Dynamic form generation method and device, computer equipment and storage medium

Abstract

The invention belongs to the field of artificial intelligence and financial science and technology, and relates to a dynamic form generation method, which comprises the following steps: preprocessing unstructured text data requested by a client form to obtain preprocessed text data; inputting the preprocessed text data into an intention recognition model to obtain an intention category; form keywords are selected according to the weights, and the incidence relation of the form keywords is obtained; retrieving a candidate record set from historical query records according to the form keyword; calculating probability distribution of the candidate record set by adopting a maximum entropy model to obtain a prediction probability; according to the prediction probability, determining a candidate record most related to the client form request as a form generation reference knowledge base; and generating a personalized form based on the form generation reference knowledge base. The invention further provides a dynamic form generation device, computer equipment and a storage medium. In addition, the invention further relates to a block chain technology, and historical query records can be stored in a block chain. According to the invention, personalized services can be provided, the business process is simplified, and the error rate is reduced.

Description

Technical Field

[0001] The present application relates to the fields of artificial intelligence and financial technology, and in particular to a dynamic form generation method, apparatus, computer equipment and storage medium. Background Art

[0002] In the insurance field, the statistics and management of multi-dimensional information such as insurance application rates and claims processing are the core of insurance companies' daily operations. With the rapid development of information technology, more and more insurance companies have begun to build self-service platforms, aiming to provide more convenient and efficient insurance services through online channels. These platforms usually integrate basic online form filling functions and policy management functions, allowing customers to independently complete the initial submission of insurance applications and the query and management of policies.

[0003] Although the existing self-service platforms have improved the convenience of insurance services to a certain extent, they still have many limitations and shortcomings. Specifically, most of these platforms rely on predefined, static forms to collect customer information. When using these platforms, customers need to manually find and fill in all relevant information, which not only increases the learning cost of customers, but also requires customers to have a certain level of professionalism in order to accurately understand and fill in various contents. The design of static forms lacks intelligent inspection and real-time feedback mechanisms, which makes it easy for customers to make mistakes during the filling process, with a high error rate.

[0004] In addition, existing platforms also have obvious shortcomings in providing guidance and support. Although some platforms have integrated simple chatbots or basic customer support systems to answer customers' common questions, these systems are often limited to providing text instructions or simple conversational interactions. They lack multimedia guidance methods, such as illustrated instructions, video tutorials, etc., making it difficult for customers to obtain intuitive and comprehensive guidance when faced with complex insurance processes. At the same time, these systems also lack intelligent guidance functions, such as AI virtual assistants, and cannot provide personalized guidance and support based on the specific circumstances and needs of customers.

[0005] In summary, the existing self-service platforms in the insurance field have problems such as lack of flexibility, insufficient guidance and high error rate. Summary of the invention

[0006] The purpose of the embodiments of the present application is to propose a dynamic form generation method, device, computer equipment and storage medium to solve the technical problems of the prior art that static insurance policies lack flexibility, have a high error rate in the filling process, and lack guidance on the service platform.

[0007] In order to solve the above technical problems, the present application embodiment provides a dynamic form generation method, which adopts the following technical solution:

[0008] Receiving a customer form request, obtaining unstructured text data in the customer form request, and preprocessing the unstructured text data to obtain preprocessed text data;

[0009] Inputting the preprocessed text data into the trained intent recognition model to determine the intent category to which the customer form request belongs;

[0010] Calculating the weight of each word in the preprocessed text data, selecting form keywords from all the words according to the weight, and obtaining the association relationship of the form keywords;

[0011] Retrieving a first preset number of historical query records having the highest similarity to the customer form request from the historical query records according to the form keyword as a candidate record set;

[0012] Using a maximum entropy model, the probability distribution of the candidate record set is calculated according to the intent category, the form keyword and the association relationship, and a predicted probability of each candidate record is obtained;

[0013] Determine, according to the predicted probability, a second preset number of candidate records that are most relevant to the customer form request as a reference knowledge base for form generation;

[0014] Based on the form generation reference knowledge base, a personalized form that meets the customer form request is dynamically generated.

[0015] In order to solve the above technical problems, the embodiment of the present application also provides a dynamic form generation device, which adopts the following technical solution:

[0016] A preprocessing module, used for receiving a customer form request, obtaining unstructured text data in the customer form request, and preprocessing the unstructured text data to obtain preprocessed text data;

[0017] An intent recognition module, used to input the pre-processed text data into a trained intent recognition model to determine the intent category to which the customer form request belongs;

[0018] An extraction module, used to calculate the weight of each word in the preprocessed text data, select form keywords from all the words according to the weight, and obtain the association relationship of the form keywords;

[0019] A retrieval module, configured to retrieve, from the historical query records according to the form keyword, a first preset number of historical query records having the highest similarity to the customer form request as a candidate record set;

[0020] A prediction module, configured to use a maximum entropy model to calculate the probability distribution of the candidate record set according to the intent category, the form keyword and the association relationship, and obtain a predicted probability of each candidate record;

[0021] A construction module, configured to determine, according to the predicted probability, a second preset number of candidate records most relevant to the customer form request as a reference knowledge base for form generation;

[0022] A generation module is used to generate a reference knowledge base based on the form and dynamically generate a personalized form that meets the customer's form request.

