Data processing method and device

Abstract

The embodiment of the invention provides a data processing method and device, and the method comprises the steps: extracting reference information corresponding to at least one reference field from to-be-processed data through a data processing model according to the to-be-processed data and corresponding first prompt information, and then constructing second prompt information based on the at least one adjustment field and the reference information, extracting adjustment information corresponding to the at least one adjustment field under the reference information from the to-be-processed data by using the data processing model according to the to-be-processed data and the second prompt information, and generating corresponding structured target data in combination with the reference information and the adjustment information. Therefore, the required reference information and the adjustment information under the reference information are automatically extracted, so that the corresponding structured target data are obtained, manual extraction one by one is not needed, missing of some key structured data is avoided, the labor input cost is low, and the accuracy and efficiency of data extraction are greatly improved.

Description

Technical Field

[0001] The embodiments of the present specification relate to the field of computer technology, and more particularly to a data processing method and device. Background Art

[0002] With the rapid development of computer technology and Internet technology, data in the form of electronic documents has shown explosive growth. The structured data included in electronic documents is often the key information for business departments to process their business. In the prior art, the required structured data is generally extracted manually from electronic documents, and then provided to downstream business departments for use to adjust the corresponding business. However, this data processing method may miss some key structured data, the accuracy of data extraction is low, and the labor cost investment is huge and the efficiency is low. Therefore, a more efficient and accurate data processing solution is urgently needed. Summary of the invention

[0003] In view of this, an embodiment of this specification provides a data processing method. One or more embodiments of this specification also relate to a data processing device, a computing device, a computer-readable storage medium and a computer program product to solve the technical defects existing in the prior art.

[0004] According to a first aspect of an embodiment of this specification, a data processing method is provided, including:

[0005] According to the data to be processed and the corresponding first prompt information, using the data processing model to extract reference information corresponding to at least one reference field from the data to be processed, wherein the first prompt information is constructed based on the at least one reference field;

[0006] constructing second prompt information based on at least one adjustment field and the reference information, and extracting adjustment information corresponding to the at least one adjustment field under the reference information from the data to be processed using the data processing model according to the data to be processed and the second prompt information;

[0007] According to the reference information and the adjustment information, structured target data corresponding to the data to be processed is generated.

[0008] According to a second aspect of an embodiment of this specification, there is provided a data processing device, including:

[0009] A first extraction module is configured to extract reference information corresponding to at least one reference field from the data to be processed using a data processing model according to the data to be processed and the corresponding first prompt information, wherein the first prompt information is constructed based on the at least one reference field;

[0010] A second extraction module is configured to construct second prompt information based on at least one adjustment field and the reference information, and extract adjustment information corresponding to the at least one adjustment field under the reference information from the data to be processed using the data processing model according to the data to be processed and the second prompt information;

[0011] A generating module is configured to generate structured target data corresponding to the data to be processed according to the benchmark information and the adjustment information.

[0012] According to a third aspect of an embodiment of this specification, a computing device is provided, including:

[0013] Memory and processor;

[0014] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the above data processing method are implemented.

[0015] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores computer-executable instructions, and when the instructions are executed by a processor, the steps of the above-mentioned data processing method are implemented.

[0016] According to a fifth aspect of the embodiments of this specification, a computer program product is provided, comprising a computer program / instruction, which implements the steps of the above-mentioned data processing method when executed by a processor.

[0017] An embodiment of the present specification realizes that, according to the data to be processed and the corresponding first prompt information, the benchmark information corresponding to at least one benchmark field is extracted from the data to be processed using a data processing model, and then the second prompt information is constructed based on at least one adjustment field and the benchmark information. According to the data to be processed and the second prompt information, the adjustment information corresponding to at least one adjustment field under the benchmark information is extracted from the data to be processed using a data processing model, and the corresponding structured target data is generated by combining the benchmark information and the adjustment information. In this way, the benchmark information can be automatically extracted from the data to be extracted based on the first prompt information, and then the adjustment information can be further automatically extracted based on the extracted benchmark information. The required benchmark information and the adjustment information under the benchmark information are automatically extracted twice, so as to obtain the corresponding structured target data, which is used for subsequent business departments to adjust and process the corresponding business. There is no need to extract them one by one manually, which avoids the omission of certain key structured data, and the labor input cost is low, which greatly improves the accuracy and efficiency of data extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is an application architecture diagram of data processing provided by an embodiment of this specification;

[0019] Figure 2 is a flow chart of a data processing method provided by an embodiment of this specification;

[0020] Figure 3a is a schematic diagram of a benchmark structured data provided by an embodiment of this specification;

[0021] Figure 3b is a schematic diagram of adjusting structured data provided by an embodiment of this specification;

[0022] Figure 4 is a processing flow chart of a data processing method provided by an embodiment of this specification;

[0023] Figure 5 is a structural schematic diagram of a data processing device provided by an embodiment of this specification;

[0024] Figure 6 It is a structural block diagram of a computing device provided by an embodiment of this specification. DETAILED DESCRIPTION

[0025] Many specific details are described in the following description to facilitate a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of this specification, so this specification is not limited to the specific implementation disclosed below.

[0026] The terms used in one or more embodiments of this specification are only for the purpose of describing specific embodiments, and are not intended to limit one or more embodiments of this specification. The singular forms of "a" and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0027] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0028] In one or more embodiments of this specification, a large model refers to a deep learning model with large-scale model parameters, which usually contains hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than 10 trillion model parameters. A large model can also be called a foundation model / foundation model. A large model is pre-trained with large-scale unlabeled corpus to produce a pre-trained model with more than 100 million parameters. This model can adapt to a wide range of downstream tasks, and the model has good generalization ability, such as a large language model (LLM), a multi-modal pre-training model, etc.

[0029] When the big model is used in practice, only a small number of samples are needed to fine-tune the pre-trained model and it can be applied to different tasks. The big model can be widely used in natural language processing (NLP), computer vision and other fields. Specifically, it can be applied to computer vision tasks such as visual question answering (VQA), image caption (IC), image generation, as well as natural language processing tasks such as text-based sentiment classification, text summary generation, and machine translation. The main application scenarios of the big model include digital assistants, intelligent robots, search, online education, office software, e-commerce, intelligent design, etc.

