Multi-channel questionnaire generation method and device and electronic equipment

Abstract

The invention discloses a multi-channel questionnaire generation method and device and electronic equipment, and the method comprises the steps: obtaining an event log corresponding to a questionnaire trigger event when the questionnaire trigger event is monitored; analyzing a target service element contained in the event log, and determining a target questionnaire intention entity corresponding to the questionnaire trigger event in a preset questionnaire knowledge graph based on the target service element; determining a target questionnaire element matched with the target questionnaire intention entity in a preset questionnaire element library, and generating an initial questionnaire based on the target questionnaire element; and obtaining interaction environment information corresponding to the questionnaire trigger event, and adjusting the initial questionnaire based on the interaction environment information to generate a target questionnaire. Through the method, intelligent and accurate matching of the initial questionnaire and the service scene can be realized, multi-dimensional adaptive adjustment based on the interaction environment can be realized, the filling rate and the data quality can be improved, and the user experience can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of finance, in particular to a multi-channel questionnaire generation method and device and electronic equipment. BACKGROUND

[0002] In bank business, collecting customer feedback through questionnaire survey is a crucial means to improve product experience and service quality. With the diversification of electronic channels, customers can interact with banks through enterprise online banking, personal online banking, mobile banking App and other channels.

[0003] At present, in order to collect customer experience data of multiple channels, a unified evaluation interface is used to receive questionnaire trigger requests of each touch point and match and push corresponding preset questionnaires, which to some extent realizes the questionnaire coverage of multiple channels. However, in this method, the questionnaire content and style of different channels are different, which cannot utilize the customer's opinion feedback, satisfaction score and the like of the same function service in different channels, and the questionnaire generation mainly depends on the pre-set fixed questionnaire template library, which cannot dynamically generate survey content that deeply fits the scene according to different application scenes, and it is difficult to accurately mine user opinions, and the actual value of the questionnaire is difficult to reflect. In addition, the questionnaire generated by this method has poor channel adaptability, and the adaptation to different channels is mostly limited to simple responsive layout adjustment, and it is also impossible to adaptively adjust the questionnaire according to dynamic factors related to the interactive environment such as real-time device performance, network status and interaction preferences of users.

[0004] At present, there is no effective solution to the problem of low adaptability of the above questionnaire generation method to electronic channels, business scenarios and interactive environment and poor user experience. SUMMARY

[0005] The purpose of the embodiments of the present application is to provide a multi-channel questionnaire generation method, device and electronic equipment to solve the problem of low adaptability of the questionnaire generation method to electronic channels, business scenarios and interactive environment and poor user experience.

[0006] To solve the above technical problems, the first aspect of the present specification provides a multi-channel questionnaire generation method, comprising: monitoring a questionnaire trigger event, obtaining an event log corresponding to the questionnaire trigger event; analyzing a target business element contained in the event log, and determining a target questionnaire intention entity corresponding to the questionnaire trigger event based on the target business element in a preset questionnaire knowledge graph, wherein the questionnaire knowledge graph is composed of a plurality of entities and the association relationship between the entities, and the plurality of entities at least include a questionnaire intention entity and a business element entity; determine a target questionnaire element matching the target questionnaire intention entity in a preset questionnaire element library, and generate an initial survey questionnaire based on the target questionnaire element; obtain interaction environment information corresponding to the questionnaire triggering event, and adjust the initial survey questionnaire based on the interaction environment information to generate a target survey questionnaire; wherein the interaction environment information includes at least one of the following: a target channel corresponding to the target survey questionnaire, a first target device, device environment information, and user behavior data of a target user corresponding to the target survey questionnaire.

[0007] In some embodiments of the present specification, the questionnaire knowledge graph is constructed by the following way: obtain multi-source heterogeneous data for constructing the questionnaire knowledge graph, wherein the multi-source heterogeneous data at least includes historical business data, historical questionnaire data, customer service data, a preset questionnaire knowledge base, and user behavior data; process the historical business data to extract key business elements related to business logic as the business element entity; process the historical questionnaire data, the customer service data, and the preset questionnaire knowledge base to extract the questionnaire intention entity; process the historical business data, the user behavior data, and the historical questionnaire data to determine the association relationship between the business element entity and the questionnaire intention entity, and the relationship attribute data corresponding to each association relationship; construct the questionnaire knowledge graph based on the business element entity, the questionnaire intention entity, the association relationship, and the relationship attribute data corresponding to each association relationship.

[0008] In some embodiments of the present specification, the business element entity includes at least one of the following: a product entity, a service entity, a user journey node entity, a user label entity, and a channel entity; The relationship attribute data includes at least one of the following: relationship weight and relationship type.

[0009] In some embodiments of the present specification, the questionnaire knowledge graph is constructed based on the business element entity, the questionnaire intention entity, the association relationship, and the relationship attribute data corresponding to each association relationship, including: vectorize the questionnaire intention entity to obtain a questionnaire intention vector; construct the questionnaire knowledge graph by taking the determined business element entity and the questionnaire intention vector as nodes, the association relationship as edges connecting the nodes, and the relationship attribute data corresponding to each association relationship as parameters of each edge.

[0010] In some embodiments of the present disclosure, the target questionnaire intention entity corresponding to the questionnaire trigger event is determined based on the target business element in a preset questionnaire knowledge graph, comprising: generating a target query vector based on the target business element; determining a target node in the questionnaire knowledge graph to which the target query vector is mapped; determining the target questionnaire intention entity based on the questionnaire intention entities associated with the target node and the relationship attribute data between the target node and each associated questionnaire intention entity.

[0011] In some embodiments of the present disclosure, the target node in the questionnaire knowledge graph to which the target query vector is mapped is determined, comprising: performing similarity matching between the target query vector and the nodes in the questionnaire knowledge graph to obtain a plurality of candidate nodes; performing logical verification on the plurality of candidate nodes based on the business processing logic corresponding to the target business element, and screening the target node from the plurality of candidate nodes based on the verification result.

[0012] In some embodiments of the present disclosure, the target questionnaire element matching the target questionnaire intention entity is determined in a preset questionnaire element library, and an initial survey questionnaire is generated based on the target questionnaire element, comprising: calculating the similarity between the target questionnaire intention entity and each questionnaire element in the questionnaire element library, and screening the questionnaire element satisfying the preset similarity as the target questionnaire element; sorting the target questionnaire elements based on the logical dependency relationship between the target questionnaire elements, and embedding the logical relationship of the sorting result based on the logical dependency relationship to obtain a questionnaire element sequence; generating the initial survey questionnaire based on the questionnaire element sequence and the target business element.

[0013] In some embodiments of the present disclosure, the initial survey questionnaire is adjusted based on the interaction environment information to generate a target survey questionnaire, comprising: determining the target adjustment parameter corresponding to the initial survey questionnaire based on the interaction environment information, the target adjustment parameter comprising at least one of the following: rendering adjustment parameter, interaction adjustment parameter, content adjustment parameter, format adjustment parameter; adjusting the initial survey questionnaire based on the target adjustment parameter to obtain the target survey questionnaire.

[0014] In some embodiments of the present disclosure, the target adjustment parameter corresponding to the initial survey questionnaire is determined based on the interaction environment information, comprising: determine a user portrait label of the target user based on the user behavior data; normalize the user portrait label, the target channel, the first target device, and the device environment information to obtain an interaction environment vector; input the interaction environment vector into a preset adjustment strategy model, and output the target adjustment parameter.

