WO2022156086A1

WO2022156086A1 - Human computer interaction method, apparatus and device, and storage medium

Info

Publication number: WO2022156086A1
Application number: PCT/CN2021/090424
Authority: WO
Inventors: 方勇
Original assignee: 平安科技（深圳）有限公司
Priority date: 2021-01-21
Filing date: 2021-04-28
Publication date: 2022-07-28
Also published as: CN112860850A; CN112860850B

Abstract

A human computer interaction method, apparatus and device, and a storage medium, which relate to the field of artificial intelligence and are used for improving human computer conversation efficiency of a financial service system. The human computer interaction method comprises: classifying financial corpora by using a K-value clustering method, obtaining multiple groups of classified corpora; sorting each group of classified corpora according to a preset flow sequence, and creating multiple groups of financial service structure trees according to the multiple groups of sorted corpora; matching a corresponding financial service structure tree by means of inputted question data, determining the corresponding financial service structure tree as a target financial structure tree, and searching the target financial structure tree for a target matching answer corresponding to the inputted question data; skipping from the target financial structure tree to a predicted financial structure tree by means of skipping signal data, and returning an operation interface of the predicted financial structure tree; querying the predicted financial structure tree for a predicted matching answer corresponding to inputted operation data, and returning the predicted matching answer. The present method further relates to a blockchain technique, and financial corpora can be stored in a blockchain.

Description

Human-computer interaction method, device, device and storage medium

This application claims the priority of the Chinese patent application filed on January 21, 2021 with the application number 202110082849.8 and the invention titled "Human-Computer Interaction Method, Apparatus, Equipment and Storage Medium", the entire contents of which are incorporated by reference in application.

technical field

The present application relates to the field of artificial intelligence, and in particular, to a human-computer interaction method, apparatus, device and storage medium.

Background technique

With the rapid development of electronic technology and the popularization of artificial intelligence, human-computer interaction systems are cited in some industries. Human-computer interaction (HCI) is a study of the interaction between systems and users. The corresponding human-computer interaction systems can be various machines or computerized systems and software. , wherein the system includes a human-computer interface for users to communicate or operate with the system. In the financial business scenario, the existing human-computer interaction dialogue system is usually established based on the knowledge base, which can match the appropriate answers in the database according to the customer's intention.

However, the inventor realizes that the existing human-computer interaction dialogue system can only match the corresponding operation results according to the operation information provided by the user, resulting in low efficiency of the human-computer dialogue in the financial business system.

SUMMARY OF THE INVENTION

The present application provides a human-computer interaction method, device, device and storage medium, which are used to improve the efficiency of human-computer dialogue in a financial business system.

A first aspect of the present application provides a human-computer interaction method, including: acquiring financial corpus in a financial business scenario, and classifying the financial corpus by using a K-value clustering method to obtain multiple groups of classified corpora, the financial corpus At least include signing the financial information corpus, supplementing the financial information corpus, and entering the financial contact corpus; sorting each group of classification corpora according to the preset process sequence, obtaining multiple sets of sorting corpus, and creating multiple sets of financial services according to the multiple sets of sorting corpus Structure tree; obtain the input problem data during human-computer interaction, match the corresponding financial business structure tree through the input problem data, and determine the corresponding financial business structure tree as the target financial structure tree, in the target financial structure tree Find the target matching answer corresponding to the input question data in , and return the target matching answer; obtain the jump signal data during human-computer interaction, and filter the multiple groups of financial business structure trees through the jump signal data. Predict the financial structure tree, and jump from the target financial structure tree to the predicted financial structure tree, and return to the operation interface of the predicted financial structure tree; obtain the input operation data when the human-computer interaction is performed on the operation interface, in the query the predicted matching answer corresponding to the input operation data in the predicted financial structure tree, and return the predicted matching answer.

A second aspect of the present application provides a human-computer interaction device, comprising a memory, a processor, and computer-readable instructions stored on the memory and executable on the processor, and the processor executes the computer-readable instructions When reading the instructions, the following steps are implemented: obtaining the financial corpus under the financial business scenario, and classifying the financial corpus using the K-value clustering method to obtain multiple groups of classified corpora, where the financial corpus at least includes signing the financial information corpus, supplementing the financial corpus. Information corpus and input financial contact corpus; sort each group of classified corpus according to the preset process sequence to obtain multiple sets of sorted corpus, and create multiple sets of financial business structure trees according to the multiple sets of sorted corpus; Input problem data, match the corresponding financial business structure tree through the input problem data, determine the corresponding financial business structure tree as the target financial structure tree, and find the corresponding financial business structure tree in the target financial structure tree. The target matching answer is returned, and the target matching answer is returned; the jump signal data during human-computer interaction is obtained, and the predicted financial structure tree is screened in the multiple groups of financial business structure trees through the jump signal data, and is obtained from the The target financial structure tree jumps to the forecast financial structure tree, and returns to the operation interface of the forecast financial structure tree; obtains the input operation data when the human-computer interaction is performed on the operation interface, and queries the forecast financial structure tree in the forecast financial structure tree. The predicted matching answer corresponding to the input operation data is returned, and the predicted matching answer is returned.