[0030] In this specification, a data processing method is provided. This specification also relates to a data processing apparatus, a computing device, a computer-readable storage medium, and a computer program product, which are described in detail one by one in the following embodiments.

[0031] Considering the huge amount of model parameters of the large model and the limited computing resources of the mobile terminal, the data processing method provided in the embodiment of the present application can be applied to Figure 1 The application scenarios shown are not limited to these. Figure 1 This is an application architecture diagram of data processing provided by an embodiment of this specification. Figure 1 In the application scenario shown, the large model is deployed in the server 10, and the server 10 can be connected to one or more client devices 20 through a local area network connection, a wide area network connection, an Internet connection, or other types of data networks. The client device 20 here may include but is not limited to: smart phones, tablet computers, laptops, PDAs, personal computers, smart home devices, vehicle-mounted devices, etc. The client device 20 can interact with the user through a graphical user interface to implement the call of the large model, thereby implementing the method provided in the embodiments of this specification.

[0032] In an embodiment of the present specification, a system composed of a client device and a server may perform the following steps: the client device receives the data to be processed input by the user and uploads it to the server. The server extracts the reference information corresponding to at least one reference field from the data to be processed using a data processing model according to the data to be processed and the corresponding first prompt information, wherein the first prompt information is obtained by constructing based on at least one reference field; constructs the second prompt information based on at least one adjustment field and the reference information, extracts the adjustment information corresponding to at least one adjustment field under the reference information from the data to be processed using the data processing model according to the data to be processed and the second prompt information; generates the structured target data corresponding to the data to be processed according to the reference information and the adjustment information.

[0033] It should be noted that, when the operating resources of the client device can meet the deployment and operating conditions of the large model, the embodiments of the present application can be carried out in the client device.

[0034] See also Figure 2 , Figure 2 A flow chart of a data processing method provided according to an embodiment of the present specification is shown, which specifically includes the following steps 202-206.

[0035] Step 202: According to the data to be processed and the corresponding first prompt information, the reference information corresponding to at least one reference field is extracted from the data to be processed using a data processing model, wherein the first prompt information is constructed based on at least one reference field.

[0036] Specifically, the data to be processed is data that needs to be extracted from structured data, and the data to be processed can be in text format or non-text format, such as image format, HTML (HyperText Markup Language) format, etc. In actual implementation, if the data to be processed is in text format, no conversion processing is required; if the data to be processed is in non-text format, the data to be processed is uploaded to a file server for conversion to obtain the data to be processed in text format, and then subsequent data extraction is performed.

[0037] It should be noted that the data to be processed can be structured data, semi-structured data or unstructured data. Key structured data such as benchmark information and adjustment information are extracted from the data to be processed to provide to downstream business departments for business adjustments. In addition, in different application scenarios, the data to be processed can have different contents. For example, in the insurance scenario, the data to be processed can be a rate table of a certain insurance; in the resume processing scenario, the data to be processed can be an electronic resume.

[0038] In actual implementation, the first prompt information is constructed based on at least one benchmark field, and is used to prompt the data processing model to extract the basic information corresponding to each benchmark field from the data to be processed, that is, the first prompt information is a kind of information that guides the data processing model to perform data processing. Among them, the benchmark field is the field corresponding to the benchmark information that needs to be extracted in the business scenario to which the data to be processed belongs. The benchmark information is the basic business information fixed in the business scenario. The basic fields and corresponding basic information are different in different business scenarios. For example, in the insurance scenario, the basic fields can be the name of insurance liability, benchmark premium, benchmark insured amount, benchmark rate, source and clause number, etc. For example, in the resume processing scenario, the basic fields can be basic information, education experience, work experience, project experience; the basic information is the detailed data corresponding to the above basic fields. In specific implementation, the benchmark fields can be configured based on actual business needs. The data processing model can be a trained large model, and the large model is used for data processing.

[0039] In an embodiment of the present specification, first prompt information can be constructed in advance based on at least one benchmark field to be extracted, and the data to be processed and the first prompt information can be input into a data processing model together. The data processing model can extract corresponding benchmark information from the data to be processed according to each benchmark field in the first prompt information.

[0040] For example, taking the insurance scenario as an example, if the data to be processed is a rate table, the benchmark fields to be extracted may include the insurance liability name, benchmark premium, benchmark insured amount, benchmark rate, source and clause number, etc. According to the above benchmark fields, a first prompt word "Please extract the insurance liability name, benchmark premium, benchmark insured amount, benchmark rate, source and clause number in the input data" is constructed, and the rate table and the first prompt information are input into the big model together. The big model extracts the benchmark information corresponding to the insurance liability name, benchmark premium, benchmark insured amount, benchmark rate, source and clause number from the input rate table according to the first prompt information. If the extracted The benchmark information can be "[{'Insurance Liability Name':'A Insurance Liability', 'Benchmark Premium':'-', 'Benchmark Sum':'M1', 'Benchmark Rate':'N1', 'Source':'XXXXXXA Insurance Clauses (Internet Clause A) Rate Table', 'Clause Number':'XXXXXX'}, {'Insurance Liability Name':'B Insurance Liability', 'Benchmark Premium':'-', 'Benchmark Sum':'M2', 'Benchmark Rate':'N2', 'Source':'XXXXXXB Insurance Clauses (Internet Clause A) Rate Table', 'Clause Number':'XXXXXX'}]".

[0041] In an optional implementation of this embodiment, before extracting the reference information corresponding to at least one reference field from the data to be processed by using the data processing model according to the data to be processed and the corresponding first prompt information, the method further includes:

[0042] Acquire at least one reference data in at least one business scenario;

[0043] For any business scenario, extract candidate fields from at least one reference data under the business scenario, and construct candidate prompt information corresponding to the business scenario based on the candidate fields;

[0044] Stores candidate prompt information corresponding to each business scenario.