[0015] In some embodiments of the present specification, the above method further comprises: construct a questionnaire identifier corresponding to the target questionnaire; obtain questionnaire update data of the target questionnaire in real time; obtain a current interaction environment corresponding to the second target device in response to the operation of the target user on the second target device; adjust the target questionnaire based on the current interaction environment and the questionnaire update data, and push the adjusted questionnaire to the second target device.

[0016] The second aspect of the present specification further provides a multi-channel questionnaire generation device, comprising: An acquisition module is configured to acquire event logs corresponding to a questionnaire triggering event when the questionnaire triggering event is monitored. A determination module is configured to analyze target business elements included in the event logs, and determine target questionnaire intention entities corresponding to the questionnaire triggering event in a preset questionnaire knowledge graph based on the target business elements, wherein the questionnaire knowledge graph is composed of a plurality of entities and association relationships between the entities, and the plurality of entities at least include questionnaire intention entities and business element entities. A generation module is configured to determine target questionnaire elements matching the target questionnaire intention entities in a preset questionnaire element library, and generate an initial questionnaire based on the target questionnaire elements. An adjustment module is configured to acquire interaction environment information corresponding to the questionnaire triggering event, and adjust the initial questionnaire based on the interaction environment information to generate a target questionnaire, wherein the interaction environment information includes at least one of the following: a target channel corresponding to the target questionnaire, a first target device, device environment information, and user behavior data of a target user to which the target questionnaire is pushed.

[0017] The third aspect of the present specification provides an electronic device, comprising: a memory and a processor, which are communicatively connected to each other, and the memory stores computer instructions, and the processor executes the computer instructions to perform the steps of the method of the first aspect.

[0018] The fourth aspect of the specification provides a computer storage medium, which stores computer program instructions, and the computer program instructions are executed to implement the steps of the method of the first aspect.

[0019] The fifth aspect of the specification provides a computer program product, which contains a computer program, and the computer program is executed by a processor to implement the steps of the method of the first aspect.

[0020] In the method for generating a multi-channel questionnaire provided in the embodiments of the specification, when a questionnaire triggering event is monitored, an event log corresponding to the questionnaire triggering event is acquired; target business elements contained in the event log are parsed, and a target questionnaire intention entity corresponding to the questionnaire triggering event is determined in a preset questionnaire knowledge graph based on the target business elements, the questionnaire knowledge graph is composed of a plurality of entities and the association relationship between the entities, and the plurality of entities at least include a questionnaire intention entity and a business element entity; a target questionnaire element matching the target questionnaire intention entity is determined in a preset questionnaire element library, and an initial survey questionnaire is generated based on the target questionnaire element; interaction environment information corresponding to the questionnaire triggering event is acquired, and the initial survey questionnaire is adjusted based on the interaction environment information to generate a target survey questionnaire; wherein the interaction environment information includes at least one of the following: a target channel corresponding to the target survey questionnaire, a first target device, device environment information, and user behavior data of a target user corresponding to the target survey questionnaire. Through the above method, the financial business knowledge graph is introduced, and intelligent and accurate matching of the initial survey questionnaire and the business scenario can be realized; further, based on the interaction environment information, multi-dimensional adaptive adjustment based on the interaction environment can be realized, so that the questionnaire can provide the best interactive experience in various channels, terminal devices and environments, and the filling rate and data quality can be improved, and the user experience can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the description of the embodiments or the prior art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0022] Figure 1 Fig. 1 shows a schematic diagram of the method for generating a multi-channel questionnaire provided in the embodiments of the specification; Figure 2 Fig. 2 shows a schematic diagram of the method for constructing a questionnaire knowledge graph provided in the embodiments of the specification; Figure 3 Fig. 3 shows a schematic diagram of the method for determining a target questionnaire intention entity provided in the embodiments of the specification; Figure 4 Fig. 1 shows a schematic diagram of a method for generating an initial questionnaire according to an embodiment of the present specification; Figure 5 Fig. 2 shows a schematic diagram of a multi-channel questionnaire generation device according to an embodiment of the present specification; Figure 6 Fig. 3 shows a schematic diagram of an electronic device according to an embodiment of the present specification. DETAILED DESCRIPTION

[0023] In order to enable those skilled in the art to better understand the technical solutions in the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should fall within the scope of protection of the present application.

[0024] It should be noted that the information and data related to the user involved in the embodiments of the present specification are information and data authorized by the user or authorized by relevant parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data comply with relevant laws, regulations and standards, necessary security measures are taken, do not violate public order and good customs, and provide corresponding operation portal for users or relevant parties to choose authorization or refusal.

[0025] It should also be noted that in the embodiments of the present specification, some software, components, models and other existing solutions in the industry may be mentioned, which should be considered as exemplary, and the purpose is only to illustrate the feasibility of the implementation of the technical solutions of the present application, but it does not mean that the applicant has or will necessarily use the solution.

[0026] The multi-channel questionnaire generation method provided by the embodiments of the present application will be introduced below in conjunction with the drawings. It can be understood that the multi-channel questionnaire generation method in the embodiments of the present specification can be applied to a specific electronic device, or can be applied to a system deployed on some devices, such as a questionnaire system, etc. The embodiments of the present specification take the multi-channel questionnaire generation method applied to the questionnaire system (hereinafter referred to as the system) as an example for description and introduction, in order to better clarify each step and its specific execution process.

[0027] Figure 1A schematic diagram of a multi-channel questionnaire generation method provided by an embodiment of the present specification is shown. Although the present specification provides method operation steps or device structures as shown in the following embodiments or drawings, more or part of the operation steps or module units can be included in the method or device based on conventional or non-creative labor. In steps or structures that do not have essential causal relationships in logic, the execution order of these steps or the module structure of the device is not limited to the execution order or module structure shown in the embodiments or drawings of the present specification. When the method or module structure is applied in actual devices, servers or terminal products, it can be sequentially executed or executed in parallel (for example, in a parallel processor or a multi-thread processing environment, even including a distributed processing, server cluster implementation environment) according to the method or module structure shown in the embodiments or drawings. For example, Figure 1 As shown, the method can include steps S101 to S104, specifically: S101: When a questionnaire trigger event is monitored, the event log corresponding to the questionnaire trigger event is obtained.

[0028] It can be understood that the questionnaire trigger event can be an event signal generated by the user when completing a specific business operation in each channel of the bank (such as mobile banking, online banking, counter system, etc.), which can trigger the generation of a survey questionnaire. The questionnaire trigger event may, for example, complete a financial product purchase, end a branch business, process an account loss, or end a customer service consultation, and then the system can determine that a survey questionnaire needs to be generated when the corresponding event is monitored, and then the corresponding event log can be obtained. It can be understood that the system will generate an event log containing business key information when executing a business process. The event log can be a structured or semi-structured data record, which can include but is not limited to: event ID, timestamp, user ID, channel type, business type, business parameters (such as business type, transaction amount, operation time, etc.), device basic information (such as device model, system version), etc. In specific implementation, the system can capture the questionnaire trigger event in real time through the listening module deployed in each business system of the bank, and after capturing the questionnaire trigger event, the system can obtain the event log corresponding to the questionnaire trigger event from each business system through a message queue (such as Kafka) or API interface, and temporarily store it in a data buffer pool for subsequent processing.

[0029] S102: Analyze the target business elements contained in the event log, and determine the target questionnaire intention entity corresponding to the questionnaire trigger event in the preset questionnaire knowledge graph based on the target business elements, wherein the questionnaire knowledge graph is composed of multiple entities and the association relationship between the entities, and the multiple entities at least include questionnaire intention entities and business element entities.

