Insurance consultation intelligent response method and device, computer device and storage medium

Abstract

The application relates to the field of medical data processing and discloses an insurance consultation intelligent response method and device, equipment and a medium. The method comprises the following steps: generating a consultation content label and a consultation sentence pattern label according to insurance consultation information, the consultation content label representing the content to be consulted by the insurance consultation sentence, and the consultation sentence pattern label representing the sentence pattern corresponding to the insurance consultation sentence; determining context information corresponding to the insurance consultation sentence according to the consultation content label; generating a virtual context based on the context information and the consultation sentence pattern label, and generating a target fusion sentence by fusing the insurance consultation sentence and the virtual context; generating target response information corresponding to the consultation content label according to the target fusion sentence; generating a target response sentence based on the target response information and the consultation sentence pattern label, and outputting the target response sentence to a target terminal, simulating the context corresponding to the consultation sentence based on the consultation sentence, accurately understanding the consultation information and giving a detailed and correct answer, and improving the user consultation experience.

Description

Technical Field

[0001] This application relates to the field of insurance consulting, and more particularly to an intelligent response method, device, computer equipment, and storage medium for insurance consulting. Background Technology

[0002] In modern insurance consultation scenarios, large language models (such as GPT-3 and GPT-4) are often used to generate corresponding responses based on the user's inquiry information, thereby helping users understand insurance-related information, especially about the types of insurance they are interested in.

[0003] However, current methods for generating responses using language models suffer from several drawbacks. The inability to determine the context of the user's inquiry leads to inaccurate understanding and makes it difficult to provide detailed and accurate answers. Furthermore, the responses often appear abrupt. Additionally, the large size of language models makes it inefficient to generate responses for each user's single inquiry. Summary of the Invention

[0004] This application provides an intelligent response method, apparatus, device, and medium for insurance consultations, which aims to accurately understand the consultation information and provide detailed and correct answers based on the context of the consultation statement, thereby improving the user's consultation experience.

[0005] In a first aspect, embodiments of this application provide an intelligent response method for insurance inquiries, including:

[0006] Obtain insurance consultation statements and generate consultation content tags and consultation sentence style tags based on the insurance consultation information. Consultation content tags represent the content to be consulted in the insurance consultation statement, and consultation sentence style tags represent the sentence style corresponding to the insurance consultation statement.

[0007] Determine the contextual information corresponding to the insurance consultation statements based on the consultation content tags;

[0008] Based on contextual information and consultation sentence style tags, a virtual context is generated for insurance consultation statements, and the insurance consultation statements and the virtual context are merged to generate a target fusion statement;

[0009] Generate target response information corresponding to the consultation content tags based on the target fusion statement;

[0010] The target response statement is generated based on the target response information and the consultation sentence tag, and then output to the target terminal.

[0011] In some implementations, the contextual information corresponding to the insurance consultation statement is determined based on the consultation content tag, including:

[0012] Based on the consultation content tags, insurance category sub-tags and consultation category sub-tags are generated. The insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, and the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement.

[0013] Determine the first contextual features corresponding to the insurance category sub-label, and determine the second contextual features corresponding to the consultation category sub-label;

[0014] Contextual information is obtained by integrating the features of the first context and the features of the second context.

[0015] In some implementations, a virtual context is generated for insurance consultation statements based on contextual information and consultation sentence style tags, including:

[0016] Based on the primary contextual features and consultation sentence style tags in the contextual information, reference passages are searched from a pre-set insurance text database;

[0017] Based on the secondary contextual features in the contextual information, determine the target replacement statement and its location coordinates corresponding to the insurance consultation statement in the reference text.

[0018] The target text range is determined based on the location coordinates, and the virtual context is extracted from the reference text based on the target text range.

[0019] In some implementations, generating target response information corresponding to consultation content tags based on the target fusion statement includes:

[0020] Search the pre-set insurance knowledge database for candidate responses that match the tags of the consultation content;

[0021] Determine the response confidence level of multiple candidate response information based on the target fusion statement;

[0022] The target response information is determined from the candidate response information based on the response confidence level.

[0023] In some implementations, candidate response information matching the consultation content tags is searched in a preset insurance knowledge database, including:

[0024] Based on the consultation content tags, insurance category sub-tags and consultation category sub-tags are generated. The insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, and the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement.

[0025] Search the insurance knowledge database for candidate hot words that match the insurance category sub-tags;

[0026] Generate keyword category sub-tags for the candidate keywords;

[0027] Candidate response information is generated based on the matching degree between the consultation category sub-tag and the hot word category sub-tag.

[0028] In some implementations, the response confidence of multiple candidate response information is determined based on the target fusion statement, including:

[0029] The target fusion statement is segmented to generate multiple preliminary vocabulary words, which include first-category words that appear in the insurance consultation statement and second-category words that do not appear in the insurance consultation statement.

[0030] The second category of words is filtered to obtain the target words;

[0031] Generate a mapping vector between the first type of vocabulary and the target vocabulary. The mapping vector represents the relationship category between the first type of vocabulary and the target vocabulary in the insurance scenario.