A third aspect of the present application provides a computer-readable storage medium, where computer instructions are stored in the computer-readable storage medium, and when the computer instructions are executed on a computer, the computer is caused to perform the following steps: obtaining a financial business scenario and classify the financial corpus using the K-value clustering method to obtain multiple groups of classified corpora, the financial corpus at least includes the corpus of signing financial information, supplementing the corpus of financial information and entering the corpus of financial contacts; according to the preset The process sequence sorts each group of classified corpus, obtains multiple sets of sorted corpus, and creates multiple sets of financial business structure trees according to the multiple sets of sorted corpus; obtains input question data during human-computer interaction, and matches the corresponding input question data through the input question data. The financial business structure tree, the corresponding financial business structure tree is determined as the target financial structure tree, the target matching answer corresponding to the input question data is searched in the target financial structure tree, and the target matching answer is returned; Obtain the jump signal data during human-computer interaction, screen the predicted financial structure tree in the multiple groups of financial business structure trees through the jump signal data, and jump from the target financial structure tree to the predicted financial structure tree, return the operation interface of the forecast financial structure tree; obtain the input operation data when the human-computer interaction is performed on the operation interface, query the forecast matching answer corresponding to the input operation data in the forecast financial structure tree, and return all The described prediction matches the answer.

A fourth aspect of the present application provides a human-computer interaction device, comprising: a classification module for acquiring financial corpus in a financial business scenario, and classifying the financial corpus by using K-value clustering method to obtain multiple groups of classified corpora , the financial corpus at least includes the signed financial information corpus, the supplementary financial information corpus, and the input financial contact corpus; the sorting module is used to sort each group of classified corpora according to the preset process sequence, and obtain multiple groups of sorted corpora, and according to the The multiple sets of sorting corpus create multiple sets of financial business structure trees; a determination module is used to obtain input problem data during human-computer interaction, and the corresponding financial business structure tree is matched by the input problem data, and the corresponding financial business structure The tree is determined as the target financial structure tree, the target matching answer corresponding to the input question data is searched in the target financial structure tree, and the target matching answer is returned; the jump module is used to obtain the jump during human-computer interaction. Signal data, screen the forecast financial structure tree in the multiple groups of financial business structure trees through the jump signal data, and jump from the target financial structure tree to the forecast financial structure tree, and return to the forecast financial structure The operation interface of the tree; the return module is used to obtain the input operation data when the human-computer interaction is performed on the operation interface, query the prediction matching answer corresponding to the input operation data in the prediction financial structure tree, and return the prediction matching Answer.

In the technical solution provided by the present application, the financial corpus under the financial business scenario is obtained, and the K-value clustering method is used to classify the financial corpus to obtain multiple groups of classified corpora, and the financial corpus at least includes the signed financial information corpus, supplementary Financial information corpus and input financial contact corpus; sort each group of classified corpus according to the preset process sequence to obtain multiple sets of sorted corpus, and create multiple sets of financial business structure trees according to the multiple sets of sorted corpus; The input problem data, match the corresponding financial business structure tree by the input problem data, determine the corresponding financial business structure tree as the target financial structure tree, and find the input problem data in the target financial structure tree corresponding to The target matching answer is returned, and the target matching answer is returned; the jump signal data during human-computer interaction is obtained, and the predicted financial structure tree is screened in the multiple groups of financial business structure trees through the jump signal data, and from all the The target financial structure tree jumps to the predicted financial structure tree, and returns to the operation interface of the predicted financial structure tree; obtains the input operation data when the human-computer interaction is performed on the operation interface, and inquires about all the information in the predicted financial structure tree. The predicted matching answer corresponding to the input operation data is returned, and the predicted matching answer is returned. In the embodiment of the present application, by classifying and sorting the financial corpus in the financial business scenario, and inputting the classified and sorted financial corpus into the structure tree builder, the financial business structure tree is constructed. When performing human-computer interaction, The query can be performed by jumping from the financial structure tree currently being queried to the financial structure tree of the next process by jumping the signal data, which improves the man-machine dialogue efficiency of the financial business system.

Description of drawings

FIG. 1 is a schematic diagram of an embodiment of a human-computer interaction method in an embodiment of the present application;

FIG. 2 is a schematic diagram of another embodiment of the human-computer interaction method in the embodiment of the present application;

FIG. 3 is a schematic diagram of an embodiment of a human-computer interaction device in an embodiment of the present application;

FIG. 4 is a schematic diagram of another embodiment of the human-computer interaction device in the embodiment of the present application;

FIG. 5 is a schematic diagram of an embodiment of a human-computer interaction device in an embodiment of the present application.

Detailed ways

Embodiments of the present application provide a human-computer interaction method, apparatus, device, and storage medium, which are used to improve the efficiency of human-computer dialogue in a financial business system.

The terms "first", "second", "third", "fourth", etc. (if any) in the description and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It is to be understood that data so used may be interchanged under appropriate circumstances so that the embodiments described herein can be practiced in sequences other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those expressly listed steps or units, but may include other steps or units not expressly listed or inherent to these processes, methods, products or devices.

For ease of understanding, the following describes the specific process of the embodiment of the present application, referring to FIG. 1 , an embodiment of the human-computer interaction method in the embodiment of the present application includes:

101. Obtain the financial corpus under the financial business scenario, and use the K-value clustering method to classify the financial corpus to obtain multiple groups of classified corpus, and the financial corpus at least includes the signed financial information corpus, the supplementary financial information corpus, and the input financial contact corpus;

It can be understood that the execution subject of the present application may be a human-computer interaction device, or may be a terminal or a server, which is not specifically limited here. The embodiments of the present application take the server as an execution subject as an example for description.

It should be noted that the financial business scenario here refers to the human-computer interaction scenario in the financial field, such as the human-computer interaction scenario of the financial lending business, in which users can handle financial lending projects through human-computer interaction. The financial corpus in the financial business scenario here is the commonly used sentences in the corresponding financial scenario. Taking the financial lending business scenario as an example, the corresponding financial corpus at least includes: loan terms corpus, credit authorization letter corpus, user basic information corpus, contact Human information corpus and other corpora.

After obtaining the financial corpus in the financial business scenario, the server needs to classify and process multiple financial corpora, and classify the corpus with the same category together, and then use the corpus with the same category to build the corresponding financial business structure tree.

