be provided with corresponding services. For example, after obtaining the answer information such as destination information, departure location information, and departure time, the user can provide the user with the ticket booking order. service.

In the embodiment of the present disclosure, by asking questions closer to the natural language, the user can communicate more smoothly and naturally with the machine during the conversation, so that the user can answer the required answer parameters more quickly. According to these second goals, The answer parameter can be completed to provide the corresponding service to the user.

Referring to FIG. 3, which illustrates a method for information interaction provided in another embodiment of the present disclosure. In this embodiment, the specific implementation of generating the question database according to the input real corpus data may include the following: Steps:

Step S301: Receive real corpus data.

The real corpus data includes the corpus data and annotated data on the corpus data. The annotated data includes the type of the corpus data and the response parameters corresponding to the corpus data.

In the process of training to generate the problem database, a large amount of real corpus data is needed. The corpus data in the real corpus data is extracted from the real conversation between the artificial customer service and the user. Annotated data can be obtained after annotating the corpus manually, or after annotating the corpus through semi-supervised machine learning. The annotated corpus data needs to be annotated one by one. The annotated data includes the type of corpus data. For example, the corpus data is a question, or an answer, or other (such as unrelated to the question and answer Greetings, chats, etc.). The annotation data also includes the answer parameters corresponding to the corpus data. When the corpus data is a question corpus, the answer parameters corresponding to the corpus data represent the answer parameters corresponding to the question corpus. For example, the corpus data is " "Where are you going?", The annotated data of this corpus data is "problem, destination information", which means that "Where are you going?" This corpus data is a question corpus corresponding to the destination information. When the corpus data is the answer corpus, the answer parameters corresponding to the corpus data represent the answer parameters included in the answer corpus.

In practical applications, each piece of real corpus data can be encapsulated into a structure, and the real corpus data is received one by one, and the subsequent steps are processed.

Step S302: Determine whether the corpus data is a question corpus according to the type of the corpus data. If yes, go to step S303; if no, go to step S304.

In the process of training to generate a question base, the emphasis is on the questioning methods that require various answer parameters. The question corpus is linked with the relevant answer parameters. Therefore, the question corpus needs to be obtained from the corpus data. In this embodiment, The type of the corpus data in the annotation data is read to determine whether the corpus data is a problem corpus.

Step S303: Obtain answer parameters corresponding to the corpus data, and store the question corpus in a question set corresponding to the corpus data.

When the corpus data is a question corpus, the answer parameters corresponding to the corpus data can be obtained, that is, the answer parameters corresponding to the question corpus are obtained, and the question corpus is stored to the corresponding corpus data. Answer parameter in question set. When the answer parameter corresponding to the question corpus is one, the question corpus is stored in the question set corresponding to the answer parameter, for example, the question corpus of "where are you going" is stored to In the question set corresponding to "Destination Information". When the question corpus has multiple corresponding answer parameters, the question corpus is stored in the question set corresponding to the answer parameters, for example, the question phrase "where are you going from?" The data is stored in the question set corresponding to the "departure point information and destination point information".

In addition, even if the question corpus appears repeatedly, the question corpus is still stored in the corresponding question set, so that the frequency of occurrence of each question in the question set is basically consistent with the real context, for example, when question destination information is asked 80% of the questions are "Where are you going?", Then "Where are you going?" The question corpus also accounts for 80% of the question set corresponding to the destination information, and it is related to the destination When selecting a question from the question set corresponding to the information, the probability of selecting the question "Where are you going?" Is also 80%, making the selected question closer to the real context.

Step S304: Determine whether the corpus data is a response corpus, and when the corpus data is a response corpus, store the response corpus.

When the corpus data is not a question corpus, the corpus data may be the answer corpus, or it may be another corpus, you can judge whether the corpus data is the answer corpus by the type of the corpus data. The data is the answer corpus, and the answer corpus is stored. In addition, it should be noted that step S304 does not belong to the process of generating the question base. When determining that the corpus data is not a question corpus according to the type of the corpus data, no processing may be performed.

After receiving the input corpus data and the annotation data of the corpus data, the embodiment of the present disclosure can automatically extract the question corpus and the corresponding answer parameters, and establish a question library including a question set with the answer parameters, so that The chat service makes questions more human and intelligent.