[0032] Response confidence is generated based on the mapping vector and the candidate response information.

[0033] In some implementations, a target response statement is generated based on the target response information and the consultation sentence style tag, including:

[0034] Match the corresponding response sentence based on the query sentence type tag;

[0035] The target response information is restructured according to the response sentence pattern to obtain the target response sentence.

[0036] Secondly, embodiments of this application also provide an intelligent response device for insurance consultation, comprising:

[0037] The tag generation module is used to obtain insurance consultation statements and generate consultation content tags and consultation sentence structure tags based on the insurance consultation information. Consultation content tags represent the content to be consulted in the insurance consultation statement, and consultation sentence structure tags represent the sentence structure corresponding to the insurance consultation statement.

[0038] The simulated context module is used to determine the context information corresponding to insurance consultation statements based on the consultation content tags.

[0039] The statement fusion module is used to generate a virtual context for insurance consultation statements based on contextual information and consultation sentence style tags, and to fuse the insurance consultation statements and the virtual context to generate the target fused statement;

[0040] The intelligent response module is used to generate target response information corresponding to the consultation content tags based on the target fusion statement;

[0041] The statement output module is used to generate a target response statement based on the target response information and the consultation sentence tag, and output the target response statement to the target terminal.

[0042] Thirdly, embodiments of this application also provide a computer device, which includes a memory and a processor;

[0043] Memory, used to store computer programs;

[0044] A processor is configured to execute a computer program and, when executing the computer program, implement the insurance consultation intelligent response method provided in any embodiment of this application specification.

[0045] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to implement the insurance consultation intelligent response method provided in any embodiment of this application specification.

[0046] This application provides an intelligent response method, apparatus, device, and medium for insurance consultations. The intelligent response method includes: acquiring an insurance consultation statement and generating consultation content tags and consultation sentence structure tags based on the consultation information. The consultation content tags represent the content to be inquired about in the insurance consultation statement, and the consultation sentence structure tags represent the sentence structure corresponding to the insurance consultation statement. The method also includes: determining the contextual information corresponding to the insurance consultation statement based on the consultation content tags; generating a virtual context for the insurance consultation statement based on the contextual information and the consultation sentence structure tags, and fusing the insurance consultation statement and the virtual context to generate a target fused statement; generating target response information corresponding to the consultation content tags based on the target fused statement; generating a target response statement based on the target response information and the consultation sentence structure tags, and outputting the target response statement to a target terminal. This simulates the context corresponding to the insurance consultation statement to accurately understand its semantics and provide a detailed and correct answer, thereby improving the user's consultation experience. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a schematic diagram illustrating an application scenario of an intelligent response method for insurance consultation provided in an embodiment of this application.

[0049] Figure 2 A schematic flowchart illustrating the steps of an intelligent response method for insurance consultation provided in this application embodiment;

[0050] Figure 3 A schematic diagram of the target fusion statement, insurance consultation statement, and virtual context in an intelligent response method for insurance consultation provided in an embodiment of this application;

[0051] Figure 4 A schematic flowchart illustrating the steps of generating target response information in an intelligent response method for insurance consultation provided in this application embodiment;

[0052] Figure 5 A schematic flowchart illustrating the steps of generating a differential medical examination report package in an intelligent response method for insurance consultation provided in this application embodiment;

[0053] Figure 6 A schematic diagram of the module structure of an intelligent insurance consultation response device provided in this application embodiment;

[0054] Figure 7 This is a schematic block diagram of a computer device provided in an embodiment of this application. Detailed Implementation

[0055] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0056] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0057] In modern insurance consultation scenarios, large language models (such as GPT-3 and GPT-4) are often used to generate corresponding responses based on the user's inquiry information, thereby helping users understand insurance-related information, especially about the types of insurance they are interested in.

[0058] However, current methods for generating responses using language models suffer from several drawbacks. The inability to determine the context of the user's inquiry leads to inaccurate understanding and makes it difficult to provide detailed and accurate answers. Furthermore, the responses often appear abrupt. Additionally, the large size of language models makes it inefficient to generate responses for each user's single inquiry.

[0059] Based on this, embodiments of this application provide an intelligent response method, apparatus, device, and medium for insurance consultations, which aims to accurately understand consultation information and provide detailed and correct answers by simulating the context of the consultation statement based on the consultation statement, thereby improving the user's consultation experience.

[0060] The intelligent response method for insurance consultation can be applied to electronic devices such as computers, intelligent robots, independent servers, or server clusters, without limitation.

[0061] In this embodiment, the application of the intelligent insurance consultation response method to an independent server is used as an example for illustration, but it is not limited to the intelligent insurance consultation response method being used only in this scenario.

[0062] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0063] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating an application scenario of an intelligent response method for insurance consultation provided in an embodiment of this application.