It should be emphasized that, in order to further ensure the privacy and security of the above-mentioned financial corpus, the above-mentioned financial corpus can also be stored in a node of a blockchain.

102. Sort each group of classification corpus according to the preset process sequence to obtain multiple sets of sorting corpus, and create multiple sets of financial business structure trees according to the multiple sets of sorting corpus;

Before establishing the financial business structure tree, it is necessary to sort each group of classified corpus. For example, in the financial business structure tree for signing the credit authorization letter, the credit authorization letter needs to be displayed first, and then the page that needs to be signed by the user is displayed as a The process needs to have a corresponding display order, so it is necessary to sort the financial corpus in each group of classified corpus according to the preset process order, and then construct the financial business structure tree after sorting.

103. Obtain the input problem data during human-computer interaction, match the corresponding financial business structure tree through the input problem data, determine the corresponding financial business structure tree as the target financial structure tree, and search for the corresponding financial business structure tree in the target financial structure tree. target matching answer, and return target matching answer;

When the server obtains the input problem data collected by the user through the financial business system during the human-computer interaction, it directly matches the corresponding financial business structure tree according to the input problem data. Here, the corresponding matching method is used. Taking the user signing the comprehensive authorization letter in the financial business system as an example, after the user enters "Apply Now" in the financial business system, it will directly match the financial business structure tree corresponding to the signing credit authorization letter in the financial business structure tree, and According to the logical order in the financial business structure tree, return to the operation interface in the financial business structure tree in turn. The operation interface here is the operation interface related to signing the comprehensive power of attorney. For example, the comprehensive power of attorney interface that has been set in advance is displayed for the user to know. , when the user browses to the last page of the final comprehensive authorization form interface, a signing interface that requires the user to sign the user's name is displayed.

104. Acquire the jump signal data during human-computer interaction, filter the forecast financial structure tree among multiple sets of financial business structure trees through the jump signal data, and jump from the target financial structure tree to the forecast financial structure tree, and return to the forecast financial structure tree interface;

It should be noted that there is not only a logical order in the financial business structure tree, but also a certain logical order between the established financial business structure trees. For example, after completing the signing of the financial business structure tree corresponding to the comprehensive power of attorney, it is necessary to enter the contact. In the human-computer interaction scenario of the financial business structure tree corresponding to people, at this time, it is necessary to jump between the financial business structure trees.

The jump signal data here refers to the signal transmitted by the financial business system to the server after the target matching answer is returned. Through the jump signal, the server can determine whether the structure tree needs to be jumped.

105. Acquire the input operation data when the human-computer interaction is performed on the operation interface, query the prediction matching answer corresponding to the input operation data in the prediction financial structure tree, and return the prediction matching answer.

The server obtains the input operation data when the user performs human-computer interaction on the operation interface of the forecast financial structure tree, queries the forecast matching answer corresponding to the input operation data in the corresponding forecast financial structure tree, and returns the forecast matching answer. It should be noted that the operation process here is the same as the operation process in the above-mentioned step 103, but the specific query content is also different due to the difference of the forecast financial structure tree to be queried.

In the embodiment of the present application, by classifying and sorting the financial corpus in the financial business scenario, and inputting the classified and sorted financial corpus into the structure tree builder, the financial business structure tree is constructed. When performing human-computer interaction, The query can be performed by jumping from the financial structure tree currently being queried to the financial structure tree of the next process by jumping the signal data, which improves the man-machine dialogue efficiency of the financial business system.

Referring to FIG. 2, another embodiment of the human-computer interaction method in the embodiment of the present application includes:

201. Obtain the financial corpus under the financial business scenario, and use the K-value clustering method to classify the financial corpus to obtain multiple groups of classified corpora, where the financial corpus at least includes the signed financial information corpus, the supplementary financial information corpus, and the input financial contact corpus;

Specifically, the server first obtains the financial corpus in the financial business scenario, and the financial corpus at least includes the corpus of signing financial information, supplementing the corpus of financial information, and entering the corpus of financial contacts; secondly, the server selects n financial corpora as the initial corpus, where n∈{ 2,3...,k-1}, k is the number of financial corpus; then the server calculates the Euclidean distance data between the keywords of the remaining corpus and the keywords of the initial corpus, and assigns the remaining corpus to the key of the initial corpus. In the cluster with the smallest Euclidean distance data between words, n basic clusters are obtained, and the remaining corpus is the financial corpus except the initial corpus; finally, the server calculates the average distance data of each basic cluster separately, and according to the average distance of each basic cluster The distance data determines multiple groups of classification corpus. Specifically, the server calculates the average distance data of each basic cluster separately, and determines the average distance data as the updated distance data; Group classification corpus.

It should be noted that the financial business scenario here refers to the human-computer interaction scenario in the financial field, such as the human-computer interaction scenario of the financial lending business, in which users can handle financial lending projects through human-computer interaction. The financial corpus in the financial business scenario here is the commonly used sentences in the corresponding financial scenario. Taking the financial lending business scenario as an example, the corresponding financial corpus at least includes: loan terms, credit authorization letter, basic user information, contact information, etc. corpus.