Referring to FIG. 4, which shows an information interaction device provided in an embodiment of the present disclosure, including: a determining unit 401, a first obtaining unit 402, a second obtaining unit 403, and a sending unit 404 .

The determining unit 401 is configured to determine a first target answer parameter, where the first target answer parameter is a parameter for which no request parameter corresponding to the first target answer parameter is obtained.

The first obtaining unit 402 is configured to obtain a target question set from a question database. The target question set is a question set corresponding to the first target answer parameter. The target question set includes at least one answer parameter, and all The target question set includes at least one request parameter, and the request parameter is used to request a user to send an answer parameter corresponding to the request parameter.

The second obtaining unit 403 is configured to obtain at least one target request parameter corresponding to the first target answer parameter in the target question set.

The sending unit 404 is configured to send the target request parameter to a terminal device.

In some possible implementation manners of the present disclosure, when there are n first target answer parameters, the first obtaining unit 402 may be specifically configured to: obtain the target question set from the question library, where The target question set corresponds to m of the first target answer parameters, where n is a positive integer greater than or equal to 2 and m is a positive integer less than or equal to n.

In some possible implementation manners of the present disclosure, the second obtaining unit 403 may be specifically configured to randomly select a request parameter from the target problem set as the target request parameter.

In some possible implementation manners of the present disclosure, the second obtaining unit 403 may be further specifically configured to input the target request parameter into a neural network model.

The sending unit 404 is further specifically configured to send the target request parameter output by the neural network model to the terminal device.

In some possible implementations of the present disclosure, the information interaction device provided in the embodiments of the present disclosure may further include a receiving unit, a recording unit, and a determining unit.

The receiving unit is configured to receive the answer information sent by a user.

The recording unit is configured to extract a second target answer parameter from the answer information, and the second target answer parameter is a parameter that has obtained a request parameter corresponding to the second target answer parameter.

The judging unit judges whether the first target answer parameter exists according to the second target answer parameter and the required answer parameter, and if the first target answer parameter exists, the determining unit obtains from the question database For a target question set, if the first target answer parameter does not exist, a corresponding service is provided according to the second target answer parameter.

In some possible implementation manners of the present disclosure, the apparatus for information interaction provided in the embodiments of the present disclosure may further include: a training unit for training to generate the question database according to the input real corpus data, the real corpus The data includes corpus data and annotated data on the real corpus data, and the annotated data includes the type of the real corpus data and answer parameters corresponding to the real corpus data.

In some possible implementations of the present disclosure, the training unit may include: a receiving subunit, a judging subunit, and a storing subunit.

The receiving subunit is configured to receive real corpus data.

The judging subunit is configured to judge whether the corpus data is a problem corpus according to the type of the corpus data.

The storage subunit is configured to obtain answer parameters corresponding to the corpus data if the corpus data is a question corpus, and store the question corpus into a question set corresponding to the corpus data. in.

In the embodiment of the present disclosure, when a question is asked about an unobtained answer parameter, a target question set corresponding to the unobtained answer parameter is obtained from the question library, and the target question set includes multiple and obtained from real corpus data. The corresponding request parameters of the unobtained answer parameters, according to the target question set in the question library and the unobtained answer parameters, obtain a target request parameter with the unobtained answer parameters and send it to the user, so that the user can receive the Questions are more natural and closer to natural language, increasing the wisdom of asking questions.

Accordingly, the embodiment of the present disclosure further provides a server, as shown in FIG. 5, including a processor 501, a memory 502, an input device 503, and an output device 504.

The number of processors 501 in the server may be one or more, and one processor is taken as an example in FIG. 5. In some embodiments of the present disclosure, the processor 501, the memory 502, the input device 503, and the output device 504 may be connected by using a bus 505 or other methods. In FIG.

The memory 502 can be used to store software programs and modules. The processor 501 executes various functional applications and data processing of the server by running software programs and modules stored in the memory 502. The memory 502 may mainly include a storage program area and a storage data area. The storage program area may store an operating system, at least one application required by a function, and the like. In addition, the memory 502 may include high-speed random access memory (Random Access Memory, RAM), and may also include non-volatile memory, such as at least one magnetic disk memory device, flash memory device, or other volatile solid state memory. Memory device. The input device 503 can be used to receive inputted digital or character information, and generate key signal inputs related to user settings and function control of the server.