[0045] It should be noted that the data content in different business scenarios is different, and the fields they contain are also different. If the same prompt information is used, key data in certain business scenarios may be missed, or certain benchmark fields may not be able to extract corresponding basic information, resulting in poor data extraction accuracy.

[0046] Therefore, in actual implementation, you can first obtain at least one reference data under at least one business scenario, and for each business scenario, extract candidate fields from at least one reference data under it. The candidate field is a common field included in at least one reference data under the business scenario. According to the candidate field, construct candidate prompt information corresponding to the business scenario. In this way, each business scenario can construct corresponding candidate prompt information and store it in the prompt information corpus.

[0047] For example, taking insurance as an example, for type A insurance business, multiple rate tables are obtained, and the multiple rate tables all include field 1, field 2, field 3, and field 4. At this time, candidate prompt information A corresponding to type A insurance business can be constructed based on field 1, field 2, field 3, and field 4; for type B insurance business, multiple rate tables are obtained, and the multiple rate tables all include field 1, field 2, field 3, field 5, and field 6. At this time, candidate prompt information B corresponding to type B insurance business can be constructed based on field 1, field 2, field 3, field 5, and field 6.

[0048] In the embodiments of the present specification, for different business scenarios, the fields included in at least one reference data can be analyzed, the candidate fields under the business scenario can be determined, and candidate prompt information corresponding to the business scenario can be constructed. Different business scenarios correspond to different candidate prompt information, which ensures that subsequent data extraction can match the corresponding business scenario and can improve the accuracy of data extraction.

[0049] In an optional implementation of this embodiment, based on the data to be processed and the corresponding first prompt information, using the data processing model to extract reference information corresponding to at least one reference field from the data to be processed includes:

[0050] Determine the target business scenario corresponding to the data to be processed;

[0051] Acquire first prompt information corresponding to the target business scenario from candidate prompt information corresponding to each business scenario;

[0052] According to the data to be processed and the first prompt information, the reference information corresponding to at least one reference field is extracted from the data to be processed using the data processing model.

[0053] In actual implementation, the prompt information corpus may pre-store candidate prompt information corresponding to each business scenario. After receiving the data to be processed, the target business scenario to which the data belongs may be determined based on the attribute information or identification information of the data to be processed. Then, the first prompt information corresponding to the target business scenario may be determined from the candidate prompt information corresponding to each business scenario stored in the prompt information corpus. Based on the data to be processed and the first prompt information, the data processing model may be used to extract benchmark information corresponding to at least one benchmark field from the data to be processed.

[0054] Continuing with the above example, assuming that the data to be processed belongs to a rate table under type A insurance business, the corresponding candidate prompt information A can be obtained from the prompt information corpus as the first prompt information, and the first prompt information and the rate table are input into the big model together, and the basic information corresponding to field 1, field 2, field 3, and field 4 are extracted from the rate table.

[0055] In the embodiments of the present specification, the corresponding first prompt information can be screened out from the candidate prompt information corresponding to each business scenario based on the target business scenario to which the data to be processed belongs, thereby extracting the benchmark information. The first prompt information matches the target business scenario to which it belongs, thereby ensuring the accuracy of the benchmark information extraction, and directly obtaining the first prompt information that pre-matches the target business scenario without the need for real-time generation, thereby improving the data extraction efficiency.

[0056] Of course, in actual implementation, in addition to pre-storing the candidate prompt information corresponding to each business scenario as mentioned above, and then querying the corresponding first prompt information based on the target business scenario to which the data to be processed belongs, the user can also upload the first prompt information corresponding to the data to be processed through the front-end device, or analyze the data to be processed to generate the corresponding first prompt information, etc. The embodiments of this specification do not limit this.

[0057] Step 204: construct second prompt information based on at least one adjustment field and the reference information, and extract adjustment information corresponding to at least one adjustment field under the reference information from the data to be processed using a data processing model according to the data to be processed and the second prompt information.

[0058] It should be noted that the data to be processed may include two types of information, namely, baseline information and adjustment information. In actual implementation, different types of information can be extracted in batches. In order to ensure that the information extracted twice is related, the baseline information obtained by extraction can be combined, and based on at least one adjustment field, the second prompt information can be constructed according to the set combination rules. The second prompt information is used to extract the adjustment information corresponding to at least one adjustment field under the baseline information from the data to be processed.

[0059] Among them, the adjustment field refers to the field involving adjustment information under the benchmark information. For example, in the insurance scenario, the adjustment field may refer to fields such as adjustment identification, adjustment factor, and adjustment coefficient. For example, in the resume processing scenario, the adjustment field may be fields such as internship experience and skill introduction. In addition, different business scenarios may be pre-configured with corresponding adjustment fields. And / or, different business scenarios may be configured with different combination rules to construct the second prompt information corresponding to the business scenario to which they belong.

[0060] In actual implementation, each extracted benchmark information can be added to the second prompt information to obtain the associated adjustment information; or, part of the information can be selected from the benchmark information and added to the second prompt information. For example, if the benchmark information includes an information identifier, the information identifier and at least one adjustment field can be combined to generate the corresponding second prompt information, thereby extracting the adjustment information corresponding to each adjustment field under the information identifier.

[0061] Continuing with the above example, taking the benchmark information including: 'Insurance Liability Name': 'A Insurance Liability' as an example, according to 'Insurance Liability Name': 'A Insurance Liability', the name of the rate adjustment coefficient, the rate adjustment factor, and the rate adjustment coefficient, construct the corresponding second prompt information, and input the rate table and the second prompt information into the big model together. The big model extracts the detailed data corresponding to the name of the rate adjustment coefficient, the rate adjustment factor, and the rate adjustment coefficient under the A Insurance Liability from the input rate table according to the second prompt information, and obtains the corresponding adjustment information: ["Insurance Liability Name":"A Insurance Liability". "Rate Adjustment Coefficient Name":"A Adjustment Coefficient":{"Rate Adjustment Factor":"Below 10% (inclusive)","Rate Adjustment Coefficient":"0.2-0.3";"Rate Adjustment Factor":"10%~20% (inclusive)","Rate Adjustment Coefficient":"0.3-0.5";"Rate Adjustment Factor":"20%~40% (inclusive)","Rate Adjustment Coefficient":"0.5-0.8";"Rate Adjustment Factor":"40%~60% (inclusive)","Rate Adjustment Coefficient":"0.8-1.0";"Rate Adjustment Factor":"60%~80% (inclusive)","Rate Adjustment Coefficient":"1.0-1.3";"Rate Adjustment Factor":"Above 80%","Rate Adjustment Coefficient":"1.3-3.0"};"B Adjustment Coefficient":"……";"C Adjustment Coefficient":"……";D Adjustment Coefficient":"……"].