[0064] like Figure 1 As shown, the target terminal establishes a communication connection with the server via wired or wireless means. The server is located in the background, and the target terminal is used for user operation. Figure 1 User A and / or User B can output insurance consultation statements to the target terminal or directly operate the target terminal, so that the target terminal sends insurance consultation statements that match the user's consultation intentions to the server. The server then generates a target response statement based on the insurance consultation statement and sends the target response statement to the target terminal so that the user can know the target response statement, thereby realizing intelligent response to insurance consultations.

[0065] Furthermore, users A and B can interact with the server through the same target terminal, or they can interact with the server through their respective target terminals to achieve intelligent response to insurance inquiries.

[0066] Please see Figure 2 , Figure 2 This is a schematic flowchart illustrating the steps of an intelligent response method for insurance consultation provided in an embodiment of this application.

[0067] like Figure 2 As shown, the intelligent response method for insurance consultation specifically includes the following steps S1-S5.

[0068] Step S1: Obtain insurance consultation statements and generate consultation content tags and consultation sentence structure tags based on the insurance consultation information. Consultation content tags represent the content to be consulted in the insurance consultation statement, and consultation sentence structure tags represent the sentence structure corresponding to the insurance consultation statement.

[0069] It should be noted that the insurance consultation statement is a natural language input by the user. The server executing this method generates consultation content tags and consultation sentence structure tags based on the insurance consultation information. Specifically, it performs semantic recognition on the insurance consultation information and determines the consultation content tags and consultation sentence structure tags based on the recognized semantics. It should be understood that the consultation content tag represents the content to be inquired about in the insurance consultation statement, and the consultation sentence structure tag represents the corresponding sentence structure of the insurance consultation statement.

[0070] For example, a user sends an insurance inquiry statement to the server through the target terminal:

[0071] "I would like to know the coverage of travel insurance."

[0072] When generating consultation content tags and consultation sentence tags, the server determines the consultation content of the insurance consultation statement as "the scope of coverage of travel insurance" based on semantic recognition, and then generates the corresponding consultation content tag; while it determines the consultation sentence of the insurance consultation statement as "I would like to know..." based on semantic recognition, and then generates the corresponding consultation sentence tag.

[0073] For example, a user sends an insurance inquiry statement to the server through the target terminal:

[0074] What is the longest coverage period for Health Insurance Plan A?

[0075] If the server determines that the content of the insurance consultation statement is "the coverage period of health insurance package A" based on semantic recognition, then it generates a corresponding consultation content tag; if the server determines that the consultation statement is a question with a clear consultation purpose based on semantic recognition, then it generates a corresponding consultation sentence tag.

[0076] Step S2: Determine the contextual information corresponding to the insurance consultation statement based on the consultation content tags.

[0077] In some implementations, step S2 determines the contextual information corresponding to the insurance consultation statement based on the consultation content tag, including:

[0078] Based on the consultation content tags, insurance category sub-tags and consultation category sub-tags are generated. The insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, and the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement.

[0079] Determine the first contextual features corresponding to the insurance category sub-label, and determine the second contextual features corresponding to the consultation category sub-label;

[0080] Contextual information is obtained by integrating the features of the first context and the features of the second context.

[0081] Specifically, the first contextual feature represents the context corresponding to the type of insurance, while the second contextual feature represents the context corresponding to the consultation category. For example, based on the consultation content tag "coverage of travel insurance", an insurance category sub-tag corresponding to "travel insurance" and a consultation category sub-tag corresponding to "coverage" can be generated.

[0082] After identifying the first and second contextual features, the first and second contextual features are fused together to obtain contextual information. For example, the contextual information is obtained by fusing the insurance type sub-label "travel insurance" and the consultation category sub-label "coverage".

[0083] Specifically, when fusing the first contextual feature and the second contextual feature, a first feature vector and a second feature vector corresponding to the first contextual feature and the second contextual feature respectively can be generated, and the first feature vector and the second feature vector can be associated and stored as contextual information.

[0084] Step S3: Generate a virtual context for the insurance consultation statement based on contextual information and consultation sentence style tags, and merge the insurance consultation statement and the virtual context to generate the target fused statement.

[0085] Please see Figure 3 , Figure 3 This is a schematic diagram of the target fusion statement, insurance consultation statement, and virtual context in an intelligent response method for insurance consultation provided in an embodiment of this application.

[0086] like Figure 3 As shown, a virtual context for insurance consultation statements is generated based on contextual information and consultation sentence tags. The virtual context includes the preceding text that is relatively positioned before the insurance consultation statement and the following text that is relatively positioned after the insurance consultation statement.

[0087] In some implementations, a virtual context is generated for insurance consultation statements based on contextual information and consultation sentence style tags, including:

[0088] Based on the primary contextual features and consultation sentence style tags in the contextual information, reference passages are searched from a pre-set insurance text database;

[0089] Based on the secondary contextual features in the contextual information, determine the target replacement statement and its location coordinates corresponding to the insurance consultation statement in the reference text.

[0090] The target text range is determined based on the target replacement statement and its location coordinates, and a virtual context is extracted from the reference text based on the target text range.