[0023] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0024] The computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the dynamic form generation method described above when executing the computer-readable instructions.

[0025] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0026] The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the dynamic form generation method described above are implemented.

[0027] Compared with the prior art, this application has the following beneficial effects:

[0028] The present application provides a dynamic form generation method, which can eliminate inconsistencies in data and improve data quality by obtaining unstructured text data in a customer form request and preprocessing the unstructured text data; perform intent recognition on the preprocessed text data, accurately understand the customer's insurance needs and improve service efficiency; extract form keywords from the preprocessed text data, and search for a candidate record set with the highest similarity to the customer's form request based on the form keywords, thereby improving retrieval efficiency; by adopting a maximum entropy model, the probability distribution of the candidate record set is calculated according to the intent category, form keywords and association relationship, so as to accurately understand the customer's query intention, improve query efficiency, avoid indiscriminate processing of all candidate records, and optimize resource allocation; obtain a form generation reference knowledge base according to the predicted probability, and dynamically generate a personalized form based on the form generation reference knowledge base, so as to provide personalized services to customers, reduce the filling of unnecessary content by customers, and improve customer experience. At the same time, only the actual relevant information of the customer needs to be filled in, which reduces the error rate and improves the accuracy and completeness of the data. In addition, complex business processes are simplified and the efficiency of the overall business process is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0031] Figure 2 is a flow chart of an embodiment of a dynamic form generation method according to the present application;

[0032] Figure 3 is a structural schematic diagram of an embodiment of a dynamic form generating device according to the present application;

[0033] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0035] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0036] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0037] like Figure 1As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables.

[0038] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0039] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to a laptop computer 1011, a tablet computer 1012 or a mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0040] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0041] It should be noted that the dynamic form generation method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the dynamic form generation device is generally set in the server / terminal device.

[0042] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0043] Continue to refer Figure 2 , shows a flow chart of an embodiment of a dynamic form generation method according to the present application, comprising the following steps:

[0044] Step S201, receiving a customer form request, obtaining unstructured text data in the customer form request, and preprocessing the unstructured text data to obtain preprocessed text data.

[0045] Customer form request refers to the data and information request submitted by the customer to the system or service platform through the form interface, which is used for subsequent form query, form filling and policy management, etc.

[0046] In this embodiment, the electronic device (eg, Figure 1 The server / terminal device shown in the figure can receive the client form request through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future.

[0047] In this embodiment, unstructured text data in the customer form request is extracted, and the unstructured text data is preprocessed. The preprocessing includes cleaning, normalization, word segmentation and formatting. The cleaning includes removing irrelevant characters and special symbols, removing stop words, and filling or deleting missing values. The normalization process uses a normalization method to scale the data value to the range of [0,1] to obtain normalized text data, and uses word segmentation technology to segment the normalized text data, divide the text into independent word units, and perform structured conversion on all word units according to a preset data format to obtain structured preprocessed text data. For example, the preset data format can be a JSON structure or a key-value pair format, which is not limited here.

[0048] Step S202: input the preprocessed text data into the trained intent recognition model to determine the intent category to which the customer form request belongs.

[0049] In this embodiment, before the step of inputting the preprocessed text data into the trained intent recognition model, historical query record data is obtained, query text in the historical query record data is extracted, and the query text is preprocessed to obtain structured query text data; word embedding technology is used to convert the structured text data into word vector representation, and a query text feature vector is constructed; according to the intent label of the query text data, the query text feature vector is divided into different intent categories to form a training data set; an initial intent recognition model is constructed, including an embedding layer, an encoder network and an output layer, wherein the encoder network can adopt a BERT encoding structure, a bidirectional LSTM network structure or a Transform encoder structure, etc., to encode the query text feature vector and the intent label respectively; by training on the training data set, the model parameters of the initial intent recognition model are optimized so that the model can learn text feature patterns under different intents; during the training process, the encoder network introduces an attention mechanism to enhance the model's ability to capture keywords and key semantics and improve the accuracy of intent recognition; the trained initial intent recognition model is evaluated, the generalization ability of the model is verified by a test data set, and the model is tuned according to the evaluation results to obtain a final intent recognition model.