[0062] As another example, the benchmark information may also include 'benchmark insured amount': 'M1', 'source': 'XXXXXXXA Insurance Clause (Internet Clause A) Rate Table', 'clause number': 'XXXXXX', and thus the corresponding second prompt information may be constructed based on 'insurance liability name': 'A insurance liability', 'benchmark insured amount': 'M1', 'source': 'XXXXXXXA Insurance Clause (Internet Clause A) Rate Table', 'clause number': 'XXXXXX', rate adjustment coefficient name, rate adjustment factor, rate adjustment coefficient, so as to extract the obtained adjustment information. In addition to the above content, it may also include: ['benchmark insured amount': 'M1', 'source': 'XXXXXXXA Insurance Clause (Internet Clause A) Rate Table', 'clause number': 'XXXXXX'].

[0063] In an optional implementation of this embodiment, constructing the second prompt information based on at least one adjustment field and the reference information includes:

[0064] Determining whether the benchmark information includes information corresponding to the set benchmark field;

[0065] If not included, the second prompt information is generated based on at least one adjustment field and the reference information, combined with the default reference information of the set reference field.

[0066] The set base fields refer to the fields that cannot be missing in the corresponding business scenario.

[0067] It should be noted that in some business scenarios, there may be some indispensable field information. If the extracted benchmark information is missing, it will affect the integrity and accuracy of the adjustment information extracted later. Therefore, in actual implementation, it can be determined whether the extracted benchmark information includes the information corresponding to the set benchmark field. If it does, the corresponding second information can be constructed based on the information; if it does not, it means that the extracted benchmark information is missing. At this time, the default benchmark information of the set benchmark field can be added to the second prompt information.

[0068] For example, in an insurance scenario, the base field is set to "insurance amount", and the default base information of the base field is set to 100,000. Assuming that the information corresponding to the insurance amount in the extracted base information is empty, such as "insurance amount": "-", it means that the detailed data corresponding to the insurance amount is not extracted, and the extracted base information lacks the insurance amount. At this time, the second prompt information can be generated based on "insurance amount": "100,000".

[0069] In an embodiment of the present specification, when the benchmark information extracted using the data processing model contains missing information of the set benchmark field, the default benchmark information of the set benchmark field can be used to complete the information, generate corresponding second prompt information, and substitute the default benchmark information of the set benchmark field as a parameter into the second prompt information, thereby ensuring the integrity and accuracy of the data extraction.

[0070] In an optional implementation of this embodiment, constructing the second prompt information based on at least one adjustment field and the reference information includes:

[0071] Obtaining an extraction sample of adjustment information, wherein the extraction sample is used to describe a method for extracting adjustment fields in sample data and the extracted sample adjustment information;

[0072] Based on the at least one adjustment field, the benchmark information and the extracted sample, second prompt information is constructed.

[0073] It should be noted that when generating the second prompt information, it can also be combined with the extraction sample of the adjustment information to describe the method of extracting the adjustment field in the sample data and the extracted sample adjustment information as an example information to guide the large model to extract data from the input data to be processed in an example manner.

[0074] For example, taking the insurance scenario as an example, assuming that at least one adjustment field includes an adjustment identifier, an adjustment factor, and an adjustment coefficient, and the benchmark information includes insurance liability A, the following second prompt information can be generated: "For the input data to be processed, extract the adjustment identifier, adjustment factor, adjustment coefficient and other detailed data under insurance liability A. You can refer to the sample data and analyze the relevant fields such as insurance liability A, adjustment identifier, adjustment factor, adjustment coefficient, etc. to obtain the corresponding adjustment information: ["Insurance liability name":"Insurance liability A". "Rate adjustment coefficient name":"A adjustment coefficient":{"Rate adjustment factor":"Below 10%","Rate adjustment coefficient":"0.2-0.3";...};"B adjustment coefficient":"...";"C adjustment coefficient":"...";D adjustment coefficient":"..."]".

[0075] In an embodiment of the present specification, an extraction sample can be added to the second prompt information to illustrate the method of extracting the adjustment field in the sample data and the extracted sample adjustment information, so as to help the data processing model understand the processing method of the adjustment information in the data to be processed and the data format, content and other information of the output adjustment information, thereby guiding the large model to extract the adjustment information for the input data to be processed in the same way as the sample data, so that the data processing model can extract the adjustment information with higher accuracy.

[0076] In actual implementation, in addition to adding extraction samples in the second prompt information, extraction samples can also be added to the above-mentioned first prompt information to illustrate the method of extracting the benchmark fields in the sample data and the extracted sample benchmark information, so as to help the data processing model understand the processing method of the benchmark information in the data to be processed, and require the data format, content and other information of the output benchmark information, so as to guide the large model to extract benchmark information for the input data to be processed in the same way as the sample data, so that the data processing model can extract benchmark information more accurately.

[0077] Step 206: Generate structured target data corresponding to the data to be processed according to the benchmark information and the adjustment information.

[0078] It should be noted that after using the data processing model to extract the baseline information and adjustment information corresponding to the data to be processed, the structured target data corresponding to the data to be processed can be generated based on the baseline information and the adjustment information, and stored in the corresponding database. The database can provide a query interface to the downstream business departments. When the downstream business departments need to obtain the corresponding structured target data, they can query the corresponding data from the database to achieve adjustment and management of the business.