[0091] Specifically, the insurance text database stores the first sample text, which is a pre-input response statement with first context features and consultation sentence style tags.

[0092] When generating the virtual context, the server executing this method first searches for a matching reference text in the sample text stored in the insurance text library based on the first context feature and the consultation sentence tag. That is, the statement in the sample text matches the insurance category of the insurance consultation statement. Then, based on the second context feature in the context information, it determines the target replacement statement and its position coordinates corresponding to the insurance consultation statement in the reference text. Based on the target replacement statement and position coordinates, it determines the range of the target text and extracts the virtual context from the reference text based on the range of the target text.

[0093] It should be understood that the target replacement statement is one or more statements in the reference text that correspond to the insurance consultation statement. After the target fusion statement is generated, the target replacement statement in the reference text is replaced with the insurance consultation statement.

[0094] For example, the target replacement statement and its position coordinates corresponding to the insurance consultation statement are determined in the reference text based on the second context features in the context information. Specifically, this includes: finding statements in the reference text that match the second context features and consultation sentence tags of the insurance consultation statement as target replacement statements. That is, the target replacement statement has the same consultation category as the insurance consultation statement or the similarity is greater than a first preset threshold, and the target replacement statement has the same consultation sentence structure as the insurance consultation statement or the similarity is greater than a second preset threshold. Therefore, it matches the content that the user wants to inquire about regarding the insurance category.

[0095] like Figure 3 As shown, after generating the virtual context, the insurance consultation statement is merged with the virtual context to generate the target merged statement. For example, merging the insurance consultation statement with the virtual context to generate the target merged statement includes: based on the positional relationship between the insurance consultation statement and its preceding and following text in the virtual context, the insurance consultation statement is pieced together with the virtual context to generate the target merged statement.

[0096] Therefore, by simulating the context of insurance consultation statements based on the insurance consultation statements, the semantics of the insurance consultation statements output by the user can be accurately understood.

[0097] Step S4: Generate target response information corresponding to the consultation content tags based on the target fusion statement.

[0098] Please see Figure 4 , Figure 4 This is a schematic flowchart illustrating the steps of generating target response information in an intelligent response method for insurance consultation provided in this application embodiment.

[0099] like Figure 4 As shown, in some embodiments, step S4 generates target response information corresponding to the consultation content tags based on the target fusion statement, including steps S41-S43:

[0100] Step S41: Search the preset insurance knowledge database for candidate response information that matches the consultation content tags;

[0101] Step S42: Determine the response confidence of multiple candidate response information based on the target fusion statement;

[0102] Step S43: Determine the target response information from the candidate response information based on the response confidence level.

[0103] Specifically, the insurance knowledge database stores a second sample text, which consists of pre-input statements tagged with consultation content. The server executing this method first searches the pre-set insurance knowledge database for statements matching the consultation content tags as candidate response information. Then, it determines the response confidence level of multiple candidate response information based on the target fusion statement. Finally, based on the response confidence level, it determines the target response information from the candidate response information, for example, selecting a pre-set number of candidate response information with the highest response confidence level as the target response information.

[0104] It should be understood that the response confidence level of the candidate response information represents the degree of similarity between the candidate response information and the target fusion statement. A higher response confidence level indicates that the candidate response information and the target fusion statement are semantically similar, while a lower response confidence level indicates that the candidate response information and the target fusion statement are semantically different. By analyzing the response confidence level of the candidate response information in conjunction with the target fusion statement, the response with higher confidence level can be selected.

[0105] In some implementations, candidate response information matching the consultation content tags is searched in a preset insurance knowledge database, including:

[0106] Based on the consultation content tags, insurance category sub-tags and consultation category sub-tags are generated. The insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, and the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement.

[0107] Search the insurance knowledge database for candidate hot words that match the insurance category sub-tags;

[0108] Generate keyword category sub-tags for the candidate keywords;

[0109] Candidate response information is generated based on the matching degree between the consultation category sub-tag and the hot word category sub-tag.

[0110] It should be understood that the insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, while the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement. Specifically, after generating the insurance category sub-tag and the consultation category sub-tag, the system searches the insurance knowledge database for candidate hot words that match the insurance category sub-tag, generates hot word category sub-tags for these candidate hot words, and generates candidate response information based on the matching degree between the consultation category sub-tag and the hot word category sub-tag.

[0111] Specifically, generating candidate response information based on the matching degree between consultation category sub-tags and hot word category sub-tags includes: calculating the matching degree between the hot word category sub-tags and consultation category sub-tags for each candidate hot word, selecting a preset number of candidate hot words with the highest matching degree, and combining them as target response information.

[0112] Please see Figure 5 , Figure 5 This application provides a schematic flowchart illustrating the steps involved in determining the confidence level of a response in generating target response information for an intelligent response method for insurance consultation.

[0113] In some implementations, step S42 determines the response confidence of multiple candidate response information based on the target fusion statement, including:

[0114] Step S421: Perform word segmentation on the target fusion statement to generate multiple preliminary words. The preliminary words include first-category words that appear in the insurance consultation statement and second-category words that do not appear in the insurance consultation statement.