[0050] In some optional implementations of this embodiment, the step of inputting the preprocessed text data into the trained intent recognition model to determine the intent category to which the customer form request belongs includes:

[0051] Input the preprocessed text data into the trained intent recognition model, which includes an embedding layer, an encoder network, and an output layer;

[0052] The preprocessed text data is converted into vectors through the embedding layer to obtain text feature vectors;

[0053] The text feature vector is input into the encoder network for attention mechanism calculation to obtain the text semantic intent feature;

[0054] The output layer calculates the probability distribution of intent categories based on the text semantic intent features;

[0055] According to the probability distribution of intent categories, the intent category with the largest probability value is determined as the target intent category for the customer form request.

[0056] Specifically, the preprocessed text data is converted into a text feature vector that can be understood by the model through the embedding layer; the text feature vector is input into the encoder network, and semantic features are extracted through the attention mechanism to capture the text context and semantic relationship, and obtain the text semantic intent features; the intent category probability distribution is calculated according to the text semantic intent features through the output layer, and the intent category with the largest probability value is determined according to the intent category probability distribution as the target intent category of the customer form request, and subsequent process processing is performed according to the target intent category.

[0057] By identifying the intent of customer form requests through the intent recognition model, we can provide efficient and accurate intent judgment capabilities, improve the efficiency and accuracy of intent recognition, and thus increase the processing speed of customer requests.

[0058] Step S203, calculating the weight of each word in the preprocessed text data, selecting form keywords from all the words according to the weight, and obtaining the association relationship between the form keywords.

[0059] In this embodiment, the word frequency statistics method is used to calculate the number of occurrences of each word in the preprocessed text data corresponding to a single customer form request to obtain preliminary word frequency data; the inverse document frequency algorithm is used to calculate the distribution of each word in the preprocessed text data corresponding to all customer form requests to obtain the inverse document frequency value; the word frequency is multiplied by the inverse document frequency to obtain the TF-IDF weight of each word; the TF-IDF weights of all words are sorted, and the words with the highest weight and the preset number of words are selected as form keywords.

[0060] The TF-IDF weight of each word is calculated through the inverse document frequency algorithm to measure the importance of the word, and form keywords are selected based on the TF-IDF weight, which can improve the efficiency and accuracy of keyword extraction.

[0061] In some optional implementations, the extracted form keywords are matched with the preset business keyword library. If the match is successful, the form keyword is marked as a high-value keyword for subsequent business analysis and decision-making. For example, the form keyword of a customer form request is "product", which is successfully matched with the business keyword library and marked as a high-value keyword. In business analysis, it is found that the high-value keyword "product" appears frequently in customer form requests, indicating that customers pay high attention to the product. The business department can optimize product introduction and after-sales service based on this to improve customer satisfaction. This method enables more accurate grasp of customer needs, improve service quality, optimize business decisions, and ultimately achieve the dual goals of business growth and improved customer satisfaction.

[0062] In some optional implementations, a keyword extraction model is constructed based on a pre-trained GPT model, the GPT model includes multiple stacked connected Transformer decoders, the decoder includes a masked multi-head attention mechanism and a fully connected layer; historical query record data is obtained, query text in the historical query record data is extracted, and the query text is preprocessed to obtain structured query text data, and query keywords in the query text data are extracted; the query keywords are used as keyword labels for the query text data to form a training data set; the keyword extraction model is trained using the training data set, and the model parameters are continuously optimized during the training process so that the model can learn keyword feature patterns under different texts; the trained keyword extraction model is evaluated, the generalization ability of the model is verified through a test data set, and the model is tuned according to the evaluation results to obtain a final keyword extraction model.

[0063] In some optional implementations of this embodiment, the step of obtaining the association relationship of the form keywords includes:

[0064] The form keywords and preprocessed text data are input into the trained conditional random field sequence labeling model, the form keywords are automatically labeled through the sequence labeling method, the semantic roles and relationships of the form keywords are identified, and the single keywords and their associations are output.

[0065] Specifically, the query text in the historical query record data is segmented and POS tagged by natural language processing technology to form a tagged training data set, and the tagged data includes the semantic roles of keywords and the relationship between keywords; in the model training stage, the CRF model is trained using the tagged training data set, and the model parameters are optimized by the maximum likelihood estimation method to ensure that the model can accurately identify and classify key information in the text, thereby improving the accuracy and efficiency of model recognition; after the model training is completed, a conditional random field sequence tagging model is obtained, and the conditional random field sequence tagging model is used to automatically tag the pre-processed text data, and the semantic roles of form keywords in the text and the relationship between form keywords are identified. In this way, the system can automatically extract structured information from a large amount of text, realize automated information extraction and data classification, and be used for subsequent data analysis and business process automation, greatly improving processing efficiency and accuracy.