[0079] In an optional implementation of this embodiment, generating structured target data corresponding to the data to be processed according to the benchmark information and the adjustment information includes:

[0080] According to a preset generation rule, fill the benchmark information corresponding to at least one benchmark field into a corresponding position to obtain benchmark structured data corresponding to the benchmark information;

[0081] According to the preset extraction rules, the adjustment information is processed to generate adjustment structured data;

[0082] The baseline structured data and the adjusted structured data are used as structured target data and stored in a database.

[0083] In actual implementation, the preset generation rule is a data filling method configured based on actual business needs. According to the preset generation rule, the benchmark information corresponding to at least one benchmark field can be filled in the corresponding position to obtain the benchmark structured data corresponding to the benchmark information. In addition, the preset extraction rule is a numerical value splitting rule configured based on actual business needs. The numerical value required by the business department can be further split from the adjustment information to generate the adjustment structured data. The benchmark structured data and the adjustment structured data are used as structured target data and stored in the database, and the obtained benchmark structured data and the adjustment structured data can also be displayed to the front-end user as structured target data.

[0084] In the embodiments of the present specification, after using the data processing model to extract the baseline information and adjustment information corresponding to the data to be processed, it is possible to further configure preset generation rules for processing the baseline information and preset extraction rules for further splitting the adjustment information based on actual business needs, and further use the rules to automatically process the baseline information and adjustment information to obtain structured target data that meets business needs, so that the obtained structured target data is more in line with actual scenarios and the data extraction accuracy is higher.

[0085] In an optional implementation of this embodiment, the adjustment information includes a baseline information identifier, a risk factor identifier, an adjustment factor, and an adjustment coefficient; the adjustment information is processed according to a preset extraction rule to generate adjustment structured data, including:

[0086] Extract the adjustment factor according to the preset extraction rule to obtain the adjustment factor range;

[0087] Extract the adjustment coefficient according to the preset extraction rule to obtain the adjustment factor value;

[0088] Adjustment structured data is generated based on the benchmark information identifier, risk factor identifier, adjustment factor, adjustment coefficient, adjustment factor range and adjustment factor value.

[0089] In actual implementation, the adjustment information may include an adjustment factor and an adjustment coefficient, wherein the adjustment factor indicates the condition for adjustment, and the adjustment coefficient indicates the corresponding adjustment value under the condition. Therefore, the adjustment factor may be extracted to obtain the adjustment factor range, and the adjustment coefficient may be extracted to obtain the adjustment factor value, and then each piece of information may be filled into the corresponding field to obtain the adjustment structured data.

[0090] Continuing with the above example, the benchmark information extracted using the data processing model can be "[{'Insurance Liability Name':'A Insurance Liability', 'Benchmark Premium':'-', 'Benchmark Amount':'M1', 'Benchmark Rate':'N1', 'Source':'XXXXXXXA Insurance Clause (Internet Clause A) Rate Table', 'Clause Number':'XXXXXX'}, {'Insurance Liability Name':'B Insurance Liability', 'Benchmark Premium':'-', 'Benchmark Amount':'M2', 'Benchmark Rate':'N2', 'Source':'XXXXXXXB Insurance Clause (Internet Clause A) Rate Table', 'Clause Number':'XXXXXX'}]", and the preset generation rule is to fill in each benchmark information into the corresponding field. At this time, the following can be obtained: Figure 3a The benchmark structured data shown, Figure 3a It is a schematic diagram of a benchmark structured data provided by an embodiment of this specification.

[0091] Assuming that the preset extraction rule is the extraction end value, for "A insurance liability", "A adjustment coefficient rate adjustment factor": "10% or less (inclusive)" can be split into the adjustment factor range 0 (lower limit) and 10 (upper limit), and the corresponding "rate adjustment coefficient": "0.2-0.3" can be split into the adjustment factor value 0.2 (value 1) and 0.3 (value 2); "rate adjustment factor": "10% ~ 20% (inclusive)" can be split into the adjustment factor range 11 (lower limit) and 20 (upper limit), and the corresponding "rate adjustment coefficient": "0.3-0.5" can be split into the adjustment factor value 0.3 (value 1) and 0.5 (value 2); "rate adjustment factor": "20% ~ 40% (inclusive)" can be split into the adjustment factor range 21 (lower limit) and 40 (upper limit), and the corresponding "rate adjustment coefficient": "0.5-0.8" can be split into the adjustment factor value 0.5 (value 1) and 0.8 (value 2); "Rate adjustment factor": "40% ~ 60% (inclusive)" can be split into an adjustment factor range of 41 (lower limit) and 60 (upper limit), and the corresponding "Rate adjustment coefficient": "0.8-1.0" can be split into adjustment factor values ​​0.8 (value 1) and 1.0 (value 2); "Rate adjustment factor": "60% ~ 80% (inclusive)" can be split into an adjustment factor range of 61 (lower limit) and 80 (upper limit), and the corresponding "Rate adjustment coefficient": "1.0-1.3" can be split into adjustment factor values ​​1.0 (value 1) and 1.3 (value 2); "Rate adjustment factor": "above 80%" can be split into an adjustment factor range of 81 (lower limit) and 100 (upper limit), and the corresponding "Rate adjustment coefficient": "1.3-3.0" can be split into adjustment factor values ​​1.3 (value 1) and 3.0 (value 2). The generated adjusted structured data is as follows Figure 3b As shown, Figure 3b It is a schematic diagram of adjusting structured data provided by an embodiment of this specification.

[0092] Of course, in actual implementation, in addition to being configured as the extraction end value, the preset extraction rule can also extract other values ​​based on actual business needs, such as extraction by proportion, etc., and the embodiments of this specification do not limit this.

[0093] In the embodiments of the present specification, when extracting structured data from the data to be processed, in addition to using the data processing model to extract the baseline information and the adjustment information twice respectively, business segmentation logic can also be introduced to perform business segmentation on the adjustment information, so that the extracted structured data can meet the actual business needs, thereby ensuring the quality of data extraction.