[0115] For example, multiple consecutive characters in the target fusion sentence are identified, and the target fusion sentence is segmented to obtain vocabulary preparatory units. Then, useless words in the vocabulary preparatory units are filtered to obtain multiple preparatory words.

[0116] Furthermore, the preset word segmentation methods include, but are not limited to, using third-party word segmentation tools or algorithms. Common third-party word segmentation tools include, but are not limited to, Stanford NLP word segmenter, ICTClAS word segmentation system, ansj word segmentation tool, and HanLP Chinese word segmentation tool. Word segmentation algorithms include, but are not limited to, Maximum Matching (MM), Reverse Direction Maximum Matching (RMM), Bi-directional Matching (BM), Hidden Markov Model (HMM), and N-gram models.

[0117] Specifically, if the pre-existing vocabulary appears in the insurance consultation statement, it is determined to be a Class I vocabulary; conversely, if the pre-existing vocabulary does not appear in the insurance consultation statement, it is determined to be a Class II vocabulary.

[0118] Step S422: Filter the second type of words to obtain the target words.

[0119] It is easy to understand that extracting target words from the second category of words can filter out some meaningless words, which is beneficial for limiting the scope of key information extraction based on the multiple preparatory words obtained later.

[0120] Step S423: Generate a mapping vector between the first type of vocabulary and the target vocabulary. The mapping vector represents the relationship category between the first type of vocabulary and the target vocabulary in the insurance scenario.

[0121] Specifically, the mapping vector represents the relationship category between the first category of words and the target words in an insurance scenario. For example, in the insurance inquiry statement "I want to understand the coverage of travel insurance," the first category of words includes: "understand," "travel insurance," and "coverage," where...

[0122] Step S424: Generate response confidence based on the mapping vector and candidate response information.

[0123] Specifically, response confidence represents the degree of similarity between the candidate response information and the target fused statement. It identifies the consultation category sub-label and hot word category sub-label of the candidate response information. If the relationship category of the first category of words and the target words in the insurance scenario is the same as or similar to the consultation category sub-label and hot word category sub-label of the candidate response information, then the response confidence is high.

[0124] Step S5: Generate the target response statement based on the target response information and the consultation sentence tag, and output the target response statement to the target terminal.

[0125] In some implementations, a target response statement is generated based on the target response information and the consultation sentence style tag, including:

[0126] Match the corresponding response sentence based on the query sentence type tag;

[0127] The target response information is restructured according to the response sentence pattern to obtain the target response sentence.

[0128] For example, for the insurance inquiry statement "I would like to know the coverage of travel insurance.", steps S1 to S4 can obtain target response information including "travel insurance", "medical expenses", "travel accident injury", "lost luggage" and "flight delay". Therefore, after obtaining the target response information, it is necessary to reorganize the target response information. Thus, the corresponding response sentence is matched according to the inquiry sentence tag, and the target response information is reorganized according to the response sentence to obtain the target response sentence.

[0129] For example, the server executing this method restructures the target response information "travel insurance," "medical expenses," "travel accident injury," "lost baggage," and "flight delay" according to the response sentence pattern, resulting in the target response sentence: "Our travel insurance provides coverage for medical expenses, travel accident injury, lost baggage, and flight delays."

[0130] like Figure 1 As shown, the target response statement is output to the target terminal after the target response statement.

[0131] Therefore, based on the target fusion statement generated from the context of the simulated insurance consultation statement, the semantics of the insurance consultation statement output by the user are accurately understood, and a detailed and correct answer is given.

[0132] In summary, this invention provides an intelligent response method for insurance consultations, comprising: acquiring an insurance consultation statement, and generating consultation content tags and consultation sentence structure tags based on the insurance consultation information. The consultation content tags represent the content to be consulted in the insurance consultation statement, and the consultation sentence structure tags represent the sentence structure corresponding to the insurance consultation statement; determining the contextual information corresponding to the insurance consultation statement based on the consultation content tags; generating a virtual context for the insurance consultation statement based on the contextual information and the consultation sentence structure tags, and merging the insurance consultation statement and the virtual context to generate a target fused statement; generating target response information corresponding to the consultation content tags based on the target fused statement; generating a target response statement based on the target response information and the consultation sentence structure tags, and outputting the target response statement to the target terminal. This simulates the corresponding context of the insurance consultation statement to accurately understand the consultation information and provide a detailed and correct answer, thereby improving the user's consultation experience.

[0133] Please see Figure 6 , Figure 6 This is a schematic diagram of the module structure of an intelligent insurance consultation response device provided in an embodiment of this application.

[0134] like Figure 6 As shown, the insurance consultation intelligent response device 600 includes:

[0135] The tag generation module 601 is used to obtain insurance consultation statements and generate consultation content tags and consultation sentence tags based on the insurance consultation information. The consultation content tags represent the content to be consulted in the insurance consultation statement, and the consultation sentence tags represent the sentence structure corresponding to the insurance consultation statement.