[0066] Step S204: Retrieve a first preset number of historical query records with the highest similarity to the customer's form request from the historical query records according to the form keyword as a candidate record set.

[0067] In this embodiment, query keywords of all historical forms in the historical query records are extracted, and by constructing an inverted index, the query keywords are associated with the form identifiers of the historical forms to obtain a form index; the form keywords and the query keywords are vectorized to obtain a form keyword vector and a query keyword vector; the similarity between the form keyword vector and the query keyword vector is calculated to obtain a similarity result; the historical query records are sorted according to the similarity results, and the first preset number of historical query records that are sorted first are selected as a candidate record set.

[0068] By constructing an inverted index, the query keywords of the historical forms in the historical query records are associated with the form identifiers, and a form index is constructed to facilitate the rapid location of historical forms containing specific keywords. Using the vector space model, the form keywords of the customer form request and the query keywords in the historical query records are converted into vectors to obtain the form keyword vector and the query keyword vector; by calculating the cosine similarity between the vectors, the historical query records with the highest matching degree with the customer form request are determined. On this basis, the historical query records with the highest matching degree of the first preset number are screened out as the candidate record set to improve the efficiency of information retrieval, wherein the first preset number is set according to the actual situation. In this way, the correlation between the customer form request and the historical query record is effectively quantified, thereby achieving the goal of efficient information retrieval. In addition, the accuracy and speed of the retrieval matching process are ensured.

[0069] It should be emphasized that in order to further ensure the privacy and security of historical query records, the above historical query records can also be stored in a blockchain node.

[0070] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.

[0071] Step S205 , using the maximum entropy model, calculates the probability distribution of the candidate record set according to the intent category, form keywords and association relationships, and obtains the predicted probability of each candidate record.

[0072] Calculate the intent relevance between the intent category and each candidate record in the candidate record set to form an intent relevance feature vector; calculate the matching degree between the form keywords and each candidate record in the candidate record set based on the association relationship to form a matching degree feature vector; concatenate the intent relevance feature vector and the matching degree feature vector to obtain a combined record feature vector; input the combined record feature vector into the trained maximum entropy model to calculate the probability distribution of each candidate record; and obtain the predicted probability of each candidate record based on the probability distribution.

[0073] Specifically, the candidate record set is input into the trained intent recognition model, the candidate records are classified for intent, the record intent category of each candidate record is obtained, the correlation between the record intent category and the intent category of the customer form request is calculated, and the intent correlation feature vector is obtained. For example, if they belong to the same intent, the correlation is 1, otherwise it is 0. The TF-IDF algorithm can be used to extract keywords from the candidate records to obtain record keywords, and the candidate records are automatically annotated through the conditional random field sequence annotation model to identify the semantic role of the record keywords and the relationship between the record keywords; the pre-trained word vector model is used to map the form keywords and their associations and the record keywords and their relationships into vector representations of fixed dimensions, and the cosine similarity between the record keyword vector and the form keyword vector is calculated to obtain the keyword matching feature; the intent correlation feature vector and the keyword matching feature vector are spliced ​​and input into the maximum entropy model (MaxEnt), and the weight coefficients of each feature are obtained through training to calculate the probability distribution of each candidate record, and the predicted probability of each candidate record is obtained according to the probability distribution. The larger the predicted probability, the higher the correlation between the candidate record and the customer form request.

[0074] By using the maximum entropy model to make probabilistic predictions on the candidate record set, it is possible to filter out redundant information that is irrelevant to customer intent, improve prediction efficiency and accuracy, more accurately match recommendation results, and improve query efficiency.

[0075] Step S206: determining a second preset number of candidate records that are most relevant to the customer's form request according to the predicted probability as a reference knowledge base for form generation.

[0076] In this embodiment, a second preset number of candidate records with the highest relevance are selected as a form generation reference knowledge base for a personalized form. The form generation reference knowledge base can provide a standardized template, simplify the configuration process, and dynamically generate forms according to customer needs, thereby enhancing the flexibility of the form.

[0077] Step S207: dynamically generate a personalized form that meets the customer's form request based on the form generation reference knowledge base.

[0078] Analyze customer-related historical records in the form generation reference knowledge base to obtain the form's key fields, layout structure, and filling rules; generate a form template based on the form's key fields, layout structure, and filling rules; determine the requirement type based on the customer's form request, dynamically adjust the form template's attribute parameters based on the requirement type, and output a personalized form.