[0030] It can be understood that the target business element can be an element associated with the business corresponding to the questionnaire trigger event, and can include but is not limited to: a product name involved (such as "XX financial product"), a service type (such as "loan face signing"), a user journey node (such as "payment confirmation"), a user attribute (such as "gold card customer"), and the like. Illustratively, from the log text "the user completes the purchase of the "stable A number" financial product, the amount is 50000 yuan", the product entity "stable A number" and the journey node entity "financial product purchase" can be extracted.

[0031] It can be understood that the questionnaire knowledge graph is a pre-constructed knowledge base stored in a graph structure, which can accurately map the business scenario and the questionnaire demand (questionnaire intention entity). The nodes in the questionnaire knowledge graph can include business element entities (such as products, services, journey nodes, channels, etc.) and questionnaire intention entities (such as "income satisfaction survey", "process convenience feedback", "alternative product preference collection", "service timeliness evaluation", etc.), and the edges connecting the nodes can represent the association relationship between entities, and the edges can be attached with relationship attribute data (such as relationship weight, association type, etc.).

[0032] In specific implementation, the system can convert the extracted target business element into a query vector, find the most matched business element node in the graph through vector similarity calculation, and then locate the target questionnaire intention entity most relevant to the business scenario according to the attributes and types of the edges and the like. For example, the "financial product purchase" node is connected to "income expectation survey" and "purchase motivation analysis" through a high-weight edge, and the intention entity can be used as the target questionnaire intention entity corresponding to the questionnaire trigger event.

[0033] In some embodiments of the present specification, the target business element contained in the parsed event log can specifically include that the system uses natural language processing (NLP) technology, especially named entity recognition (NER) and relationship extraction technology, to extract key business elements from the unstructured text field of the event log.

[0034] S103: Determine a target questionnaire element matching the target questionnaire intention entity in a preset questionnaire element library, and generate an initial survey questionnaire based on the target questionnaire element.

[0035] It can be understood that the questionnaire element library can be a database storing questionnaire basic components, including question atoms (such as single-choice questions, multiple-choice questions, open questions), option sets (such as satisfaction rating options "1-5 points", product type options "fixed income type / equity type / mixed type"), logic control rules (such as "jump to reason survey question after selecting 'unsatisfied'"), and the like. All elements can support flexible combination and calling.

[0036] In practice, the system can filter out elements with a similarity higher than a preset threshold as target questionnaire elements based on the similarity between the target questionnaire intention entity and each element in the questionnaire element library. Subsequently, the filtered elements can be sorted and logically assembled based on logical dependency relationships between the target questionnaire elements (e.g., question B must appear after question A selects a specific option), forming an initial survey questionnaire that is structured and consistent with the survey logic.

[0037] In S104, the interactive environment information corresponding to the questionnaire triggering event is obtained, and the initial survey questionnaire is adjusted based on the interactive environment information to generate a target survey questionnaire.

[0038] The interactive environment information includes at least one of the following: a target channel corresponding to the target survey questionnaire, a first target device, device environment information, and user behavior data of a target user corresponding to the target survey questionnaire.

[0039] It can be understood that the interactive environment information can be used to describe the current interactive environment of the user, including but not limited to the channel (i.e., the target channel) to which the survey questionnaire is to be pushed, the device (i.e., the first target device) to which the questionnaire is to be pushed, the device-related environment information, and the user behavior data of the target user to which the questionnaire is to be pushed. The target channel can include mobile banking APP, PC online banking, WeChat mini program, etc., the first target device can include device type (e.g., old machine, smart phone, notebook computer, self-service terminal in bank branch, etc.), screen size, operating system, device hardware configuration (e.g., whether it is a touch screen, whether it has operation keys), etc., the device environment information can include network status (4G / 5G / WiFi, delay), power, ambient light intensity, etc., and the user behavior data can include user historical operation records (e.g., questionnaire filling time, preferred questionnaire type, common interaction mode), user historical interaction preferences (e.g., whether the user often uses voice, whether the user tends to operate quickly), etc.

[0040] It can be understood that the system can adjust the initial survey questionnaire based on the interactive environment information, such as adjusting the rendering style, the interaction mode, the content length, etc., to generate a target survey questionnaire suitable for the current interactive environment, and can push the target survey questionnaire to the first target electronic device of the target channel to display the target survey questionnaire to the target user through the first target electronic device.

[0041] In the embodiments of the present specification, by introducing the financial business knowledge graph, intelligent and accurate matching of the initial survey questionnaire and the business scenario can be realized. Furthermore, based on the interactive environment information, multi-dimensional adaptive adjustment based on the interactive environment can be realized, so that the questionnaire can provide the best interactive experience in various channels, terminal devices and environments, and the filling rate and data quality can be improved, and the user experience can be improved.

[0042] It can be understood that the above method provided by the embodiments of the present specification can be applied to an electronic device. The electronic device can refer to an electronic device with data computing, processing and storage capabilities. The electronic device can be a terminal such as a PC (Personal Computer), a tablet computer, a smart phone, a wearable device, a smart robot, etc. The electronic device can also be a server. The server can be a standalone physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0043] In some embodiments of the present specification, the monitoring of the relevant event can be achieved by registering and configuring the questionnaire trigger event. Specifically, the corresponding preset questionnaire trigger event can be pre-configured in the event registration center for the bank core business system (such as the financial transaction system, the credit card system, and the branch service system). For each type of event, its event identifier, trigger condition, trigger delay, associated business system identifier, etc. can be defined. Further, after completing the event registration and configuration, real-time monitoring can be achieved by using event monitoring technology, for example, by using a combination of distributed event bus and message queue architecture. After monitoring the questionnaire trigger event, the event log corresponding to the event can be obtained, and the event log can be further analyzed to achieve the adaptability of the questionnaire to the corresponding business scenario.

[0044] Among them, the event identifier can be used to distinguish different questionnaire trigger events, such as EVT_FIN_TRADE_REDEEM representing a financial redemption event, EVT_CRD_REPAY representing a credit card repayment event, etc. The trigger condition can be used to represent that the corresponding event is triggered under this condition, such as “the financial redemption transaction state changes to ‘success’”, “the credit card repayment amount is greater than or equal to 1 yuan and the repayment state is ‘accounted’”, etc. The trigger delay can be, for example, triggered 3 seconds after the transaction is successful, avoiding pushing the questionnaire when the user has not completed the operation. The configuration value can be dynamically adjusted through the background management system. The associated business system identifier can be used to represent the business system triggered by the corresponding event, such as the financial system through IDSYS_FIN identifier, the credit card system through IDSYS_CRD identifier, etc.

[0045] In a specific implementation, the distributed event bus is combined with the message queue architecture to realize the process of real-time monitoring as follows: the business system implants event markers at key transaction nodes (such as a financial redemption confirmation interface and a credit card repayment callback interface) to mark the triggering method of the questionnaire triggering event; after the event is serialized, it is sent to the specified Topic of the Kafka message queue; the questionnaire system deploys a consumer instance, subscribes to the target Topic, consumes events through a partition load balancing strategy, and stores the consumption offset in Redis; if the consumption fails (such as network interruption or message format error), a retry mechanism can be triggered: the first retry is delayed for 1 second, the subsequent retry intervals are increased by 2 seconds (n is the number of retries), the maximum number of retries is 5, and if the retry fails, the message is transferred to the dead letter queue for manual verification by the operation and maintenance personnel. n

[0046] After monitoring the questionnaire triggering event, the event log corresponding to the event can be obtained. Specifically, the questionnaire system provides a corresponding log receiving interface and realizes the acquisition and reception of the event log through the interface. After the interface receives the log, the event log can be subjected to preliminary data processing, which can include data cleaning and log storage. Data cleaning can include deduplication, completion, format verification, etc.; log storage can include real-time caching and persistent storage. The specific data cleaning process and log storage process can be selected and designed based on application requirements, and the present specification does not limit this.