[0136] The simulated context module 602 is used to determine the context information corresponding to the insurance consultation statement based on the consultation content tags.

[0137] The statement fusion module 603 is used to generate a virtual context for insurance consultation statements based on contextual information and consultation sentence tags, and to fuse the insurance consultation statements and the virtual context to generate a target fused statement;

[0138] The intelligent response module 604 is used to generate target response information corresponding to the consultation content tags based on the target fusion statement;

[0139] The statement output module 605 is used to generate a target response statement based on the target response information and the consultation sentence tag, and output the target response statement to the target terminal.

[0140] In some implementations, the audio recording extraction module 702 generates a video to be processed corresponding to the target person based on the dialogue recording and the dialogue video, specifically including:

[0141] Extract the valid audio track from the dialogue recording;

[0142] Obtain the target voiceprint features corresponding to the target person, and extract the dialogue audio track from the effective audio track based on the target voiceprint features;

[0143] The dialogue audio track and dialogue video are used to determine the dialogue image track, and the dialogue audio track and dialogue image track are synthesized to obtain the video to be processed.

[0144] In some implementations, when the differential comparison module 604 generates at least one differential package of medical examination reports corresponding to each release priority based on the differential data, it includes:

[0145] Based on differential data, identify the medical examination data change items in the target medical examination items where there are discrepancies between the target medical examination information and the reference medical examination information;

[0146] Determine the data change information corresponding to the changed items in the physical examination data;

[0147] Package at least one data change message with the same release priority into a differential package of medical examination reports.

[0148] In some implementations, the instruction parsing module 601 determines the release priority of the target physical examination item, including:

[0149] Retrieve pre-stored physical examination report browsing history of terminal devices and corresponding user preference items;

[0150] Based on the browsing history of the physical examination report and the corresponding user's preference items, the target physical examination items are assigned the first priority coefficient;

[0151] Assign a second priority coefficient to the target medical examination items based on the target medical examination information corresponding to the target medical examination items;

[0152] The release priority of target physical examination items is determined based on the first priority coefficient and the second priority coefficient.

[0153] In some implementations, the instruction parsing module 601 assigns a second priority coefficient to the target medical examination item based on the target medical examination information corresponding to the target medical examination item, including:

[0154] Extract the actual medical values ​​from the target physical examination information;

[0155] Obtain the health range of the medical values ​​corresponding to the target physical examination items;

[0156] The second priority coefficient of the target physical examination item whose actual medical value exceeds the corresponding medical value health range is set as the first value;

[0157] The second priority coefficient of the target physical examination item whose actual medical value does not exceed the corresponding medical value health range is set as the second value;

[0158] Among them, the first value is greater than or equal to the second value.

[0159] In some implementations, the version lookup module 602 determines a reference medical examination report from a set of historical medical examination report versions, including:

[0160] Retrieve version information of historical versions of physical examination reports stored in the terminal device;

[0161] Based on the version information, match it with multiple historical versions of the medical examination report to determine at least one candidate version of the medical examination report, and determine the reference medical examination report from the candidate versions of the medical examination report.

[0162] In some implementations, the version lookup module 602 determines a reference medical examination report from the candidate versions of the medical examination report, including:

[0163] The target physical examination information is compared differentially with the candidate versions of the physical examination report to generate differential test data;

[0164] The number of target physical examination items that differ from the target physical examination information and the candidate versions of the physical examination report is determined based on the differential test data.

[0165] The medical examination report with the fewest items should be selected as the reference medical examination report.

[0166] In some implementations, the version lookup module 602 determines a reference medical examination report from the candidate versions of the medical examination report, including:

[0167] Obtain the release time corresponding to each candidate version of the physical examination report;

[0168] The latest version of the physical examination report will be used as the reference report.

[0169] Please see Figure 7 , Figure 7 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application.

[0170] like Figure 7 As shown, the computer device 700 includes a processor 701 and a memory 702, which are connected via a bus 703, such as an I2C (Inter-integrated Circuit) bus.

[0171] Specifically, processor 701 provides computing and control capabilities to support the operation of the entire computer device. Processor 701 can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0172] Specifically, the memory 702 can be a Flash chip, a read-only memory (ROM) disk, an optical disk, a USB flash drive, or a portable hard drive, etc.

[0173] Those skilled in the art will understand that Figure 7 The structures shown are merely block diagrams of some structures related to the embodiments of this application and do not constitute a limitation on the computer devices to which the embodiments of this application are applied. Specific computer devices may include more or fewer components than those shown in the figures, or combine certain components, or have different component arrangements.

[0174] The processor is used to run a computer program stored in the memory, and implements any of the insurance consultation intelligent response methods provided in the embodiments of this application when executing the computer program.

[0175] In some embodiments, processor 701 is used to run a computer program stored in memory 702, and performs the following steps when executing the computer program:

[0176] When a pre-set terminal device outputs a medical examination report retrieval instruction, the instruction is parsed to determine the corresponding target medical examination items and the release priority of the target medical examination items;

[0177] Based on the instructions to retrieve the physical examination report, find the corresponding user's set of historical physical examination reports, and determine the reference physical examination report from the multiple historical physical examination reports contained in the set.