[0079] Specifically, by analyzing the candidate records in the form generation reference knowledge base, natural language processing technology is used to extract keywords, such as using the TF-IDF algorithm to calculate the importance of words in the candidate records, and a preset number of keywords are selected as the key fields of the form; the layout structure of the form of the candidate record is analyzed, the layout structure features are extracted, and the layout mode of the form is determined based on the extracted layout structure features to obtain the layout structure template of the form; the filling rule metadata of the candidate record is obtained, and the value range, mandatory and other constraints of the key fields of the form are analyzed using regular expression matching, and the verification rules of the key fields of the form are generated to form a filling rule template; the weighted fusion algorithm is used to fuse the key fields of the form, the layout structure template and the filling rule template to form a complete form template. The form template is optimized using a genetic algorithm, and the template parameters are continuously optimized iteratively; according to the optimized form template, the page generation framework is used to dynamically render and generate a personalized form page.

[0080] By generating personalized forms, we can provide personalized services to customers, reduce the need for customers to fill in unnecessary content, and improve customer experience. Personalized forms dynamically adjust form content based on customer needs. Customers only need to fill in information related to their actual situation, which simplifies the operation process and reduces the possibility of errors.

[0081] In some optional implementations, customer feedback and filling status on the personalized form are obtained and used as new training samples. Incremental learning and online learning mechanisms are used to update model parameters in real time. The model includes parameters of the intent recognition model and the maximum entropy model, including:

[0082] By obtaining the data filled out by customers and using it as form content to form training samples for incremental learning and online learning, the parameters of the intent recognition model can be adjusted in real time. A keyword extraction model is used to extract key information from the form content to optimize the maximum entropy model. In this way, model parameters are updated in real time to improve accuracy and response speed. If the performance of the model on new samples does not reach the preset threshold, the learning algorithm is further adjusted to ensure that the model can effectively respond to various data changes and maintain the efficient operation of the system.

[0083] By preprocessing text data, the present application can eliminate inconsistencies in data and improve data quality; by performing intent recognition on preprocessed text data, it can accurately understand the customer's insurance needs and improve service efficiency; by extracting form keywords from preprocessed text data, and searching for a set of candidate records with the highest similarity to the customer's form request based on the form keywords, it can improve retrieval efficiency; by adopting a maximum entropy model, the probability distribution of the candidate record set is calculated based on intent categories, form keywords and associations, it can accurately understand the customer's query intent, improve query efficiency, avoid indiscriminate processing of all candidate records, and optimize resource allocation; according to the predicted probability, a form generation reference knowledge base is obtained, and personalized forms are dynamically generated based on the form generation reference knowledge base, which can provide personalized services to customers, improve customer experience, reduce error rates, and improve data accuracy and completeness. In addition, complex business processes are simplified and the efficiency of the overall business process is improved.

[0084] In some optional implementations, after the step of dynamically generating a personalized form that meets the customer form request, the method further includes:

[0085] Test the personalized form;

[0086] When the test passes, the personalized form is returned to the customer, and the customer form request and the generation record of the personalized form are stored in the form generation reference knowledge base.

[0087] In this embodiment, the generated personalized form page is automatically tested, and the usability and stability of the form are evaluated. The form template and constraints are dynamically adjusted according to the evaluation results to continuously optimize the personalized form. The generated personalized form is returned to the customer to meet the customer's business needs and provide an interactive and friendly information entry interface.

[0088] During the form information entry process, an intelligent feedback mechanism is used to verify and prompt form fields to guide users to fill in the information correctly.

[0089] By testing personalized forms, real-time content adjustments can be achieved, and the intelligent feedback mechanism can promptly detect and correct customer errors during the filling process, reducing the error rate in policy application and processing, and greatly improving business efficiency.

[0090] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0091] The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. Artificial intelligence software technology mainly includes computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning. A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it may include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium may be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0092] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0093] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a dynamic form generation device, and the embodiment of the device is Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0094] like Figure 3 As shown, the dynamic form generation device 300 described in this embodiment includes: a pre-processing module 301, an intention recognition module 302, an extraction module 303, a retrieval module 304, a prediction module 305, a construction module 306 and a generation module 307. Among them:

[0095] The preprocessing module 301 is used to receive a customer form request, obtain unstructured text data in the customer form request, and preprocess the unstructured text data to obtain preprocessed text data;

[0096] The intent recognition module 302 is used to input the pre-processed text data into the trained intent recognition model to determine the intent category to which the customer form request belongs;

[0097] The extraction module 303 is used to calculate the weight of each word in the preprocessed text data, select form keywords from all the words according to the weight, and obtain the association relationship of the form keywords;

[0098] The retrieval module 304 is used to retrieve a first preset number of historical query records with the highest similarity to the customer form request from the historical query records according to the form keyword as a candidate record set;

[0099] The prediction module 305 is used to use a maximum entropy model to calculate the probability distribution of the candidate record set according to the intent category, the form keyword and the association relationship, and obtain the predicted probability of each candidate record;

[0100] The construction module 306 is used to determine, according to the predicted probability, a second preset number of candidate records that are most relevant to the customer form request as a reference knowledge base for form generation;

[0101] The generation module 307 is used to generate a reference knowledge base based on the form, and dynamically generate a personalized form that meets the customer's form request.