[0047] Reference Figure 2 As shown in the accompanying drawings, in some embodiments of the present specification, the questionnaire knowledge graph can be constructed in the following manner: S201: Obtain multi-source heterogeneous data for constructing the questionnaire knowledge graph, wherein the multi-source heterogeneous data at least includes historical business data, historical questionnaire data, customer service data, a preset questionnaire knowledge base, and user behavior data; S201: Process the historical business data to extract key business elements related to business logic as the business element entities; S203: Process the historical questionnaire data, the customer service data, and the preset questionnaire knowledge base to extract the questionnaire intention entities; S204: Process the historical business data, the user behavior data, and the historical questionnaire data to determine the association relationship between the business element entities and the questionnaire intention entities, and the relationship attribute data corresponding to each association relationship; S205: Construct the questionnaire knowledge graph based on the business element entities, the questionnaire intention entities, the association relationship, and the relationship attribute data corresponding to each association relationship.

[0048] ​It can be understood that the historical business data can include historical transaction data and historical business logs of the business system. The multi-source heterogeneous data can include structured data and unstructured data. Further, when processing the historical business data and extracting business element entities, it can include processing of structured data and processing of unstructured data. For structured data, fields therein can be directly extracted, and business element entities therein can be determined based on the association between the fields and business logic; for unstructured data, it can be analyzed and processed through natural language processing (NLP), and elements related to business logic therein can be extracted as business element entities.

[0049] In some embodiments of the present specification, the fields or elements related to business logic can include products, services, user journey nodes, user label portraits, channels, and the like, and further, the business element entities extracted based on the multi-source heterogeneous data can include at least one of the following: product entities, service entities, user journey node entities, user label entities, and channel entities.

[0050] Further, for the extraction processes of the product entities, service entities, user journey node entities, user label entities, and channel entities, the following can be shown: for structured historical business data, product IDs, names, types, risk levels, and the like fields can be directly extracted from database tables, and through predefined mapping rules, “product entities” can be automatically created and filled; for unstructured historical business data, optical character recognition (OCR) and natural language processing (NLP) techniques can be used to identify product names, yields, terms, and the like key information in the text, and associate them with existing product entities or create new entities; for historical business logs and user behavior data, common and key interaction steps (such as “login -> browse product details -> input purchase amount -> confirm transaction” and the like) can be identified by analyzing user behavior sequences, the identified key interaction steps can be abstracted as user journey node entities, and the business flowchart of the business system can be parsed, and the “task” or “activity” nodes in the chart can be automatically converted into user journey node entities.

[0051] For the questionnaire intention entity, the intention list (such as "fee dispute", "process complexity complaint") can be initially formed by using a topic model (such as LDA) and a keyword extraction algorithm to automatically cluster and discover common customer concerns and business pain points based on the obtained historical questionnaire data, customer service data and a preset questionnaire knowledge base, and then the questionnaire intention entity can be obtained by deduplication and standardized naming of the extracted intention list. Further, the questionnaire intention entity extracted can be further vectorized, for example, the text description of each intention is converted into an intention vector by using a natural language model, and is stored in the entity attribute for subsequent semantic matching.

[0052] In some embodiments of the present specification, in addition to the need to extract entities in multi-source heterogeneous data, the association relationship between entities also needs to be extracted and analyzed. Specifically, relationship reasoning can be performed based on pre-defined business logic rules, and association rule mining can be performed based on historical user behavior data and questionnaire result data to obtain the association relationship between entities. For example, when performing association rule mining, mutual information can be used to calculate the correlation between entities, it can be determined that there is an association relationship between entities when the correlation between entities meets a preset threshold, and the calculated correlation can be used as the relationship attribute data of the entity.

[0053] In some embodiments of the present specification, the association relationship between the extracted entities can include at least one of the following: a first association relationship between the user journey node entity and the questionnaire intention entity, a second association relationship between the product entity and the questionnaire intention entity, a third association relationship between the service entity and the questionnaire intention entity, a fourth association relationship between the product entity and the service entity, a fifth association relationship between the product entity and the user journey node entity, a sixth association relationship between the service entity and the user journey node entity, and the like. Further, for some entities, there are preconditions between them, for example, for the user journey node entity, there is a process sequence between at least some entities, for example, the user journey node A needs to be executed before the user journey node B, therefore, when extracting the association relationship between entities, the extraction of the association relationship between different types of entities can be included, and the extraction of the association relationship between the same type of entities can also be included.

[0054] In some embodiments of the present specification, the association relationship between entities can be further quantified by relationship attribute data, which can include at least one of the following: relationship weight, relationship type. The relationship weight can be used to represent the strength of the association relationship between entities, and the relationship type can represent the type of association between entities, which can be used for subsequent mapping of target business elements and determination of target questionnaire intention entities.

[0055] Further, the questionnaire knowledge graph can be dynamically updated by real-time acquisition of multi-source heterogeneous data. Specifically, the completion result data corresponding to the historically generated questionnaire in the system can be acquired, which can include questionnaire completion rate, user answering duration, and open question text correlation analysis result, etc. The relationship weight is dynamically adjusted based on the completion result data. In other embodiments, the entity nodes in the questionnaire knowledge graph can also be dynamically updated by the acquired multi-source heterogeneous data, which is not limited in the present specification.

[0056] In some embodiments of the present specification, based on the business element entity, the questionnaire intention entity, the association relationship, and the relationship attribute data corresponding to each association relationship, the questionnaire knowledge graph can be constructed, which can include: vectorizing the questionnaire intention entity to obtain a questionnaire intention vector; taking the determined business element entity and the questionnaire intention vector as nodes, taking the association relationship as an edge connecting between the nodes, and taking the relationship attribute data corresponding to each association relationship as a parameter of each edge to construct the questionnaire knowledge graph.

[0057] Specifically, the constructed questionnaire knowledge graph can be displayed to the user in the form of a visual graph. In the visual display, the business element entity and the questionnaire intention vector can be taken as nodes, the association relationship between entities can be taken as an edge connecting the nodes, and the attribute parameters of each edge can be configured based on the relationship attribute data. The constructed questionnaire knowledge graph can also be stored in a graph database for system to call the questionnaire knowledge graph for mapping of the target business element.

[0058] In the embodiments of the present specification, through processing of multi-source heterogeneous data, the efficiency and accuracy of graph construction can be realized. Through the online learning mechanism based on real user feedback, the questionnaire knowledge graph can be automatically and dynamically updated by self-adaptive adjustment and continuous optimization of the questionnaire knowledge graph according to the business development and user behavior changes.

[0059] Reference Figure 3 As shown in some embodiments of the present specification, based on the target business element, the target questionnaire intention entity corresponding to the questionnaire trigger event can be determined in the preset questionnaire knowledge graph, which can include: S301: generating a target query vector based on the target business element; S302: determining a target node mapped by the target query vector in the questionnaire knowledge graph; S303: determining the target questionnaire intention entity based on the questionnaire intention entity associated with the target node and the relationship attribute data between the target node and each associated questionnaire intention entity.