[0178] Find the target physical examination information corresponding to the target physical examination item in the physical examination data to be released, and find the reference physical examination information corresponding to the target physical examination item in the reference physical examination report;

[0179] The target physical examination information is compared with the reference physical examination information to generate differential data. Based on the differential data, at least one differential package of physical examination report corresponding to each release priority is generated.

[0180] Based on the release priority, the differential packages of each physical examination report are sequentially released to the terminal device, so that the terminal device can load the target physical examination information based on the received differential packages of the physical examination reports.

[0181] In some implementations, the processor 701 generates at least one differential package of medical examination reports corresponding to each release priority based on the differential data, including:

[0182] Based on differential data, identify the medical examination data change items in the target medical examination items where there are discrepancies between the target medical examination information and the reference medical examination information;

[0183] Determine the data change information corresponding to the changed items in the physical examination data;

[0184] Package at least one data change message with the same release priority into a differential package of medical examination reports.

[0185] In some implementations, processor 701 determines the release priority of the target health checkup items, including:

[0186] Retrieve pre-stored physical examination report browsing history of terminal devices and corresponding user preference items;

[0187] Based on the browsing history of the physical examination report and the corresponding user's preference items, the target physical examination items are assigned the first priority coefficient;

[0188] Assign a second priority coefficient to the target medical examination items based on the target medical examination information corresponding to the target medical examination items;

[0189] The release priority of target physical examination items is determined based on the first priority coefficient and the second priority coefficient.

[0190] In some implementations, the processor 701 assigns a second priority coefficient to the target medical examination item based on the target medical examination information corresponding to the target medical examination item, including:

[0191] Extract the actual medical values ​​from the target physical examination information;

[0192] Obtain the health range of the medical values ​​corresponding to the target physical examination items;

[0193] The second priority coefficient of the target physical examination item whose actual medical value exceeds the corresponding medical value health range is set as the first value;

[0194] The second priority coefficient of the target physical examination item whose actual medical value does not exceed the corresponding medical value health range is set as the second value;

[0195] Among them, the first value is greater than or equal to the second value.

[0196] In some implementations, the processor 701 determines a reference medical examination report from a set of historical medical examination report versions, including:

[0197] Retrieve version information of historical versions of physical examination reports stored in the terminal device;

[0198] Based on the version information, match it with multiple historical versions of the medical examination report to determine at least one candidate version of the medical examination report, and determine the reference medical examination report from the candidate versions of the medical examination report.

[0199] In some implementations, the processor 701 determines a reference medical examination report from a pool of candidate versions of the medical examination report, including:

[0200] The target physical examination information is compared differentially with the candidate versions of the physical examination report to generate differential test data;

[0201] The number of target physical examination items that differ from the target physical examination information and the candidate versions of the physical examination report is determined based on the differential test data.

[0202] The medical examination report with the fewest items should be selected as the reference medical examination report.

[0203] In some implementations, the processor 701 determines a reference medical examination report from a pool of candidate versions of the medical examination report, including:

[0204] Obtain the release time corresponding to each candidate version of the physical examination report;

[0205] The latest version of the physical examination report will be used as the reference report.

[0206] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the computer equipment described above can be referred to the corresponding process in the aforementioned embodiment of the intelligent response method for insurance consultation, and will not be repeated here.

[0207] This application also provides a storage medium storing a computer program that can be executed by one or more processors to implement the steps of any of the insurance consultation intelligent response methods provided in the specification of this application.

[0208] The storage medium can be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The storage medium can also be an external storage device of the computer device, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0209] It will be understood by those skilled in the art that all or some of the steps, systems, or apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware embodiments, the division between functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0210] In the description of this application, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0211] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0212] It should also be understood that the term "and / or" as used in this specification and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. It should be noted that, herein, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0213] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. The above descriptions are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A smart response method for insurance consultation, characterized in that, The method includes: Obtain insurance consultation statements and generate consultation content tags and consultation sentence tags based on the insurance consultation information. The consultation content tags represent the content to be consulted in the insurance consultation statement, and the consultation sentence tags represent the sentence structure corresponding to the insurance consultation statement. Determine the contextual information corresponding to the insurance consultation statement based on the consultation content tags; Based on the contextual information and the consultation sentence tags, a virtual context is generated for the insurance consultation statement, and the insurance consultation statement and the virtual context are merged to generate a target fusion statement; Generate target response information corresponding to the consultation content tags based on the target fusion statement; Based on the target response information and the consultation sentence tag, a target response statement is generated and output to the target terminal. The step of determining the contextual information corresponding to the insurance consultation statement based on the consultation content tag includes: Based on the consultation content tags, insurance category sub-tags and consultation category sub-tags are generated, wherein the insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, and the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement; Determine the first contextual feature corresponding to the insurance category sub-label, and determine the second contextual feature corresponding to the consultation category sub-label; The contextual information is obtained by fusing the first contextual feature and the second contextual feature; The process of generating a virtual context for the insurance consultation statement based on the contextual information and the consultation sentence style tags includes: Based on the first contextual feature in the contextual information and the consultation sentence tag, a reference passage is searched from a preset insurance text database; Based on the second contextual features in the contextual information, determine the target replacement statement and its position coordinates corresponding to the insurance consultation statement in the reference text. The target text segment range is determined based on the target replacement statement and the location coordinates, and the virtual context is extracted from the reference text segment according to the target text segment range; The process of fusing the insurance consultation statement with the virtual context to generate the target fused statement includes: The target replacement statement in the reference text is replaced with the insurance consultation statement, and the insurance consultation statement and the virtual context are pasted together according to the positional relationship between the virtual context and the insurance consultation statement to generate the target fusion statement.