Specifically, in this embodiment, the processor 501 loads the executable files corresponding to one or more application processes into the memory 502 according to the following instructions, and the processor 501 runs and stores the executable files in the memory 502. An application in the memory 502 to implement various functions: determining a first target answer parameter, the first target answer parameter being a parameter for which no request parameter corresponding to the first target answer parameter has been obtained; obtained from the question base A target question set, the target question set being a question set corresponding to the first target answer parameter, the target question set including at least one answer parameter, and the target question set including at least one request parameter, the request The parameter is used to request the user to send an answer parameter corresponding to the request parameter; obtain at least one target request parameter corresponding to the first target answer parameter in the target question set; and send the target request parameter to Terminal Equipment.

Correspondingly, when there are n first target answer parameters, the obtaining the target question set from the question library includes: obtaining the target question set from the question library, and the target question set is related to m The first target answer parameter corresponds, where n is a positive integer greater than or equal to 2 and m is a positive integer less than or equal to n.

Accordingly, the obtaining at least one target request parameter corresponding to the first target answer parameter in the target question set includes: randomly selecting a request parameter from the target question set as the target request parameter.

Accordingly, the obtaining at least one target request parameter corresponding to the first target answer parameter in the target question set includes: inputting the target request parameter into a neural network model.

Accordingly, the sending the target request parameter to a terminal device includes: sending the target request parameter output by the neural network model to the terminal device.

Correspondingly, the instruction further includes: receiving the answer information sent by the user; extracting a second target answer parameter from the answer information, the second target answer parameter being obtained and the second target answer parameter Parameters of corresponding request parameters; determining whether the first target answer parameter exists according to the second target answer parameter and the required answer parameter; if the first target answer parameter exists, obtaining a target from the question base A set of questions; and if the first target answer parameter does not exist, a corresponding service is provided according to the second target answer parameter.

Correspondingly, the instruction further includes: generating the question database according to the input real corpus data training, the real corpus data including the corpus data and annotated data on the real corpus data, the annotated data includes The type of the real corpus material and the response parameters corresponding to the real corpus material.

Accordingly, the training to generate the question database based on the input real corpus data includes: receiving real corpus data; judging whether the corpus data is a question corpus according to the type of the corpus data; if the language If the data is a question corpus, the answer parameters corresponding to the corpus are obtained; and the question corpus is stored in a question set corresponding to the corpus.

In the embodiment of the present disclosure, when a question is asked about the first target answer parameter, a target question set corresponding to the first target answer parameter is obtained from the question library, and the target question set includes multiple and obtained from real corpus data. According to the target request parameter corresponding to the first target answer parameter, according to the target question set in the question library and the first target answer parameter, obtain a target request parameter corresponding to the first target answer parameter and send it to the user. The target request parameters are more natural and close to natural language, which improves the wisdom of asking questions.

It should be noted that each embodiment in this specification is described in a progressive manner. Each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments may refer to each other. For the system or device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant part may refer to the description of the method.

It should also be noted that in this disclosure, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that these entities or operations There is any such actual relationship or order between operations. Moreover, the terms "including", "comprising", or any other variation thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or device that includes a series of elements includes not only those elements but also those that are not explicitly listed Or other elements inherent to such a process, method, article, or device. Without more restrictions, the elements defined by the sentence "including a ..." do not exclude the existence of other identical elements in the process, method, article, or equipment including the elements.

The steps of the method or algorithm described in connection with the embodiments disclosed in this disclosure may be implemented directly by hardware or a software module executed by a processor, or a combination of the two. Software modules can be placed in Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), electrical Erase Electrically Erasable Programmable Read-Only Memory (EEPROM), registers, hard drives, removable disks, Compact Disc Read-Only Memory (CD-ROM), or this disclosure In any other form of storage medium known in the art.

Although the present disclosure has been disclosed as above with a preferred embodiment, it is not intended to limit the present disclosure. Those with ordinary knowledge in the technical field to which this disclosure belongs can make various changes and modifications without departing from the spirit and scope of this disclosure. Therefore, the scope of protection of this disclosure shall be determined by the scope of the appended patent application.