[0094] In an optional implementation of this embodiment, generating structured target data corresponding to the data to be processed according to the benchmark information and the adjustment information includes:

[0095] Displaying the data to be processed in the first area, and displaying the benchmark information and the adjustment information in the second area;

[0096] In response to the review pass instruction of the benchmark information and the adjustment information, structured target data corresponding to the benchmark information and the adjustment information are generated and stored in the database.

[0097] The first area and the second area are different areas in the front-end device, such as a left area, a right area, and the like.

[0098] In actual implementation, the data to be processed can be displayed in the first area, and the benchmark information and adjustment information can be displayed in the second area, so that the user can compare the original data before extraction and the data extraction results to review the data extraction results. If there are errors, they can be modified; if there are no errors, the review pass instruction can be triggered, which means that the benchmark information and adjustment information have passed manual review and the structured target data is relatively accurate. At this time, the structured target data is stored in the database.

[0099] It should be noted that manual review was added to correct the automated processing results of the model and program to ensure the accuracy of the structured target data stored in the database.

[0100] Of course, in actual implementation, the obtained structured target data can be directly stored in the database without manual review. The subsequent business department can pull the corresponding structured target data from the database center, and the business department can manually review the structured target data. If there are any modifications, the modifications can be synchronously updated to the database. The embodiments of this specification do not limit this.

[0101] In an optional implementation of this embodiment, after generating structured target data corresponding to the data to be processed according to the benchmark information and the adjustment information, the method further includes:

[0102] Parse structured target data to obtain sample benchmark information and sample adjustment information;

[0103] The data to be processed corresponding to the structured target data is used as training samples, and the sample benchmark information and sample adjustment information are used as sample labels to tune and train the data processing model.

[0104] It should be noted that the structured target data is the processing result obtained by structured extraction of the data to be processed. The structured target data is relatively accurate, so the structured target data can be used as sample data for tuning training to further tune the data processing model.

[0105] In actual implementation, the stored structured target data can be obtained from the database, and the sample benchmark information and sample adjustment information can be parsed. The data to be processed corresponding to the structured target data can be used as training samples, and the sample benchmark information and sample adjustment information can be used as sample labels to perform supervised tuning training on the data processing model, thereby improving the model performance of the data processing model and thereby improving the accuracy of data processing.

[0106] Of course, in another optional implementation, in addition to the above-mentioned tuning and training of the data processing model based on the structured target data stored in the database, the obtained structured target data can also be provided to the downstream business department to obtain their feedback information, which is information indicating whether the structured target data meets the business requirements, and construct training samples based on the feedback information to further tune the data processing model. In this way, through the interactive feedback mechanism, the sample data is updated, the data processing model is further tuned and trained, and the model performance of the data processing model is further improved.

[0107] The embodiments of the present specification provide a data processing method, which can first automatically extract benchmark information from the data to be extracted based on the first prompt information, and then automatically extract adjustment information based on the extracted benchmark information. The benchmark information extracted in the benchmark part can be substituted as a parameter into the second prompt information, so that the extraction of the adjustment information is more accurate; and, a business segmentation logic is introduced to perform business segmentation on the adjustment information, so that the extracted structured data can meet the actual business needs and ensure the quality of data extraction; secondly, the default benchmark information of the set benchmark field can be substituted as a parameter into the second prompt information to complete the information of the set benchmark field of the missing information, thereby ensuring the integrity and accuracy of the data extraction; furthermore, extraction samples can be added to the prompt information, and examples are given for the extraction method of the benchmark information and adjustment information in the sample data and the extracted information, so as to help the data processing model understand the processing method of the data to be processed, the required output data format, content and other information, thereby guiding the large model to extract data from the input data to be processed in the same way as the extraction sample, so that the data processing model has higher accuracy in data extraction.

[0108] It should be noted that the rate table of insurance products is the key data in the pricing, risk management, sales and service processes of insurance products, and the structuring of rate factors is particularly important.

[0109] The following combination Figure 4 , taking the application of the data processing method provided in this specification in the insurance scenario as an example, the data processing method is further explained. Figure 4A processing flow chart of a data processing method provided by an embodiment of the present specification is shown, which specifically includes the following steps.

[0110] Step 402: Obtain the HTML file of the original rate table, upload the HTML file to the file server for format conversion, and obtain the text file of the original rate table.

[0111] Step 404: Determine the insurance type to which the original rate table belongs, and obtain first prompt information corresponding to the insurance type.

[0112] The first prompt information is generated based on the basic rate factor required for the insurance type.

[0113] Step 406: Input the text file of the original rate table and the first prompt information into the large model, and use the large model to extract the base rate factor corresponding to at least one base field from the text file of the original rate table.

[0114] Step 408: Construct a second prompt information based on at least one adjustment field and a base rate factor, input the data to be processed and the second prompt information into the big model, use the big model to extract the adjustment rate factor corresponding to at least one adjustment field under the base rate factor from the text file of the original rate table, and generate a structured rate factor based on the base rate factor and the adjustment rate factor.

[0115] Step 410: Display the original rate table in the first area, and display the structured rate factor in the second area. In response to the review approval instruction of the structured rate factor, store the structured rate factor in the database.

[0116] Step 412: Use the structured rate factors in the database and the corresponding original rate table as sample data to perform tuning training on the large model.

[0117] It should be noted that the rate factors in the manual extraction rate table are limited by personal professional knowledge, work status, working hours and other factors. For example, if an error is copied from the original rate table or an adjustment coefficient is mistyped, the quality of the rate factor will be reduced. When the data reaches tens of millions, manual extraction has a huge human cost, is extremely inefficient and time-consuming.

[0118] The data processing method provided in the embodiments of this specification can first automatically extract the base rate factor from the original rate table based on the first prompt information, and then automatically extract the adjustment rate factor based on the extracted base rate factor. The base rate factor extracted in the base part can be used as a parameter to substitute into the second prompt information, so that the extraction of the adjustment rate factor is more accurate. The rate table is automatically parsed into structured rate factors, and the rate factors can be quickly and accurately extracted from the rate table, with low labor input costs, which greatly improves the accuracy and efficiency of the extraction of structured rate factors.