[0102] It should be emphasized that in order to further ensure the privacy and security of historical query records, the above historical query records can also be stored in a blockchain node.

[0103] Based on the above-mentioned dynamic form generation device 300, by preprocessing the text data, inconsistencies in the data can be eliminated and data quality can be improved; by performing intent recognition on the preprocessed text data, the customer's insurance needs can be accurately understood and service efficiency can be improved; by extracting form keywords from the preprocessed text data, and searching for a set of candidate records with the highest similarity to the customer's form request based on the form keywords, the retrieval efficiency can be improved; by adopting the maximum entropy model, the probability distribution of the candidate record set is calculated according to the intent category, form keywords and association relationship, the customer's query intention can be accurately understood, the query efficiency can be improved, the indiscriminate processing of all candidate records can be avoided, and resource allocation can be optimized; a form generation reference knowledge base is obtained according to the predicted probability, and personalized forms are dynamically generated based on the form generation reference knowledge base, which can provide personalized services to customers, improve customer experience, reduce error rates, and improve data accuracy and completeness. In addition, complex business processes are simplified and the efficiency of the overall business process is improved.

[0104] In some optional implementations, the intention identification module 302 is further configured to:

[0105] Inputting the preprocessed text data into a trained intent recognition model, wherein the intent recognition model includes an embedding layer, an encoder network, and an output layer;

[0106] The preprocessed text data is vectorized by the embedding layer to obtain a text feature vector;

[0107] Inputting the text feature vector into the encoder network to perform attention mechanism calculation to obtain text semantic intent features;

[0108] The output layer calculates the intent category probability distribution according to the text semantic intent features, and determines the intent category with the largest probability value according to the intent category probability distribution as the target intent category of the customer form request.

[0109] By identifying the intent of customer form requests through the intent recognition model, we can provide efficient and accurate intent judgment capabilities, improve the efficiency and accuracy of intent recognition, and thus increase the processing speed of customer requests.

[0110] In this embodiment, the extraction module 303 is further used to:

[0111] By using a word frequency statistical method, the number of occurrences of each word in the pre-processed text data corresponding to a single customer form request is calculated to obtain preliminary word frequency data;

[0112] Using an inverse document frequency algorithm, calculate the distribution of each word in the preprocessed text data corresponding to all the customer form requests to obtain an inverse document frequency value;

[0113] Multiply the word frequency by the inverse document frequency to get the TF-IDF weight of each word;

[0114] The TF-IDF weights of all the words are sorted, and the words with the highest weight and the preset number of words are selected as form keywords.

[0115] The TF-IDF weight of each word is calculated through the inverse document frequency algorithm to measure the importance of the word, and form keywords are selected based on the TF-IDF weight, which can improve the efficiency and accuracy of keyword extraction.

[0116] In some optional implementations, the retrieval module 304 is further configured to:

[0117] Extracting query keywords of all historical forms in the historical query records, and associating the query keywords with form identifiers of the historical forms by constructing an inverted index to obtain a form index;

[0118] Performing vector conversion on the form keyword and the query keyword to obtain a form keyword vector and a query keyword vector;

[0119] Calculating the similarity between the form keyword vector and the query keyword vector to obtain a similarity result;

[0120] The historical query records are sorted according to the similarity results, and a first preset number of historical query records that are sorted first are selected as a candidate record set.

[0121] By retrieving historical query records related to customer form requests, the correlation between customer form requests and historical query records can be effectively quantified, thereby achieving the goal of efficient information retrieval and ensuring the accuracy and speed of the retrieval and matching process.

[0122] In some optional implementations of this embodiment, the prediction module 305 is further used to:

[0123] Calculating the intention relevance between the intention category and each candidate record in the candidate record set to form an intention relevance feature vector;

[0124] Calculating the matching degree between the form keyword and each candidate record in the candidate record set according to the association relationship to form a matching degree feature vector;

[0125] Concatenate the intention relevance feature vector and the matching degree feature vector to obtain a record combined feature vector;

[0126] Input the record and feature vector into the trained maximum entropy model to calculate the probability distribution of each candidate record;

[0127] The predicted probability of each candidate record is obtained according to the probability distribution.