Claims (12)

A method for information interaction includes: determining a first target answer parameter, wherein the first target answer parameter does not have a request parameter, the request parameter is used to request a user to send an answer parameter; from a question database Obtaining a target question set, wherein the target question set is a question set corresponding to the first target answer parameter, the target question set includes at least one request parameter; and obtaining the target question set from the first question set At least one target request parameter corresponding to the target answer parameter; and sending the target request parameter to a terminal device, wherein there are n first target answer parameters, the target question set and m first target answers The parameters correspond to that n is a positive integer greater than or equal to 2 and m is a positive integer less than or equal to n. The step of obtaining the target question set from the question library includes: obtaining n from the question library and n A set of target questions corresponding to the first target answer parameters. The method of claim 1, wherein the step of obtaining at least one target request parameter corresponding to the first target response parameter in the target question set includes: randomly selecting from the target question set A request parameter is selected as the target request parameter. The method of claim 1, wherein the step of obtaining at least one target request parameter corresponding to the first target answer parameter in the target question set includes: entering the target request parameter into a In a neural network model, the step of sending the target request parameter to the terminal device includes: sending the target request parameter output by the neural network model to the terminal device. The method according to item 1 of the scope of patent application, further comprising: receiving the answer information sent by the user; extracting a second target answer parameter from the answer information, wherein the second target answer The parameter is a parameter that has obtained one of the request parameters corresponding to the second target answer parameter; judging whether the first target answer parameter exists according to the second target answer parameter and the required answer parameter; if the first target answer parameter exists; A first target answer parameter, returning to performing the step of obtaining the target question set from the question library; and if the first target answer parameter does not exist, providing a corresponding one according to the second target answer parameter service. The method according to item 1 of the scope of patent application, further comprising: generating the question database according to the inputted real corpus data training, wherein the real corpus data includes a corpus data and the real corpus data. A batch of annotation data, the annotation data includes the type of the real corpus data and the answer parameters corresponding to the real corpus data. The method according to item 5 of the scope of patent application, wherein the step of generating the question database according to the input real corpus data includes: receiving real corpus data; judging the language according to the type of the corpus data Whether the corpus data is a question corpus; if the corpus data is a question corpus, obtaining answer parameters corresponding to the corpus data; and storing the question corpus to a corresponding corpus data Question collection. An information interaction device includes a determining unit for determining a first target response parameter, wherein the first target response parameter does not have a request parameter, and the request parameter is used to request a user to send a response parameter A first acquisition unit for acquiring a target question set from a question library, wherein the target question set is a question set corresponding to the first target answer parameter, and the target question set includes at least one request parameter A second obtaining unit, configured to obtain at least one target request parameter corresponding to the first target answer parameter in the target question set; and A sending unit, configured to send the target request parameter to a terminal device, wherein there are n first target response parameters, the target question set corresponds to m first target response parameters, and n is greater than Or a positive integer equal to 2, and m is a positive integer less than or equal to n, and the first obtaining unit is specifically configured to obtain a target question set corresponding to n first target answer parameters from the question database. The device according to item 7 of the scope of patent application, wherein the second obtaining unit is specifically configured to randomly select a request parameter from the target problem set as the target request parameter. The device according to item 7 of the scope of patent application, wherein the second obtaining unit is specifically configured to: input the target request parameter into a neural network model; and the sending unit is specifically configured to: input the neural network The target request parameter output by the model is sent to the terminal device. The device according to item 7 of the scope of patent application, further comprising: a receiving unit for receiving the answer information sent by the user; a recording unit for extracting a second target from the answer information Answer parameter, wherein the second target answer parameter is a parameter that has obtained one of the request parameters corresponding to the second target answer parameter; a judgment unit determines according to the second target answer parameter and the required answer parameter Whether the first target answer parameter exists, and if the first target answer parameter exists, the determining unit obtains the number of target question sets from the question library; if the first target answer parameter does not exist, The second target answer parameter provides a corresponding service. The device according to item 7 of the scope of the patent application, further comprising: a training unit for generating the question database according to the input real corpus data, wherein the real corpus data includes a corpus data and the Capital injection The annotation data includes the type of the real corpus data and the answer parameters corresponding to the real corpus data. The device according to item 11 of the scope of patent application, wherein the training unit includes: a receiving sub-unit for receiving real corpus data; and a judging sub-unit for judging the type according to the type of the corpus data Whether the corpus data is a question corpus; and a storage sub-unit for obtaining the answer parameters corresponding to the corpus data if the corpus data is a question corpus, and storing the question corpus to and In the question set corresponding to the corpus data.