[0119] Corresponding to the above method embodiment, this specification also provides a data processing device embodiment, Figure 5 FIG. 1 is a schematic diagram showing the structure of a data processing device provided by an embodiment of the present specification. Figure 5 As shown, the device comprises:

[0120] A first extraction module 502 is configured to extract reference information corresponding to at least one reference field from the data to be processed using a data processing model according to the data to be processed and the corresponding first prompt information, wherein the first prompt information is constructed based on the at least one reference field;

[0121] The second extraction module 504 is configured to construct second prompt information based on at least one adjustment field and the reference information, and extract adjustment information corresponding to at least one adjustment field under the reference information from the data to be processed using a data processing model according to the data to be processed and the second prompt information;

[0122] The generating module 506 is configured to generate structured target data corresponding to the data to be processed according to the benchmark information and the adjustment information.

[0123] Optionally, the device further comprises a construction module configured to:

[0124] Acquire at least one reference data in at least one business scenario;

[0125] For any business scenario, extract candidate fields from at least one reference data under the business scenario, and construct candidate prompt information corresponding to the business scenario based on the candidate fields;

[0126] Stores candidate prompt information corresponding to each business scenario.

[0127] Optionally, the second extraction module 504 is further configured to:

[0128] Determine the target business scenario corresponding to the data to be processed;

[0129] Acquire first prompt information corresponding to the target business scenario from candidate prompt information corresponding to each business scenario;

[0130] According to the data to be processed and the first prompt information, the reference information corresponding to at least one reference field is extracted from the data to be processed using the data processing model.

[0131] Optionally, the second extraction module 504 is further configured to:

[0132] Determining whether the benchmark information includes information corresponding to the set benchmark field;

[0133] If not included, the second prompt information is generated based on at least one adjustment field and the reference information, combined with the default reference information of the set reference field.

[0134] Optionally, the second extraction module 504 is further configured to:

[0135] Obtaining an extraction sample of adjustment information, wherein the extraction sample is used to describe a method for extracting adjustment fields in sample data and the extracted sample adjustment information;

[0136] Based on the at least one adjustment field, the benchmark information and the extracted sample, second prompt information is constructed.

[0137] Optionally, the generating module 506 is further configured to:

[0138] According to a preset generation rule, fill the benchmark information corresponding to at least one benchmark field into a corresponding position to obtain benchmark structured data corresponding to the benchmark information;

[0139] According to the preset extraction rules, the adjustment information is processed to generate adjustment structured data;

[0140] The baseline structured data and the adjusted structured data are used as structured target data and stored in a database.

[0141] Optionally, the adjustment information includes a reference information identifier, a risk factor identifier, an adjustment factor, and an adjustment coefficient; the generating module 506 is further configured to:

[0142] Extract the adjustment factor according to the preset extraction rule to obtain the adjustment factor range;

[0143] Extract the adjustment coefficient according to the preset extraction rule to obtain the adjustment factor value;

[0144] Adjustment structured data is generated based on the benchmark information identifier, risk factor identifier, adjustment factor, adjustment coefficient, adjustment factor range and adjustment factor value.

[0145] Optionally, the generating module 506 is further configured to:

[0146] Displaying the data to be processed in the first area, and displaying the benchmark information and the adjustment information in the second area;

[0147] In response to the review approval instruction of the benchmark information and the adjustment information, structured target data corresponding to the benchmark information and the adjustment information are generated and stored in the database.

[0148] Optionally, the device further includes a tuning module configured to:

[0149] Parse structured target data to obtain sample benchmark information and sample adjustment information;

[0150] The data to be processed corresponding to the structured target data is used as training samples, and the sample benchmark information and sample adjustment information are used as sample labels to tune and train the data processing model.

[0151] The data processing device provided in the embodiment of the present specification includes a first extraction module, a second extraction module and a generation module. Through the interactive configuration of the above three functional modules, the benchmark information can be automatically extracted from the data to be extracted based on the first prompt information, and then the adjustment information can be further automatically extracted based on the extracted benchmark information. The required benchmark information and the adjustment information under the benchmark information are automatically extracted twice, so as to obtain the corresponding structured target data for subsequent business departments to adjust the corresponding business. There is no need for manual extraction one by one, which avoids the omission of certain key structured data, has low labor input costs, and greatly improves the accuracy and efficiency of data extraction.

[0152] The above is a schematic scheme of a data processing device of this embodiment. It should be noted that the technical scheme of the data processing device and the technical scheme of the above data processing method belong to the same concept, and the details of the technical scheme of the data processing device that are not described in detail can be referred to the description of the technical scheme of the above data processing method.

[0153] Figure 6 The block diagram of a computing device according to an embodiment of the present specification is shown. The components of the computing device 600 include but are not limited to a memory 610 and a processor 620. The processor 620 is connected to the memory 610 via a bus 630, and the database 650 is used to store data.

[0154] The computing device 600 also includes an access device 640, which enables the computing device 600 to communicate via one or more networks 660. Examples of these networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 640 may include one or more of any type of network interface (e.g., a network interface card (NIC)) of wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a world-wide interoperability for microwave access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, and a near field communication (NFC).

[0155] In one embodiment of the present specification, the above components of the computing device 600 and Figure 6 Other components not shown in the figure may also be connected to each other, for example, via a bus. It should be understood that Figure 6 The computing device structure block diagram shown is only for the purpose of illustration, and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0156] The computing device 600 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook, etc.), a mobile phone (e.g., a smart phone), a wearable computing device (e.g., a smart watch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or a personal computer (PC). The computing device 600 may also be a mobile or stationary server.

[0157] The processor 620 is used to execute the following computer executable instructions, which implement the steps of the above data processing method when executed by the processor.

[0158] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above data processing method belong to the same concept, and the details not described in detail in the technical scheme of the computing device can be referred to the description of the technical scheme of the above data processing method.

[0159] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions, which can implement the steps of the above-mentioned data processing method when executed by a processor.

[0160] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the above data processing method belong to the same concept, and the details not described in detail in the technical scheme of the storage medium can be referred to the description of the technical scheme of the above data processing method.