When building a discourse structure tree for human-computer interaction, it is first necessary to classify the financial corpus in the financial business scenario. Here, the K-value clustering method is used to classify the financial corpus in multiple financial business scenarios. Clustering is an attempt to Divide the samples in the dataset into several disjoint subsets, each of which is called a cluster. Through this division, each cluster may correspond to some different categories. In the clustering algorithm, the k-means algorithm is the most commonly used clustering algorithm. The k-means uses the distance as an evaluation index for similarity. The samples are clustered into different clusters. The closer the distance between the two points, the greater the similarity, so as to obtain compact and independent clusters as the clustering target. Its working principle is as follows: first, k objects are randomly selected in the data point set D, and each object represents the initial mean or center of a cluster; secondly, for each remaining object, according to its Euclidean distance from the center of each cluster , assign it to the most similar cluster; then use the k-means algorithm to iteratively improve the inner variation, for each cluster, use the objects assigned to the cluster in the last iteration to calculate the new mean; then use the updated mean as the new mean The cluster centers of , reallocate all objects; finally continue to iterate until the allocation is stable. In the present application, the K-value clustering method is used to classify the financial corpus under multiple financial business scenarios, so as to obtain multiple groups of classified corpus.

202. Sort each group of classification corpora according to the preset process sequence, obtain multiple sets of sorted corpora, and create multiple sets of financial business structure trees according to the multiple sets of sorted corpus;

Specifically, the server first sorts each group of classification corpora in the order of the preset process sequence to obtain multiple groups of sorted corpus; then the server inputs each group of sorted corpus into the structure tree builder to generate multiple sets of financial business structure trees. , the server firstly inputs each group of sorted corpus into the tree structure table of the structure tree builder, and the tree structure table includes the node attributes and node fields of each group of sorted corpus; secondly, the server uses the node attributes and calling functions to sort multiple groups of The node fields of the corpus are loaded into the initial financial structure tree, and the basic financial structure tree is obtained; then the server traverses the node fields whose node attributes are the first level, and loads the sorting corpus corresponding to the node fields of the first level to the corresponding nodes of the first level On the field, the first traversal structure tree is obtained; finally, the server traverses the node field whose node attribute is the second level, and loads the sorting corpus corresponding to the second level node field to the corresponding second level node field, and obtains the second traversal Structure tree, until the node fields of all node attributes are traversed, and multiple groups of financial business structure trees are generated.

Before establishing the financial business structure tree, it is necessary to sort each group of classified corpus. For example, in the financial business structure tree for signing the credit authorization letter, the credit authorization letter needs to be displayed first, and then the page that needs to be signed by the user is displayed as the A process needs to have a corresponding display order, so it is necessary to sort the financial corpus in each group of classified corpus according to the preset process order, and then construct the financial business structure tree after sorting.

When building a financial business structure tree, you need to first input the sorting corpus into the structure tree builder for creating the structure tree, and record the node attributes and nodes of each group of sorting corpus in the tree structure table in the structure tree builder Field, the node attribute here refers to the attribute corresponding to the sorting corpus, for example: the signing of the credit authorization letter for the sorting corpus belongs to the signature attribute, and the contact input of the sorting corpus belongs to the information input attribute, and the node field here refers to the key corresponding to the sorting corpus Words, for example: the keyword corresponding to the ordering corpus to sign the authorization letter for credit investigation is the signing authorization letter. In the constructed financial business structure tree, the sorting corpus is distributed on the nodes of the financial business structure tree, and the keywords corresponding to the sorting corpus are displayed on the corresponding node. It should be noted that, the display order of the financial business structure tree is to display the branch nodes hierarchically from the root node.

203. Obtain the input problem data during human-computer interaction, match the corresponding financial business structure tree through the input problem data, determine the corresponding financial business structure tree as the target financial structure tree, and search for the corresponding financial business structure tree in the target financial structure tree. target matching answer, and return target matching answer;

204. Acquire the jump signal data during human-computer interaction, filter the forecast financial structure tree among multiple sets of financial business structure trees through the jump signal data, and jump from the target financial structure tree to the forecast financial structure tree, and return to the forecast financial structure tree interface;

Specifically, the server obtains the jump signal data during human-computer interaction, selects the financial business structure tree whose process sequence is located after the target financial structure tree among the multiple groups of financial business structure trees according to the preset process sequence, and places the process sequence in the target financial structure tree. The financial business structure tree after the structure tree is determined as the forecast financial structure tree; the server obtains the node field whose node attribute is the first level of the forecast financial structure tree, and inputs the node field whose node attribute is the first level into the jump function, using The jump function jumps from the target financial structure tree to the forecast financial structure tree, and returns to the operation interface of the forecast financial structure tree.

First, the server obtains the jump signal data generated during human-computer interaction, and uses the jump signal data as a signal to query multiple financial business structure trees as the predicted financial structure tree after the target financial structure tree, and obtain the predicted financial structure tree The node attribute is the node field of the first level, and the node field is the jump address. Entering the node field into the jump function can realize the jump effect of jumping from the target financial structure tree to the forecast financial structure tree. After the jump, return to the operation interface of the forecast financial structure tree, so that the user can perform the next operation on the operation interface of the forecast financial structure tree.

205. Acquire the input operation data when the human-computer interaction is performed on the operation interface, query the prediction matching answer corresponding to the input operation data in the prediction financial structure tree, and return the prediction matching answer;

The server obtains the input operation data when the user performs human-computer interaction on the operation interface of the forecast financial structure tree, queries the forecast matching answer corresponding to the input operation data in the corresponding forecast financial structure tree, and returns the forecast matching answer. It should be noted that the operation process here is the same as the operation process in the above-mentioned step 203, but the specific query content is also different due to the difference of the predicted financial structure tree to be queried.

206. Use a display system to display the predicted matching answer.

After obtaining the predicted matching answer, the server will display the obtained predicted matching answer through the display system, so that the user can better clarify the query content.