[0128] By using the maximum entropy model to make probabilistic predictions on the candidate record set, it is possible to filter out redundant information that is irrelevant to customer intent, improve prediction efficiency and accuracy, more accurately match recommendation results, and improve query efficiency.

[0129] In some optional implementations of this embodiment, the generating module 307 includes:

[0130] An analysis submodule, used to analyze the customer-related historical records in the form generation reference knowledge base to obtain key fields, layout structure and filling rules of the form;

[0131] A template generation submodule, used to generate a form template according to the form key fields, the layout structure and the filling rules;

[0132] The form generation submodule is used to determine the requirement type according to the customer form request, dynamically adjust the attribute parameters of the form template based on the requirement type, and output a personalized form.

[0133] By generating personalized forms, we can provide personalized services to customers, reduce the need for customers to fill in unnecessary content, and improve customer experience. Personalized forms dynamically adjust form content based on customer needs. Customers only need to fill in information related to their actual situation, which simplifies the operation process and reduces the possibility of errors.

[0134] In some optional implementations, the dynamic form generating apparatus 300 further includes a testing module, which is used to:

[0135] Testing the personalized form;

[0136] When the test is passed, the personalized form is returned to the customer, and the customer form request and the generation record of the personalized form are stored in the form generation reference knowledge base.

[0137] By testing personalized forms, real-time content adjustments can be achieved, and the intelligent feedback mechanism can promptly detect and correct customer errors during the filling process, reducing the error rate in policy application and processing, and greatly improving business efficiency.

[0138] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0139] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 4 with a memory 41, a processor 42, and a network interface 43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0140] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0141] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk equipped on the computer device 4, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (FlashCard), etc. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the dynamic form generation method, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.

[0142] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run the computer-readable instructions stored in the memory 41 or process data, such as running the computer-readable instructions of the dynamic form generation method.

[0143] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0144] By preprocessing text data, inconsistencies in the data can be eliminated and data quality can be improved; by performing intent recognition on the preprocessed text data, the customer's insurance needs can be accurately understood and service efficiency can be improved; by extracting form keywords from the preprocessed text data and finding the candidate record set with the highest similarity to the customer's form request based on the form keywords, the retrieval efficiency can be improved; by adopting the maximum entropy model, the probability distribution of the candidate record set is calculated according to the intent category, form keywords and association relationship, the customer's query intention can be accurately understood, the query efficiency can be improved, the indiscriminate treatment of all candidate records can be avoided, and resource allocation can be optimized; the form generation reference knowledge base is obtained according to the predicted probability, and personalized forms are dynamically generated based on the form generation reference knowledge base, which can provide personalized services to customers, improve customer experience, reduce error rates, and improve data accuracy and completeness. In addition, complex business processes are simplified and the efficiency of the overall business process is improved.

[0145] The present application also provides another implementation, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the dynamic form generation method as described above.

[0146] By preprocessing text data, inconsistencies in the data can be eliminated and data quality can be improved; by performing intent recognition on the preprocessed text data, the customer's insurance needs can be accurately understood and service efficiency can be improved; by extracting form keywords from the preprocessed text data and finding the candidate record set with the highest similarity to the customer's form request based on the form keywords, the retrieval efficiency can be improved; by adopting the maximum entropy model, the probability distribution of the candidate record set is calculated according to the intent category, form keywords and association relationship, the customer's query intention can be accurately understood, the query efficiency can be improved, the indiscriminate treatment of all candidate records can be avoided, and resource allocation can be optimized; the form generation reference knowledge base is obtained according to the predicted probability, and personalized forms are dynamically generated based on the form generation reference knowledge base, which can provide personalized services to customers, improve customer experience, reduce error rates, and improve data accuracy and completeness. In addition, complex business processes are simplified and the efficiency of the overall business process is improved.

[0147] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0148] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A dynamic form generation method, characterized in that: The steps include: Receiving a customer form request, obtaining unstructured text data in the customer form request, and preprocessing the unstructured text data to obtain preprocessed text data; Inputting the preprocessed text data into the trained intent recognition model to determine the intent category to which the customer form request belongs; Calculating the weight of each word in the preprocessed text data, selecting form keywords from all the words according to the weight, and obtaining the association relationship of the form keywords; Retrieving a first preset number of historical query records having the highest similarity to the customer form request from the historical query records according to the form keyword as a candidate record set; Using a maximum entropy model, the probability distribution of the candidate record set is calculated according to the intent category, the form keyword and the association relationship, and a predicted probability of each candidate record is obtained; Determine, according to the predicted probability, a second preset number of candidate records that are most relevant to the customer form request as a reference knowledge base for form generation; Based on the form generation reference knowledge base, a personalized form that meets the customer form request is dynamically generated.