[0161] An embodiment of the present specification further provides a computer program, wherein when the computer program is executed in a computer, the computer is caused to execute the steps of the above-mentioned data processing method.

[0162] The above is a schematic scheme of a computer program of this embodiment. It should be noted that the technical scheme of the computer program and the technical scheme of the above data processing method belong to the same concept, and the details not described in detail in the technical scheme of the computer program can be referred to the description of the technical scheme of the above data processing method.

[0163] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0164] Computer instructions include computer program codes, which may be in source code form, object code form, executable files or some intermediate form, etc. Computer readable media may include: any entity or device capable of carrying computer program codes, recording media, USB flash drives, mobile hard disks, magnetic disks, optical disks, computer memories, read-only memories (ROM), random access memories (RAM), electric carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the contents of computer readable media may be appropriately increased or decreased according to the requirements of patent practice. For example, in some regions, according to patent practice, computer readable media do not include electric carrier signals and telecommunication signals.

[0165] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0166] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0167] The preferred embodiments of this specification disclosed above are only used to help explain this specification. The optional embodiments do not describe all the details in detail, nor do they limit the invention to only the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that technicians in the relevant technical field can well understand and use this specification. This specification is only limited by the claims and their full scope and equivalents.

Claims

1. A data processing method, characterized in that: include: According to the data to be processed and the corresponding first prompt information, using the data processing model to extract reference information corresponding to at least one reference field from the data to be processed, wherein the first prompt information is constructed based on the at least one reference field; constructing second prompt information based on at least one adjustment field and the reference information, and extracting adjustment information corresponding to the at least one adjustment field under the reference information from the data to be processed using the data processing model according to the data to be processed and the second prompt information; According to the reference information and the adjustment information, structured target data corresponding to the data to be processed is generated.

2. The method according to claim 1, characterized in that: Before extracting the reference information corresponding to at least one reference field from the data to be processed by using the data processing model according to the data to be processed and the corresponding first prompt information, the method further includes: Acquire at least one reference data in at least one business scenario; For any business scenario, extract candidate fields from at least one reference data under the business scenario, and construct candidate prompt information corresponding to the business scenario according to the candidate fields; Stores candidate prompt information corresponding to each business scenario.

3. The method according to claim 2, characterized in that The extracting, according to the data to be processed and the corresponding first prompt information, benchmark information corresponding to at least one benchmark field from the data to be processed by using a data processing model includes: Determine the target business scenario corresponding to the data to be processed; Acquire first prompt information corresponding to the target business scenario from candidate prompt information corresponding to each business scenario; According to the data to be processed and the first prompt information, a data processing model is used to extract reference information corresponding to at least one reference field from the data to be processed.

4. The method according to claim 1, characterized in that: The constructing the second prompt information based on the at least one adjustment field and the reference information includes: Determining whether the benchmark information includes information corresponding to a set benchmark field; If not included, the second prompt information is generated based on the at least one adjustment field and the reference information, combined with the default reference information of the set reference field.

5. The method according to claim 1, characterized in that The constructing the second prompt information based on the at least one adjustment field and the reference information includes: Acquire an extraction sample of the adjustment information, wherein the extraction sample is used to describe a method for extracting an adjustment field in sample data and the extracted sample adjustment information; The second prompt information is constructed based on the at least one adjustment field, the reference information and the extracted sample.

6. The method according to claim 1, characterized in that The step of generating structured target data corresponding to the data to be processed according to the reference information and the adjustment information includes: According to a preset generation rule, fill the reference information corresponding to the at least one reference field into a corresponding position to obtain reference structured data corresponding to the reference information; Processing the adjustment information according to preset extraction rules to generate adjustment structured data; The reference structured data and the adjusted structured data are used as the structured target data and stored in a database.

7. The method according to claim 6, characterized in that The adjustment information includes a benchmark information identifier, a risk factor identifier, an adjustment factor, and an adjustment coefficient; the adjustment information is processed according to a preset extraction rule to generate adjustment structured data, including: Extracting the adjustment factor according to a preset extraction rule to obtain an adjustment factor range; Extract the adjustment coefficient according to a preset extraction rule to obtain an adjustment factor value; The adjustment structured data is generated according to the benchmark information identifier, the risk factor identifier, the adjustment factor, the adjustment coefficient, the adjustment factor range, and the adjustment factor value.

8. The method according to claim 1, characterized in that: The step of generating structured target data corresponding to the data to be processed according to the reference information and the adjustment information includes: Displaying the data to be processed in a first area, and displaying the benchmark information and the adjustment information in a second area; In response to an instruction that the reference information and the adjustment information have been reviewed and approved, structured target data corresponding to the reference information and the adjustment information are generated and stored in a database.

9. The method according to claim 1, characterized in that: After generating structured target data corresponding to the data to be processed according to the reference information and the adjustment information, the method further includes: Parsing the structured target data to obtain sample benchmark information and sample adjustment information; The data to be processed corresponding to the structured target data is used as a training sample, and the sample benchmark information and sample adjustment information are used as sample labels to perform tuning training on the data processing model.

10. A data processing device, characterized in that: include: A first extraction module is configured to extract reference information corresponding to at least one reference field from the data to be processed using a data processing model according to the data to be processed and the corresponding first prompt information, wherein the first prompt information is constructed based on the at least one reference field; A second extraction module is configured to construct second prompt information based on at least one adjustment field and the reference information, and extract adjustment information corresponding to the at least one adjustment field under the reference information from the data to be processed using the data processing model according to the data to be processed and the second prompt information; A generating module is configured to generate structured target data corresponding to the data to be processed according to the benchmark information and the adjustment information.

11. A computing device, characterized in that: include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the data processing method according to any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium, characterized in that: It stores computer executable instructions, which, when executed by a processor, implement the steps of the data processing method described in any one of claims 1 to 9.

13. A computer program product, characterized in that The invention comprises a computer program / instruction, which, when executed by a processor, implements the steps of the data processing method according to any one of claims 1 to 9.

Images