The human-computer interaction method in the embodiment of the present application has been described above, and the human-computer interaction device in the embodiment of the present application is described below. Please refer to FIG. 3 . An embodiment of the human-computer interaction device in the embodiment of the present application includes:

The classification module 301 is used to obtain financial corpus under the financial business scenario, and use the K-value clustering method to classify the financial corpus to obtain multiple groups of classified corpora, and the financial corpus at least includes signing the financial information corpus and supplementing the financial information. Corpus and input financial contact corpus;

The sorting module 302 is configured to sort each group of classification corpora according to the preset process sequence, obtain multiple sets of sorted corpora, and create multiple sets of financial business structure trees according to the multiple sets of sorted corpora;

The determination module 303 is used to obtain the input problem data during human-computer interaction, match the corresponding financial business structure tree through the input problem data, and determine the corresponding financial business structure tree as the target financial structure tree. Find the target matching answer corresponding to the input question data in the financial structure tree, and return the target matching answer;

The jump module 304 is used to obtain the jump signal data during human-computer interaction, screen and predict the financial structure tree in the multiple groups of financial business structure trees through the jump signal data, and jump from the target financial structure tree Go to the forecast financial structure tree, and return to the operation interface of the forecast financial structure tree;

Returning module 305 is configured to acquire input operation data when performing human-computer interaction on the operation interface, query the prediction matching answer corresponding to the input operation data in the prediction financial structure tree, and return the prediction matching answer.

Referring to FIG. 4, another embodiment of the human-computer interaction device in the embodiment of the present application includes:

Optionally, the classification module 301 includes:

The obtaining unit 3011 is configured to obtain financial corpus under the financial business scenario, where the financial corpus at least includes signing financial information corpus, supplementing financial information corpus, and entering financial contact corpus;

A selection unit 3012, configured to select n of the financial corpora as initial corpora, where n∈{2,3...,k-1}, k is the number of financial corpora;

Assigning unit 3013, for calculating the Euclidean distance data between the keywords of the remaining corpus and the keywords of the initial corpus, and assigning the remaining corpus to the cluster with the smallest Euclidean distance data between the keywords of the initial corpus , to obtain n basic clusters, and the remaining corpus is the financial corpus except the initial corpus;

The determining unit 3014 is configured to separately calculate the average distance data of each basic cluster, and determine multiple groups of classification corpora according to the average distance data of each basic cluster.

Optionally, the determining unit 3014 is specifically configured to:

Calculate the average distance data of each basic cluster respectively, and determine the average distance data as the update distance data;

The remaining corpus is redistributed by using the updated distance data until the distribution is stable, and multiple groups of classification corpora are obtained.

Optionally, the sorting module 302 includes:

The sorting unit 3021 is used to sort each group of classified corpora according to the sequence of the preset process sequence, and obtain multiple groups of sorted corpora;

The generating unit 3022 is configured to input each group of sorted corpus into the structure tree builder to generate multiple groups of financial business structure trees.

Optionally, the generating unit 3022 is specifically used for:

Input each group of sorting corpus into the tree structure table of the structure tree builder, and the tree structure table includes node attributes and node fields of each group of sorting corpus;

Load the node fields of multiple groups of sorted corpora into the initial financial structure tree by using the node attributes and calling functions to obtain the basic financial structure tree;

Traversing the node fields whose attributes are the first level, and loading the sorting corpus corresponding to the node fields of the first level onto the corresponding node fields of the first level, to obtain the first traversed structure tree;

Traverse the node fields whose node attributes are the second level, load the sorting corpus corresponding to the node fields of the second level to the corresponding node fields of the second level, and obtain the second traversal structure tree until all the node attributes are traversed. The node field of , generates multiple groups of financial business structure trees.

Optionally, the jumping module 304 is specifically used for:

Obtain the jump signal data during human-computer interaction, screen the financial business structure tree whose process order is located after the target financial structure tree in the multiple groups of financial business structure trees according to the preset process sequence, and place the process sequence in the financial business structure tree. The financial business structure tree after the target financial structure tree is determined as the predicted financial structure tree;

Obtain the node field whose node attribute is the first level of the predicted financial structure tree, input the node field whose node attribute is the first level into the jump function, and use the jump function to retrieve the target financial structure tree from the target financial structure tree. Jump to the forecast financial structure tree, and return to the operation interface of the forecast financial structure tree.

Optionally, the human-computer interaction device further includes:

The display module 306 is configured to display the predicted matching answer by using the display system.

3 and 4 above describe in detail the human-computer interaction device in the embodiment of the present application from the perspective of modular functional entities, and the following describes the human-computer interaction device in the embodiment of the present application in detail from the perspective of hardware processing.

5 is a schematic structural diagram of a human-computer interaction device provided by an embodiment of the present application. The human-computer interaction device 500 may vary greatly due to different configurations or performances, and may include one or more central processing units (central processing units). , CPU) 510 (eg, one or more processors) and memory 520, one or more storage media 530 (eg, one or more mass storage devices) storing application programs 533 or data 532. Among them, the memory 520 and the storage medium 530 may be short-term storage or persistent storage. The program stored in the storage medium 530 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the human-computer interaction device 500 . Furthermore, the processor 510 may be configured to communicate with the storage medium 530 to execute a series of instruction operations in the storage medium 530 on the human-computer interaction device 500 .

Human-computer interaction device 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input and output interfaces 560, and/or, one or more operating systems 531, such as Windows Server , Mac OS X, Unix, Linux, FreeBSD and more. Those skilled in the art can understand that the structure of the human-computer interaction device shown in FIG. 5 does not constitute a limitation on the human-computer interaction device, and may include more or less components than those shown in the figure, or combine some components, or different Component placement.

The present application also provides a human-computer interaction device, the computer device includes a memory and a processor, the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, causes the processor to execute the above-mentioned various embodiments. The steps of the human-computer interaction method.