2. The dynamic form generation method according to claim 1, characterized in that: The step of inputting the pre-processed text data into the trained intent recognition model to determine the intent category to which the customer form request belongs includes: Inputting the preprocessed text data into a trained intent recognition model, wherein the intent recognition model includes an embedding layer, an encoder network, and an output layer; The preprocessed text data is vectorized by the embedding layer to obtain a text feature vector; Inputting the text feature vector into the encoder network to perform attention mechanism calculation to obtain text semantic intent features; The output layer calculates the intent category probability distribution according to the text semantic intent features, and determines the intent category with the largest probability value according to the intent category probability distribution as the target intent category of the customer form request.

3. The dynamic form generation method according to claim 1, characterized in that: The step of calculating the weight of each word in the preprocessed text data and selecting form keywords from all the words according to the weight comprises: By using a word frequency statistical method, the number of occurrences of each word in the pre-processed text data corresponding to a single customer form request is calculated to obtain preliminary word frequency data; Using an inverse document frequency algorithm, calculate the distribution of each word in the preprocessed text data corresponding to all the customer form requests to obtain an inverse document frequency value; Multiply the word frequency by the inverse document frequency to get the TF-IDF weight of each word; The TF-IDF weights of all the words are sorted, and the words with the highest weight and the preset number of words are selected as form keywords.

4. The dynamic form generation method according to claim 1, characterized in that: The step of retrieving a first preset number of historical query records having the highest similarity to the customer form request from the historical query records according to the form keyword as a candidate record set comprises: Extracting query keywords of all historical forms in the historical query records, and associating the query keywords with form identifiers of the historical forms by constructing an inverted index to obtain a form index; Performing vector conversion on the form keyword and the query keyword to obtain a form keyword vector and a query keyword vector; Calculating the similarity between the form keyword vector and the query keyword vector to obtain a similarity result; The historical query records are sorted according to the similarity results, and a first preset number of historical query records that are sorted first are selected as a candidate record set.

5. The dynamic form generation method according to claim 1, characterized in that: The step of using the maximum entropy model to calculate the probability distribution of the candidate record set according to the intent category, the form keyword and the association relationship to obtain the predicted probability of each candidate record includes: Calculating the intention relevance between the intention category and each candidate record in the candidate record set to form an intention relevance feature vector; Calculating the matching degree between the form keyword and each candidate record in the candidate record set according to the association relationship to form a matching degree feature vector; Concatenate the intention relevance feature vector and the matching degree feature vector to obtain a record combined feature vector; Input the record and feature vector into the trained maximum entropy model to calculate the probability distribution of each candidate record; The predicted probability of each candidate record is obtained according to the probability distribution.

6. The dynamic form generation method according to claim 1, characterized in that: The step of dynamically generating a personalized form that meets the customer's form request based on the form generation reference knowledge base includes: Analyze the customer-related historical records in the form generation reference knowledge base to obtain key form fields, layout structure and filling rules; Generate a form template according to the form key fields, the layout structure and the filling rules; The requirement type is determined according to the customer form request, the attribute parameters of the form template are dynamically adjusted based on the requirement type, and a personalized form is output.

7. The dynamic form generation method according to claim 1, characterized in that: After the step of dynamically generating a personalized form that meets the customer form request, the method further includes: Testing the personalized form; When the test is passed, the personalized form is returned to the customer, and the customer form request and the generation record of the personalized form are stored in the form generation reference knowledge base.

8. A dynamic form generation device, characterized in that: include: A preprocessing module, used for receiving a customer form request, obtaining unstructured text data in the customer form request, and preprocessing the unstructured text data to obtain preprocessed text data; An intent recognition module, used to input the preprocessed text data into a trained intent recognition model to determine the intent category to which the customer form request belongs; An extraction module, used to calculate the weight of each word in the preprocessed text data, select form keywords from all the words according to the weight, and obtain the association relationship of the form keywords; A retrieval module, configured to retrieve, from the historical query records according to the form keyword, a first preset number of historical query records having the highest similarity to the customer form request as a candidate record set; A prediction module, configured to use a maximum entropy model to calculate the probability distribution of the candidate record set according to the intent category, the form keyword and the association relationship, and obtain a predicted probability of each candidate record; A construction module, configured to determine, according to the predicted probability, a second preset number of candidate records most relevant to the customer form request as a reference knowledge base for form generation; A generation module is used to generate a reference knowledge base based on the form and dynamically generate a personalized form that meets the customer's form request.

9. A computer device, characterized in that: The method comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of the dynamic form generation method according to any one of claims 1 to 7 when executing the computer-readable instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the steps of the dynamic form generation method according to any one of claims 1 to 7 are implemented.

Images