[0060] It can be understood that when determining the target questionnaire intention entity in the questionnaire knowledge graph, the target business element can be first converted into a target query vector, so as to facilitate the mapping of the target business element to the node based on the target query vector, and then the corresponding target node can be determined based on the mapping result, and the target questionnaire intention entity can be determined based on the nodes having connection edges with the target node and the attributes of each connection edge.

[0061] In specific implementation, when the target business element is determined, the target business element can be screened to select key features related to the type corresponding to the business element entity node in the questionnaire knowledge graph, so as to improve the data processing efficiency. The selected business elements can include business types, products corresponding to the business, current user journey nodes, channels triggering the event of generating the questionnaire, user labels corresponding to triggering the event, and the like. After screening the target business element, for text type business elements, they can be converted into vectors through coding techniques such as One-Hot coding, and for numerical type business elements, they can be converted into vector form through normalization processing. After vectorizing the target business element to obtain a plurality of vectors, the plurality of vectors can be spliced, and the splicing result can be used as a target query vector for vector mapping in the questionnaire knowledge graph. After generating the target query vector, the target query vector can be mapped to the target node in the questionnaire knowledge graph through similarity matching, and through the edges connected to the target node and the parameters of each edge, the target questionnaire intention entity corresponding to the current business scenario can be further determined.

[0062] In specific implementation, after the target node is determined, the questionnaire intention entity nodes having connection edges with the target node (which can be used as candidate intention entities) and the relationship attribute data corresponding to the connection edges can be queried, and a preset number of questionnaire intention entities can be selected as the target questionnaire intention entities based on the relationship weights corresponding to each candidate intention entity; or the questionnaire intention entities meeting a preset weight threshold can be selected as the target questionnaire intention entities. Further, after the questionnaire intention entities meeting the rule conditions are selected based on the preset rules (the preset number, the preset weight threshold, and the like), the questionnaire intention entities meeting the preset rules can be further screened based on the user labels to obtain the target questionnaire intention entities. For example, the user label is “old users”, and the relevance of “redemption process convenience survey” is higher (old users are more sensitive to operation complexity), and finally “replacement product preference collection” and “income gap reason investigation” are determined as the target questionnaire intention entities.

[0063] In some embodiments of the present specification, determining the target node to which the target query vector is mapped in the questionnaire knowledge graph can include: performing similarity matching between the target query vector and the nodes in the questionnaire knowledge graph to obtain a plurality of candidate nodes; performing logical verification on the plurality of candidate nodes based on the business processing logic corresponding to the target business element, and screening the target node from the plurality of candidate nodes based on the verification result.

[0064] Specifically, the feature vector corresponding to each node can be extracted from the questionnaire knowledge graph, the cosine similarity between the feature vector of each node and the target query vector can be calculated, and the node whose similarity meets the preset similarity threshold can be taken as the plurality of candidate nodes. After determining the candidate nodes, logical verification can be performed on each candidate node. The process of logical verification can be set based on the business logic rules of the business system. Based on the logical verification result, the target node can be screened. For example, the retrieved "execute redemption" node must have an indirect connection established through an association relationship with the retrieved "wealth product stable A number" node. Through logical verification, the logical consistency of the mapping result can be ensured, and the target node can be finally determined. The process of logical verification can be realized through a pre-designed rule engine. The rule engine can define business logic rules for a plurality of business scenarios, and thus when performing logical verification, the rule engine can be executed to screen the valid candidate nodes as the target node.

[0065] Through the above method, an unstructured business event log can be automatically parsed and mapped to one or more nodes with explicit semantics in the questionnaire knowledge graph, thereby providing accurate input for subsequent determination of a target questionnaire intention entity and questionnaire generation.

[0066] Reference Figure 4 In some embodiments of the present specification, determining the target questionnaire element matching the target questionnaire intention entity in the preset questionnaire element library, and generating an initial survey questionnaire based on the target questionnaire element can include: S401: calculating the similarity between the target questionnaire intention entity and each questionnaire element in the questionnaire element library, and screening a questionnaire element meeting a preset similarity as a target questionnaire element; S402: sorting the target questionnaire elements based on the logical dependency relationship between the target questionnaire elements, and performing logical relationship embedding on the sorting result based on the logical dependency relationship to obtain a questionnaire element sequence; S403: generating the initial survey questionnaire based on the questionnaire element sequence and the target business element.

[0067] It can be understood that a large number of question atoms can be stored in the questionnaire element library, the question atoms can be questions that users are more concerned about in multiple business scenarios or related to system and service optimization, and the target questionnaire intention entity is a question that a user is more concerned about and related to system and service optimization determined for the current business scenario. Then, the target questionnaire elements used to generate the initial survey questionnaire can be determined by matching.

[0068] Specifically, the questionnaire element library can at least include the following elements: question elements (for example, question atoms), option elements, logic control elements, and the like. Among them, the question elements can include but are not limited to: question identifier, question type, question content, difficulty level, whether to fill in, creation time, update time, and the like; the option elements can include but are not limited to: option identifier, associated question (which can be represented by an associated question identifier), option content, option value, sorting number, and the like; the logic control elements can include but are not limited to: control identifier, trigger type (for example, option trigger, question trigger), trigger identifier, action type (for example, jump / display / hide, etc.), target question identifier (i.e., the question identifier to be jumped to), trigger condition expression, and the like.

[0069] When matching the target questionnaire intention entity, the target questionnaire intention entity and the question content in the question element can be semantically matched to determine whether each question element is a question to be investigated for the corresponding target questionnaire intention entity. After determining a plurality of target question elements, the option elements associated with each target question element can be determined, the target question element and the corresponding option element can be used as a target questionnaire element, and then the plurality of target questionnaire elements can be sorted, and the logic control elements associated with the target question elements and / or the option elements can be embedded in the plurality of sorted target questionnaire elements to obtain a questionnaire element sequence.

[0070] In some embodiments of the present specification, adjusting the initial survey questionnaire based on the interaction environment information to generate a target survey questionnaire can include: determining a target adjustment parameter corresponding to the initial survey questionnaire based on the interaction environment information, the target adjustment parameter including at least one of the following: rendering adjustment parameter, interaction adjustment parameter, content adjustment parameter, format adjustment parameter; adjusting the initial survey questionnaire based on the target adjustment parameter to obtain the target survey questionnaire.

[0071] Specifically, the rendering adjustment parameter can adjust the rendering style, such as adjusting the layout (single column / multi-column), font size, element spacing according to the screen size; the interaction adjustment parameter can adjust the interaction mode, such as selecting the input mode (touch large button, keyboard navigation, voice input) according to the device support condition, etc., the content adjustment parameter can adjust the content length, such as dynamically compressing the problem quantity or simplifying the complex question type (such as splitting the matrix question into multiple single-choice questions) when the network condition is poor or the device performance is low; the format adjustment parameter can adjust the format of the output survey questionnaire to adapt to different channels, such as different channels can correspond to different adaptation interfaces, and then the format of the survey questionnaire can be adjusted based on different interfaces, such as providing sdk for mobile terminals (app, applet), providing H5 embedded interface for PC terminal, and then the format of the survey questionnaire can be converted based on the receiving standards of each interface.