The present application also provides a computer-readable storage medium, and the computer-readable storage medium may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores computer instructions, and when the computer instructions are executed on the computer, the computer performs the following steps: acquiring financial corpus under a financial business scenario, and classifying the financial corpus by using a K-value clustering method , to obtain multiple sets of classified corpus, the financial corpus at least includes the signed financial information corpus, the supplementary financial information corpus, and the input financial contact corpus; sort each group of classified corpus according to the preset process sequence, and obtain multiple sets of sorted corpus, and according to The multiple sets of sorted corpora create multiple sets of financial business structure trees; the input question data during human-computer interaction is obtained, the corresponding financial business structure tree is matched by the input question data, and the corresponding financial business structure tree is determined as a target Financial structure tree, find the target matching answer corresponding to the input question data in the target financial structure tree, and return the target matching answer; obtain the jump signal data during human-computer interaction, and pass the jump signal data through the jump signal data. Screening the forecast financial structure tree among the multiple groups of financial business structure trees, and jumping from the target financial structure tree to the forecast financial structure tree, and returning to the operation interface of the forecast financial structure tree; the acquisition is performed on the operation interface Input operation data during human-computer interaction, query the prediction matching answer corresponding to the input operation data in the prediction financial structure tree, and return the prediction matching answer.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system, device and unit described above may refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

The blockchain referred to in this application is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm. Blockchain, essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block. The blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

The integrated unit, if implemented in the form of a software functional unit and sold or used as an independent product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the present application can be embodied in the form of software products in essence, or the parts that contribute to the prior art, or all or part of the technical solutions, and the computer software products are stored in a storage medium , including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk and other media that can store program codes .

As mentioned above, the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand: The technical solutions described in the embodiments are modified, or some technical features thereof are equivalently replaced; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions in the embodiments of the present application.

Claims

A human-computer interaction method, the human-computer interaction method comprising:

Obtain the financial corpus under the financial business scenario, and use the K-value clustering method to classify the financial corpus to obtain multiple groups of classified corpora, where the financial corpus at least includes signing the financial information corpus, supplementing the financial information corpus and entering financial contacts corpus;

Sort each group of classification corpus according to the preset process sequence to obtain multiple sets of sorting corpus, and create multiple sets of financial business structure trees according to the multiple sets of sorting corpus;

Obtain the input problem data during human-computer interaction, match the corresponding financial business structure tree through the input problem data, determine the corresponding financial business structure tree as the target financial structure tree, and search for the target financial structure tree in the target financial structure tree. The target matching answer corresponding to the input question data is returned, and the target matching answer is returned;

Obtain the jump signal data during human-computer interaction, screen the predicted financial structure tree in the multiple groups of financial business structure trees through the jump signal data, and jump from the target financial structure tree to the predicted financial structure tree, returning the operation interface of the predicted financial structure tree;

Obtain the input operation data when performing human-computer interaction on the operation interface, query the forecast matching answer corresponding to the input operation data in the forecast financial structure tree, and return the forecast matching answer.
The human-computer interaction method according to claim 1, wherein the obtaining financial corpus in a financial business scenario, and classifying the financial corpus using a K-value clustering method to obtain multiple groups of classified corpora, the financial corpus At least the corpus of signing financial information, supplementing the corpus of financial information and entering the financial contact corpus includes:

Obtain financial corpus under the financial business scenario, where the financial corpus at least includes the corpus of signing financial information, the corpus of supplementary financial information, and the corpus of entering financial contacts;

Select n of the financial corpora as the initial corpus, where n∈{2,3...,k-1}, k is the number of financial corpora;

Calculate the Euclidean distance data between the keywords of the remaining corpus and the keywords of the initial corpus, and assign the remaining corpus to the cluster with the smallest Euclidean distance data between the keywords of the initial corpus, and obtain n basic clusters , the remaining corpus is the financial corpus other than the initial corpus;

Calculate the average distance data of each basic cluster separately, and determine multiple groups of classification corpus according to the average distance data of each basic cluster.
The human-computer interaction method according to claim 2, wherein said calculating the average distance data of each basic cluster respectively, and determining multiple groups of classification corpora according to the average distance data of each basic cluster comprises:

Calculate the average distance data of each basic cluster respectively, and determine the average distance data as the update distance data;

The remaining corpus is redistributed by using the updated distance data until the distribution is stable, and multiple groups of classification corpora are obtained.
The human-computer interaction method according to claim 1, wherein the sorting of each group of classified corpora according to a preset process sequence is performed to obtain multiple sets of sorted corpora, and multiple sets of financial business structure trees are created according to the multiple sets of sorted corpus include:

Sort each group of classified corpus according to the sequence of the preset process sequence to obtain multiple groups of sorted corpus;

Input each set of ranking corpus into the structure tree builder to generate multiple sets of financial business structure trees.
The human-computer interaction method according to claim 4, wherein, inputting each group of sorted corpus into a structure tree builder to generate multiple groups of financial business structure trees comprises:

Input each group of sorting corpus into the tree structure table of the structure tree builder, and the tree structure table includes node attributes and node fields of each group of sorting corpus;

Load the node fields of multiple groups of sorted corpora into the initial financial structure tree by using the node attributes and calling functions to obtain the basic financial structure tree;

Traversing the node fields whose attributes are the first level, and loading the sorting corpus corresponding to the node fields of the first level onto the corresponding node fields of the first level, to obtain the first traversed structure tree;

Traverse the node fields whose node attributes are the second level, load the sorting corpus corresponding to the node fields of the second level to the corresponding node fields of the second level, and obtain the second traversal structure tree until all the node attributes are traversed. The node field of , generates multiple groups of financial business structure trees.
The human-computer interaction method according to claim 5, wherein, in the process of acquiring jump signal data during human-computer interaction, screening and predicting financial structure trees among the multiple groups of financial business structure trees by using the jump signal data, And jump from the target financial structure tree to the predicted financial structure tree, and the operation interface for returning to the predicted financial structure tree includes:

Obtain the jump signal data during human-computer interaction, screen the financial business structure tree whose process order is located after the target financial structure tree in the multiple groups of financial business structure trees according to the preset process sequence, and place the process sequence in the financial business structure tree. The financial business structure tree after the target financial structure tree is determined as the predicted financial structure tree;

Obtain the node field whose node attribute is the first level of the predicted financial structure tree, input the node field whose node attribute is the first level into the jump function, and use the jump function to retrieve the target financial structure tree from the target financial structure tree. Jump to the forecast financial structure tree, and return to the operation interface of the forecast financial structure tree.
The human-computer interaction method according to any one of claims 1-6, wherein, in the acquisition of the input operation data when the human-computer interaction is performed on the operation interface, the input operation is queried in the forecast financial structure tree After the predicted matching answer corresponding to the data is returned, and the predicted matching answer is returned, the human-computer interaction method further includes:

The predicted matching answer is displayed using the display system.
A human-computer interaction device, comprising a memory, a processor, and computer-readable instructions stored on the memory and executable on the processor, wherein the processor implements the following steps when executing the computer-readable instructions:

Obtain the financial corpus under the financial business scenario, and use the K-value clustering method to classify the financial corpus to obtain multiple groups of classified corpora, and the financial corpus at least includes the signing of the financial information corpus, the supplementary financial information corpus, and the entry of financial contacts. corpus;

Sort each group of classification corpus according to the preset process sequence to obtain multiple sets of sorting corpus, and create multiple sets of financial business structure trees according to the multiple sets of sorting corpus;

Obtain the input problem data during human-computer interaction, match the corresponding financial business structure tree through the input problem data, determine the corresponding financial business structure tree as the target financial structure tree, and search for the target financial structure tree in the target financial structure tree. The target matching answer corresponding to the input question data is returned, and the target matching answer is returned;

Obtain the jump signal data during human-computer interaction, screen the predicted financial structure tree in the multiple groups of financial business structure trees through the jump signal data, and jump from the target financial structure tree to the predicted financial structure tree, returning the operation interface of the predicted financial structure tree;

Obtain the input operation data when performing human-computer interaction on the operation interface, query the forecast matching answer corresponding to the input operation data in the forecast financial structure tree, and return the forecast matching answer.
The human-computer interaction device according to claim 8, wherein the processor executes the computer-readable instructions to realize the obtaining of the financial corpus under the financial business scenario, and uses a K-value clustering method to perform the financial corpus on the financial corpus. Classification to obtain multiple groups of classified corpus, the financial corpus at least includes the following steps when signing the financial information corpus, supplementing the financial information corpus and entering the financial contact corpus:

Obtain financial corpus under the financial business scenario, where the financial corpus at least includes the corpus of signing financial information, the corpus of supplementary financial information, and the corpus of entering financial contacts;

Select n of the financial corpora as the initial corpus, where n∈{2,3...,k-1}, k is the number of financial corpora;

Calculate the Euclidean distance data between the keywords of the remaining corpus and the keywords of the initial corpus, and assign the remaining corpus to the cluster with the smallest Euclidean distance data between the keywords of the initial corpus, and obtain n basic clusters , the remaining corpus is the financial corpus other than the initial corpus;

Calculate the average distance data of each basic cluster separately, and determine multiple groups of classification corpus according to the average distance data of each basic cluster.
The human-computer interaction device according to claim 9, wherein the processor executes the computer-readable instructions to realize the calculation of the average distance data of each basic cluster respectively, and calculate the average distance data of each basic cluster according to the average distance data of each basic cluster. When the distance data is used to determine multiple groups of classification corpus, the following steps are included:

Calculate the average distance data of each basic cluster respectively, and determine the average distance data as the update distance data;

The remaining corpus is redistributed by using the updated distance data until the distribution is stable, and multiple groups of classification corpora are obtained.
The human-computer interaction device according to claim 8, wherein the processor executes the computer-readable instructions to achieve the sorting of each group of classified corpora according to a preset sequence of processes, to obtain multiple groups of sorted corpora, and according to the When multiple sets of sorting corpus create multiple sets of financial business structure trees, the following steps are included:

Sort each group of classified corpus according to the sequence of the preset process sequence to obtain multiple groups of sorted corpus;

Input each set of ranking corpus into the structure tree builder to generate multiple sets of financial business structure trees.
The human-computer interaction device according to claim 11, wherein, when the processor executes the computer-readable instructions to realize the inputting each group of sorted corpora into the structure tree builder, and when generating multiple groups of financial business structure trees, including The following steps:

Input each group of sorting corpus into the tree structure table of the structure tree builder, and the tree structure table includes node attributes and node fields of each group of sorting corpus;

Load the node fields of multiple groups of sorted corpora into the initial financial structure tree by using the node attributes and calling functions to obtain the basic financial structure tree;

Traversing the node fields whose attributes are the first level, and loading the sorting corpus corresponding to the node fields of the first level onto the corresponding node fields of the first level, to obtain the first traversed structure tree;

Traverse the node fields whose node attributes are the second level, load the sorting corpus corresponding to the node fields of the second level to the corresponding node fields of the second level, and obtain the second traversal structure tree until all the node attributes are traversed. The node field of , generates multiple groups of financial business structure trees.
The human-computer interaction device according to claim 12, wherein the processor executes the computer-readable instructions to realize the acquisition of jump signal data during human-computer interaction, and the jump signal data is used in the multiple Screening the forecast financial structure tree in the group financial business structure tree, and jumping from the target financial structure tree to the forecast financial structure tree, and returning to the operation interface of the forecast financial structure tree, the following steps are also included:

Obtain the jump signal data during human-computer interaction, screen the financial business structure tree whose process order is located after the target financial structure tree in the multiple groups of financial business structure trees according to the preset process sequence, and place the process sequence in the financial business structure tree. The financial business structure tree after the target financial structure tree is determined as the predicted financial structure tree;

Obtain the node field whose node attribute is the first level of the predicted financial structure tree, input the node field whose node attribute is the first level into the jump function, and use the jump function to retrieve the target financial structure tree from the target financial structure tree. Jump to the forecast financial structure tree, and return to the operation interface of the forecast financial structure tree.
The human-computer interaction device according to any one of claims 8-13, wherein the processor executes the computer-readable instructions to achieve the acquisition of input operation data when performing human-computer interaction on the operation interface, When querying the predicted matching answer corresponding to the input operation data in the predicted financial structure tree and returning the predicted matching answer, the following steps are also included:

The predicted matching answer is displayed using the display system.
A computer-readable storage medium, storing computer instructions in the computer-readable storage medium, when the computer instructions are executed on a computer, the computer is made to perform the following steps:

Obtain the financial corpus under the financial business scenario, and use the K-value clustering method to classify the financial corpus to obtain multiple groups of classified corpora, and the financial corpus at least includes the signing of the financial information corpus, the supplementary financial information corpus, and the entry of financial contacts. corpus;

Sort each group of classification corpus according to the preset process sequence to obtain multiple sets of sorting corpus, and create multiple sets of financial business structure trees according to the multiple sets of sorting corpus;

Obtain the input problem data during human-computer interaction, match the corresponding financial business structure tree through the input problem data, determine the corresponding financial business structure tree as the target financial structure tree, and search for the target financial structure tree in the target financial structure tree. The target matching answer corresponding to the input question data is returned, and the target matching answer is returned;

Obtain the jump signal data during human-computer interaction, screen the predicted financial structure tree in the multiple groups of financial business structure trees through the jump signal data, and jump from the target financial structure tree to the predicted financial structure tree, returning the operation interface of the predicted financial structure tree;

Obtain the input operation data when performing human-computer interaction on the operation interface, query the forecast matching answer corresponding to the input operation data in the forecast financial structure tree, and return the forecast matching answer.
The computer-readable storage medium of claim 15, when the computer instructions are executed on a computer, causing the computer to further perform the following steps:

Obtain financial corpus under the financial business scenario, where the financial corpus at least includes the corpus of signing financial information, the corpus of supplementary financial information, and the corpus of entering financial contacts;

Select n of the financial corpora as the initial corpus, where n∈{2,3...,k-1}, k is the number of financial corpora;

Calculate the Euclidean distance data between the keywords of the remaining corpus and the keywords of the initial corpus, and assign the remaining corpus to the cluster with the smallest Euclidean distance data between the keywords of the initial corpus, and obtain n basic clusters , the remaining corpus is the financial corpus other than the initial corpus;

Calculate the average distance data of each basic cluster separately, and determine multiple groups of classification corpus according to the average distance data of each basic cluster.
The computer-readable storage medium of claim 16, when the computer instructions are executed on a computer, causing the computer to further perform the following steps:

Calculate the average distance data of each basic cluster respectively, and determine the average distance data as the update distance data;

The remaining corpus is redistributed by using the updated distance data until the distribution is stable, and multiple groups of classification corpora are obtained.
The computer-readable storage medium of claim 15, when the computer instructions are executed on a computer, causing the computer to further perform the following steps:

Sort each group of classified corpus according to the sequence of the preset process sequence to obtain multiple groups of sorted corpus;

Input each set of ranking corpus into the structure tree builder to generate multiple sets of financial business structure trees.
The computer-readable storage medium of claim 18, the computer instructions, when executed on a computer, cause the computer to further perform the following steps:

Input each group of sorting corpus into the tree structure table of the structure tree builder, and the tree structure table includes node attributes and node fields of each group of sorting corpus;

Load the node fields of multiple groups of sorted corpora into the initial financial structure tree by using the node attributes and calling functions to obtain the basic financial structure tree;

Traversing the node fields whose attributes are the first level, and loading the sorting corpus corresponding to the node fields of the first level onto the corresponding node fields of the first level, to obtain the first traversed structure tree;

Traverse the node fields whose node attributes are the second level, load the sorting corpus corresponding to the node fields of the second level to the corresponding node fields of the second level, and obtain the second traversal structure tree until all the node attributes are traversed. The node field of , generates multiple groups of financial business structure trees.
A human-computer interaction device, the human-computer interaction device comprising:

The classification module is used to obtain the financial corpus under the financial business scenario, and use the K-value clustering method to classify the financial corpus to obtain multiple groups of classified corpora, and the financial corpus at least includes the signed financial information corpus and the supplementary financial information corpus and input financial contact corpus;

The sorting module is used for sorting each group of classification corpus according to the preset process sequence, obtaining multiple sets of sorting corpus, and creating multiple sets of financial business structure trees according to the multiple sets of sorting corpus;

The determination module is used to obtain the input problem data during human-computer interaction, and matches the corresponding financial business structure tree through the input problem data, and determines the corresponding financial business structure tree as the target financial structure tree. Find the target matching answer corresponding to the input question data in the structure tree, and return the target matching answer;

The jump module is used to obtain the jump signal data during human-computer interaction, screen and predict the financial structure tree in the multiple groups of financial business structure trees through the jump signal data, and jump from the target financial structure tree To the predicted financial structure tree, return the operation interface of the predicted financial structure tree;

The returning module is used for acquiring input operation data when performing human-computer interaction on the operation interface, querying the forecast financial structure tree for the forecast matching answer corresponding to the input operation data, and returning the forecast matching answer.