2. The method according to claim 1, characterized in that, The step of generating target response information corresponding to the consultation content tags based on the target fusion statement includes: Search the preset insurance knowledge database for candidate response information that matches the consultation content tags; Determine the response confidence level of multiple candidate response information based on the target fusion statement; The target response information is determined from the candidate response information based on the response confidence level.

3. The method according to claim 2, characterized in that, The step of searching for candidate response information that matches the consultation content tags in a preset insurance knowledge database includes: Based on the consultation content tags, insurance category sub-tags and consultation category sub-tags are generated, wherein the insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, and the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement; Search the insurance knowledge database for candidate hot words that match the insurance category sub-tag; Generate hot word category sub-tags for the candidate hot words; The candidate response information is generated based on the matching degree between the consultation category sub-tag and the hot word category sub-tag.

4. The method according to claim 2, characterized in that, The step of determining the response confidence of multiple candidate response information based on the target fusion statement includes: The target fusion statement is segmented to generate multiple preliminary vocabulary words, wherein the preliminary vocabulary words include a first category of vocabulary words that appear in the insurance consultation statement and a second category of vocabulary words that do not appear in the insurance consultation statement; The second type of vocabulary is filtered to obtain the target vocabulary; Generate a mapping vector between the first type of vocabulary and the target vocabulary, wherein the mapping vector represents the relationship category between the first type of vocabulary and the target vocabulary in an insurance scenario; The response confidence score is generated based on the mapping vector and the candidate response information.

5. The method according to any one of claims 1-4, characterized in that, The process of generating a target response statement based on the target response information and the consultation sentence tag includes: Match the corresponding response sentence based on the consultation sentence type tag; The target response information is restructured according to the response pattern to obtain the target response statement.

6. An intelligent response device for insurance consultation, characterized in that, include: The tag generation module is used to obtain insurance consultation statements and generate consultation content tags and consultation sentence tags based on the insurance consultation information. The consultation content tags represent the content to be consulted in the insurance consultation statement, and the consultation sentence tags represent the sentence structure corresponding to the insurance consultation statement. The simulated context module is used to determine the context information corresponding to the insurance consultation statement based on the consultation content tags. The statement fusion module is used to generate a virtual context for the insurance consultation statement based on the context information and the consultation sentence tag, and to fuse the insurance consultation statement and the virtual context to generate a target fused statement; The intelligent response module is used to generate target response information corresponding to the consultation content tags based on the target fusion statement; The statement output module is used to generate a target response statement based on the target response information and the consultation sentence tag, and output the target response statement to the target terminal; The step of determining the contextual information corresponding to the insurance consultation statement based on the consultation content tag includes: Based on the consultation content tags, insurance category sub-tags and consultation category sub-tags are generated, wherein the insurance category sub-tag represents the type of insurance corresponding to the insurance consultation statement, and the consultation category sub-tag represents the consultation item corresponding to the insurance consultation statement; Determine the first contextual feature corresponding to the insurance category sub-label, and determine the second contextual feature corresponding to the consultation category sub-label; The contextual information is obtained by fusing the first contextual feature and the second contextual feature; The process of generating a virtual context for the insurance consultation statement based on the contextual information and the consultation sentence style tags includes: Based on the first contextual feature in the contextual information and the consultation sentence tag, a reference passage is searched from a preset insurance text database; Based on the second contextual features in the contextual information, determine the target replacement statement and its position coordinates corresponding to the insurance consultation statement in the reference text. The target text segment range is determined based on the target replacement statement and the location coordinates, and the virtual context is extracted from the reference text segment according to the target text segment range; The process of fusing the insurance consultation statement with the virtual context to generate the target fused statement includes: The target replacement statement in the reference text is replaced with the insurance consultation statement, and the insurance consultation statement and the virtual context are pasted together according to the positional relationship between the virtual context and the insurance consultation statement to generate the target fusion statement.

7. A computer device, characterized in that, The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is configured to execute the computer program and, in executing the computer program, implement the insurance consultation intelligent response method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, causes the processor to implement the insurance consultation intelligent response method as described in any one of claims 1 to 5.

Images