[0072] In some embodiments of the present specification, based on the interaction environment information, determining the target adjustment parameter corresponding to the initial survey questionnaire can include: determining the user portrait label of the target user based on the user behavior data; normalizing the user portrait label, the target channel, the first target device and the device environment information to obtain an interaction environment vector; inputting the interaction environment vector into a preset adjustment strategy model to output the target adjustment parameter.

[0073] It can be understood that adjusting the initial survey questionnaire based on the user portrait label can make the generated target survey questionnaire better adapt to the user preference, and adjusting the initial survey questionnaire based on the target channel, the first electronic device and the device environment information, etc. can make the generated target survey questionnaire better adapt to the electronic device to be pushed. The normalization processing can include normalization of text information and normalization of numerical value information. For text information, normalization can be realized by mapping the text into a vector form, and for numerical value information, normalization can be realized by maximum value, minimum value, etc.

[0074] It can be understood that the adjustment strategy model can make a decision on the adjustment parameter based on the input interaction environment vector to obtain the target adjustment parameter.

[0075] Specifically, the adjustment strategy model can be trained in the following ways: (1) Model selection: Gradient Boosting Decision Tree (e.g., XGBoost) can be used as the adjustment strategy model. Gradient Boosting Decision Tree can support classification and regression tasks, can output multiple adjustment parameters (such as rendering parameters, content parameters, etc.) at the same time, and can be robust to missing values without complex data preprocessing; in addition, feature importance analysis can be used to determine key influencing factors (such as the influence weight of device type on font size). (2) Data set construction: training data can be collected from historical data of questionnaire surveys in multiple business scenarios within a preset time period, each training data can include an interaction environment vector as an input feature and an adjustment parameter (such as font size: 24, number of questions: 3, interaction mode: voice + touch) as an output label; then the training data can be divided into a training set, a validation set, and a test set. (3) Model training: parameter configuration can be performed first, including learning rate (learning_rate), tree depth (max_depth), tree number (n_estimators), and loss function: multi: softmax, etc. The training set model is trained, and the MAE (mean absolute error) of the validation set is monitored. When the MAE does not decrease for n consecutive rounds (e.g., 5 rounds), the training is stopped to prevent overfitting. Model evaluation can be performed using the test set, and the MAE (parameter prediction error) and accuracy of the model on the test set can be set.

[0076] After the adjustment strategy model is trained, the normalized interaction environment vector can be input into the model, the trained adjustment strategy model is loaded, and the target adjustment parameter is output.

[0077] In some embodiments of the present specification, the generated target survey questionnaire can also synchronize data on multiple channels or multiple devices, and the target survey questionnaire and the current filling progress can be synchronized to another device when the user switches devices. That is, the above method can further include: constructing a questionnaire identifier corresponding to the target survey questionnaire; acquiring questionnaire update data of the target survey questionnaire in real time; in response to the operation of the target user on the second target device, acquiring the current interaction environment corresponding to the second target device; adjusting the target survey questionnaire based on the current interaction environment and the questionnaire update data, and pushing the adjusted survey questionnaire to the second target device.

[0078] Specifically, when adjusting the target survey questionnaire, changes in the current interaction environment can be determined, updated environment information can be determined, and new adjustment parameters corresponding to the target survey questionnaire can be determined based on the updated environment information. The target survey questionnaire can be adjusted based on the newly determined adjustment parameters to push the adjusted survey questionnaire to the second target device.

[0079] In the embodiments of the present specification, the association between the business elements and the questionnaire intention is established through the questionnaire knowledge graph, and the real-time business scene is analyzed in combination with the NLP technology, so as to realize the accurate matching of the questionnaire content and the user operation. Compared with the fixed template questionnaire in the prior art, the questionnaire content has higher compatibility, the real needs of the user can be effectively mined, and the accurate adaptability of the questionnaire and the business scene can be improved. Based on the channel characteristic database and the interaction environment information, the style, the interaction mode and the content complexity of the questionnaire can be automatically adjusted (such as large font and voice interaction for the elderly machine, double-column layout and detailed questions for the PC terminal), so that the problem of large difference in the questionnaire style and poor adaptability in the prior art can be solved, the multi-channel adaptive adaptation can be realized, and the user experience can be improved. The unified template library and the channel adaptation interface are adopted to avoid independent development of the questionnaire system by each channel, reduce repeated investment, and at the same time, the data real-time synchronization and unified analysis function reduce the data integration cost and improve the utilization efficiency of the feedback data, so that the development and operation cost can be reduced. Through the questionnaire identification and real-time monitoring of the interaction environment, the questionnaire synchronization and adaptive adjustment of the user when switching between different devices (such as mobile banking and personal online banking) are realized, repeated filling is avoided, the coherence and convenience of the questionnaire filling are improved, cross-device seamless switching can be supported, and the user experience is optimized. Based on the user portrait and the historical data, the personalized questionnaire is generated, and in combination with the unified data collection of multiple channels, the feedback opinions and the satisfaction scores of the user on the same function in different channels can be comprehensively and accurately collected, data support is provided for the bank product optimization and service upgrade, and finally the product experience level is improved.

[0080] Based on the above-described multi-channel questionnaire generation method, one or more embodiments of the present specification also provide a multi-channel questionnaire generation device. The device can include a device (including a distributed system), software (application), module, plug-in, server, client, etc. using the method described in the embodiments of the present specification, and a device combined with necessary implementation hardware. Based on the same innovative concept, the device in one or more embodiments of the present specification is described in the following embodiments. Since the implementation scheme of the device to solve the problem is similar to the method, the implementation of the specific device in the embodiments of the present specification can refer to the implementation of the foregoing method, and the repeated parts will not be described herein. The term "unit" or "module" used below can be a combination of software and / or hardware that realizes a predetermined function. Although the device described in the following embodiments is preferably realized in software, the realization of hardware or a combination of software and hardware is also possible and conceived. Figure 5 A schematic diagram of a multi-channel questionnaire generation device provided by the embodiments of the present specification is shown. As shown in the figure, Figure 5 As shown, the multi-channel questionnaire generation device 500 can include.

[0081] The acquisition module 501 is configured to acquire an event log corresponding to the questionnaire triggering event when the questionnaire triggering event is monitored. The determination module 502 is configured to parse a target business element contained in the event log, and determine a target questionnaire intention entity corresponding to the questionnaire triggering event in a preset questionnaire knowledge graph based on the target business element. The questionnaire knowledge graph is composed of a plurality of entities and association relationships between the entities. The plurality of entities at least include a questionnaire intention entity and a business element entity. The generation module 503 is configured to determine a target questionnaire element matching the target questionnaire intention entity in a preset questionnaire element library, and generate an initial survey questionnaire based on the target questionnaire element. The adjustment module 504 is configured to acquire interaction environment information corresponding to the questionnaire triggering event, and adjust the initial survey questionnaire based on the interaction environment information to generate a target survey questionnaire. The interaction environment information includes at least one of the following: a target channel corresponding to the target survey questionnaire, a first target device, device environment information, and user behavior data of a target user corresponding to the target survey questionnaire.

[0082] In some embodiments of the present specification, the questionnaire knowledge graph is constructed by: acquiring multi-source heterogeneous data for constructing the questionnaire knowledge graph, the multi-source heterogeneous data at least including historical business data, historical questionnaire data, customer service data, a preset questionnaire knowledge base, and user behavior data; processing the historical business data to extract key business elements related to business logic as the business element entity; processing the historical questionnaire data, the customer service data, and the preset questionnaire knowledge base to extract the questionnaire intention entity; processing the historical business data, the user behavior data, and the historical questionnaire data to determine the association relationship between the business element entity and the questionnaire intention entity, and relationship attribute data corresponding to each association relationship; and constructing the questionnaire knowledge graph based on the business element entity, the questionnaire intention entity, the association relationship, and the relationship attribute data corresponding to each association relationship.

[0083] In some embodiments of the present specification, the business element entity includes at least one of the following: a product entity, a service entity, a user journey node entity, a user label entity, and a channel entity. The relationship attribute data includes at least one of the following: a relationship weight and a relationship type.

[0084] In some embodiments of the present specification, based on the business element entity, the questionnaire intention entity, the association relationship, and the relationship attribute data corresponding to each association relationship, constructing the questionnaire knowledge graph can include: vectorizing the questionnaire intention entity to obtain a questionnaire intention vector; taking the determined business element entity and the questionnaire intention vector as nodes, taking the association relationship as an edge connecting between nodes, and taking the relationship attribute data corresponding to each association relationship as a parameter of each edge to construct the questionnaire knowledge graph.

[0085] In some embodiments of the present specification, the determining module 502 determines the target questionnaire intention entity corresponding to the questionnaire trigger event in the preset questionnaire knowledge graph based on the target business element, and specifically can be used for: generating a target query vector based on the target business element; determining a target node mapped by the target query vector in the questionnaire knowledge graph; determining the target questionnaire intention entity based on the questionnaire intention entity associated with the target node and the relationship attribute data between the target node and each associated questionnaire intention entity.

[0086] In some embodiments of the present specification, the determining module 502 determines the target node mapped by the target query vector in the questionnaire knowledge graph, and specifically can be used for: performing similarity matching between the target query vector and the nodes in the questionnaire knowledge graph to obtain a plurality of candidate nodes; performing logical verification on the plurality of candidate nodes based on the business processing logic corresponding to the target business element, and screening the target node from the plurality of candidate nodes based on the verification result.

[0087] In some embodiments of the present specification, the generating module 503 determines the target questionnaire element matching the target questionnaire intention entity in the preset questionnaire element library, and generates an initial survey questionnaire based on the target questionnaire element, and specifically can be used for: calculating the similarity between the target questionnaire intention entity and each questionnaire element in the questionnaire element library, and screening the questionnaire element satisfying the preset similarity as the target questionnaire element; sorting the target questionnaire element based on the logical dependency relationship between the target questionnaire elements, and embedding the logical relationship in the sorting result based on the logical dependency relationship to obtain a questionnaire element sequence; generating the initial survey questionnaire based on the questionnaire element sequence and the target business element.

[0088] In some embodiments of this specification, when the adjustment module 504 adjusts the initial questionnaire based on the interaction environment information to generate a target questionnaire, it may specifically be used to: determine the target adjustment parameters corresponding to the initial questionnaire based on the interaction environment information, wherein the target adjustment parameters include at least one of the following: rendering adjustment parameters, interaction adjustment parameters, content adjustment parameters, and format adjustment parameters; and adjust the initial questionnaire based on the target adjustment parameters to obtain the target questionnaire.

[0089] In some embodiments of this specification, when the adjustment module 504 determines the target adjustment parameters corresponding to the initial questionnaire based on the interaction environment information, it may specifically be used to: determine the user profile tags of the target user based on the user behavior data; normalize the user profile tags, target channels, first target devices, and device environment information to obtain an interaction environment vector; input the interaction environment vector into a preset adjustment strategy model, and output the target adjustment parameters.

[0090] In some embodiments of this specification, the above-described apparatus may also be used to: construct a questionnaire identifier corresponding to the target questionnaire; acquire questionnaire update data of the target questionnaire in real time; in response to the target user's operation on the second target device, acquire the current interaction environment corresponding to the second target device; adjust the target questionnaire based on the current interaction environment and the questionnaire update data, and push the adjusted questionnaire to the second target device.

[0091] The descriptions and functions of the above modules can be found in the section on multi-channel questionnaire generation methods, and will not be repeated here.

[0092] This application also provides an electronic device, such as... Figure 6 As shown, the electronic device 600 may include a processor 601 and a memory 602, wherein the processor 601 and the memory 602 may be connected via a bus 603 or other means. Figure 6 Taking the connection via bus 603 as an example.

[0093] In some embodiments of the present specification, the processor 601 can be a central processing unit (CPU). In other embodiments, the processor 601 can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic components, discrete hardware components, or a combination thereof.

[0094] The memory 602, as a kind of non-transient computer readable storage medium, can be used to store non-transient software programs, non-transient computer executable programs and modules, such as the program instructions / modules corresponding to the multi-channel questionnaire generation method in the embodiments of the present application. The processor 601 performs various functional applications and data processing of the processor by running the non-transient software programs, instructions and modules stored in the memory 602, that is, implements the multi-channel questionnaire generation method in the above method embodiments.

[0095] The memory 602 can include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function; the data storage area can store data created by the processor 601 and the like. In addition, the memory 602 can include a high-speed random access memory, and can also include a non-transient memory, such as at least one magnetic disk storage device, a flash memory device, or other non-transient solid-state memory device. In some embodiments, the memory 602 can optionally include a memory remotely arranged with respect to the processor 601, which can be connected to the processor 601 through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0096] The one or more modules are stored in the memory 602, and when executed by the processor 601, the following multi-channel questionnaire generation method is performed: In response to monitoring a questionnaire triggering event, an event log corresponding to the questionnaire triggering event is obtained; a target business element included in the event log is parsed, and a target questionnaire intention entity corresponding to the questionnaire triggering event is determined in a preset questionnaire knowledge graph based on the target business element, the questionnaire knowledge graph being composed of a plurality of entities and association relationships between the entities, and the plurality of entities including at least a questionnaire intention entity and a business element entity; a target questionnaire element matching the target questionnaire intention entity is determined in a preset questionnaire element library, and an initial survey questionnaire is generated based on the target questionnaire element; interaction environment information corresponding to the questionnaire triggering event is obtained, and the initial survey questionnaire is adjusted based on the interaction environment information to generate a target survey questionnaire; and the interaction environment information includes at least one of the following: a target channel corresponding to the target survey questionnaire, a first target device, device environment information, and user behavior data of a target user corresponding to the target survey questionnaire.

[0097] The above electronic device 600 can be understood in detail by referring to the corresponding descriptions and effects of the above method embodiments, and will not be described again here.

[0098] The present specification also provides a computer storage medium storing computer program instructions, which, when executed, implement the steps of the above multi-channel questionnaire generation method.

[0099] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the program can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD) or a solid state drive (SSD), etc. The storage medium can also include a combination of the above types of memories.

[0100] Each embodiment in the present specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other. Each embodiment mainly describes the differences from other embodiments.

[0101] The system, device, module or unit described in the above embodiments can be specifically implemented by a computer chip or entity, or by a product with certain functions.

[0102] For ease of description, the above apparatus is described as various units respectively in function for separate description. Of course, functions of the units can be implemented in one or more software and / or hardware in implementing the present application.

[0103] From the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented by means of software plus necessary universal hardware platforms. Based on such an understanding, the technical solutions of the present application can be embodied in the form of a software product, and the computer software product can be stored in a storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a plurality of instructions to cause a computer device (which can be a personal computer, a server, or a network device, etc.) to execute methods of some parts of the embodiments of the present application.

[0104] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, and the like.

[0105] The present application can be described in the general context of computer-executable instructions, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, and the like, which perform particular tasks or implement particular abstract data types. The present application can also be practiced in distributed computing environments where tasks are performed by remote processing devices that are connected through a communication network. In a distributed computing environment, program modules can be located in both local and remote computer storage media including storage devices.

[0106] Although the present application is depicted through embodiments, those skilled in the art know that the present application has many variations and changes without departing from the spirit of the present application, and it is intended that the appended claims include these variations and changes without departing from the spirit of the present application.

Claims

1. A multi-channel questionnaire generation method, characterized by, The method comprises the following steps: monitoring a questionnaire triggering event, obtaining an event log corresponding to the questionnaire triggering event; parsing a target business element contained in the event log, and determining a target questionnaire intention entity corresponding to the questionnaire triggering event in a preset questionnaire knowledge graph based on the target business element, the questionnaire knowledge graph being composed of a plurality of entities and association relationships between the entities, the plurality of entities at least including a questionnaire intention entity and a business element entity; determining a target questionnaire element matching the target questionnaire intention entity in a preset questionnaire element library, and generating an initial survey questionnaire based on the target questionnaire element; obtaining interaction environment information corresponding to the questionnaire triggering event, and adjusting the initial survey questionnaire based on the interaction environment information to generate a target survey questionnaire; wherein the interaction environment information includes at least one of the following: a target channel corresponding to the target survey questionnaire, a first target device, device environment information, and user behavior data of a target user corresponding to the target survey questionnaire.

2. The multi-channel questionnaire generation method according to claim 1, wherein, The questionnaire knowledge graph is constructed by the following method: obtaining multi-source heterogeneous data for constructing the questionnaire knowledge graph, the multi-source heterogeneous data at least including historical business data, historical questionnaire data, customer service data, a preset questionnaire knowledge base, and user behavior data; processing the historical business data to extract key business elements related to business logic as the business element entity; processing the historical questionnaire data, the customer service data, and the preset questionnaire knowledge base to extract the questionnaire intention entity; processing the historical business data, the user behavior data, and the historical questionnaire data to determine association relationships between the business element entity and the questionnaire intention entity, and relationship attribute data corresponding to each association relationship; constructing the questionnaire knowledge graph based on the business element entity, the questionnaire intention entity, the association relationships, and the relationship attribute data corresponding to each association relationship.

3. The multi-channel questionnaire generation method according to claim 2, wherein The business element entity includes at least one of the following: a product entity, a service entity, a user journey node entity, a user tag entity, and a channel entity; The relationship attribute data includes at least one of the following: relationship weight and relationship type.

4. The multi-channel questionnaire generation method of claim 2, wherein, Constructing the questionnaire knowledge graph based on the business element entity, the questionnaire intention entity, the association relationships, and the relationship attribute data corresponding to each association relationship comprises: vectorizing the questionnaire intention entity to obtain a questionnaire intention vector; constructing the questionnaire knowledge graph by taking the determined business element entity and the questionnaire intention vector as nodes, the association relationships as edges connecting the nodes, and the relationship attribute data corresponding to each association relationship as parameters of each edge.

5. The multi-channel questionnaire generation method of claim 1, wherein, Determining a target questionnaire intention entity corresponding to the questionnaire triggering event in a preset questionnaire knowledge graph based on a target business element comprises: generating a target query vector based on the target business element; determining a target node mapped by the target query vector in the questionnaire knowledge graph; Determine the target questionnaire intention entity based on the questionnaire intention entity associated with the target node and the relationship attribute data between the target node and each associated questionnaire intention entity.

6. The multi-channel questionnaire generation method according to claim 5, wherein, Determine the target node to which the target query vector is mapped in the questionnaire knowledge graph, including: Match the target query vector with the nodes in the questionnaire knowledge graph based on similarity to obtain a plurality of candidate nodes; Based on the business processing logic corresponding to the target business element, perform logical verification on the plurality of candidate nodes, and based on the verification result, filter the target node from the plurality of candidate nodes.

7. The multi-channel questionnaire generation method of claim 1, wherein, Determine the target questionnaire element matching the target questionnaire intention entity in the preset questionnaire element library, and generate an initial survey questionnaire based on the target questionnaire element, including: Calculate the similarity between the target questionnaire intention entity and each questionnaire element in the questionnaire element library, and filter out the questionnaire element that meets the preset similarity as the target questionnaire element; Based on the logical dependency relationship between the target questionnaire elements, sort the target questionnaire elements, and embed the logical relationship in the sorting result based on the logical dependency relationship to obtain a questionnaire element sequence; Based on the questionnaire element sequence and the target business element, generate the initial survey questionnaire.

8. The multi-channel questionnaire generation method of claim 1, wherein, Adjust the initial survey questionnaire based on the interaction environment information to generate a target survey questionnaire, including: Based on the interaction environment information, determine the target adjustment parameter corresponding to the initial survey questionnaire, the target adjustment parameter including at least one of the following: rendering adjustment parameter, interaction adjustment parameter, content adjustment parameter, format adjustment parameter; Based on the target adjustment parameter, adjust the initial survey questionnaire to obtain the target survey questionnaire.

9. The multi-channel questionnaire generation method of claim 8, wherein, Based on the interaction environment information, determine the target adjustment parameter corresponding to the initial survey questionnaire, including: Based on the user behavior data, determine the user portrait label of the target user; Normalize the user portrait label, target channel, first target device, and device environment information to obtain an interaction environment vector; Input the interaction environment vector into a preset adjustment strategy model to output the target adjustment parameter.

10. The multi-channel questionnaire generation method of claim 1, wherein, Further comprising: Construct a questionnaire identifier corresponding to the target survey questionnaire; Real-time acquisition of questionnaire update data of the target survey questionnaire; In response to the operation of the target user on the second target device, acquire the current interaction environment corresponding to the second target device; Based on the current interaction environment and the questionnaire update data, adjust the target survey questionnaire, and push the adjusted survey questionnaire to the second target device.

11. A multi-channel questionnaire generation apparatus characterized by comprising: Comprising: An acquisition module configured to, when a questionnaire triggering event is monitored, acquire an event log corresponding to the questionnaire triggering event; A determination module configured to analyze a target business element included in the event log, and determine a target questionnaire intention entity corresponding to the questionnaire triggering event in a preset questionnaire knowledge graph based on the target business element, the questionnaire knowledge graph being composed of a plurality of entities and association relationships between the entities, the plurality of entities including at least a questionnaire intention entity and a business element entity; The generating module is configured to determine a target questionnaire element matched with the target questionnaire intention entity from a preset questionnaire element library, and generate an initial survey questionnaire based on the target questionnaire element; The adjusting module is configured to acquire interaction environment information corresponding to a questionnaire triggering event, and adjust the initial survey questionnaire based on the interaction environment information to generate a target survey questionnaire; wherein the interaction environment information includes at least one of the following: a target channel corresponding to the target survey questionnaire, a first target device, device environment information, and user behavior data of a target user corresponding to the target survey questionnaire.

12. An electronic device, comprising: The method comprises: A memory and a processor, which are in communication connection with each other, the memory stores computer instructions, and the processor implements the steps of the method according to any one of claims 1 to 10 by executing the computer instructions.

13. A computer storage medium, characterized in that The computer storage medium stores computer program instructions, and the computer program instructions are executed to implement the steps of the method according to any one of claims 1 to 10.

14. A computer program product, characterised in that, The computer program is executed by the processor to implement the steps of the method according to any one of claims 1 to 10.

Images