Work assisting method and device, medium and product

Abstract

The invention discloses a work assisting method and device, a medium and a product. Relates to the field of big data and the field of artificial intelligence, in particular to application of a large model in the field of financial science and technology, and the method comprises the following steps: in response to a work assistance request for a target salesman, determining a candidate dialogue text corresponding to the target salesman, and performing text preprocessing on the candidate dialogue text by adopting a large language model; determining a business scene corresponding to the candidate dialogue text, and desensitizing and screening the candidate dialogue text according to the business scene and the identity type of the target salesman to obtain a target dialogue text; and determining a target analysis text according to the target dialogue text and the business data corresponding to the business scene, and generating a structured abstract corresponding to the target analysis text by adopting a large language model to assist related salesmen in business processing work. According to the invention, the salesman can be assisted in business processing, and the business processing efficiency is improved.

Description

Technical Field

[0001] This invention relates to the fields of big data and artificial intelligence, specifically to the application of large models in the field of financial technology, and particularly to a work assistance method, device, medium, and product. Background Technology

[0002] In banking scenarios, sales staff frequently communicate about business matters through office automation systems and various instant messaging tools. These exchanges often present pain points such as information fragmentation, difficulties in cross-modal processing, security risks, and efficiency bottlenecks. Specifically, the large volume and dispersed nature of chat messages from office automation systems, emails, and other channels make it easy to overlook key requirements and task milestones, leading to information fragmentation; the diverse nature of business communication content makes it difficult to overlook key requirements and task milestones, resulting in difficulties in cross-modal processing; the financial industry's requirements for sensitive data anonymization and access control raise high security risks; and manual processing is time-consuming and error-prone, leading to efficiency bottlenecks.

[0003] Therefore, how to analyze work documents and business data to assist sales staff in their business processing and improve efficiency is an urgent problem to be solved. Summary of the Invention

[0004] This invention provides a work assistance method, apparatus, medium, and product for analyzing work texts and business data, assisting salespersons in business processing, and improving business processing efficiency.

[0005] According to one aspect of the present invention, a work assistance method is provided, comprising:

[0006] In response to a work assistance request from a target salesperson, candidate dialogue texts corresponding to the target salesperson are identified, and a large language model is used to preprocess the candidate dialogue texts.

[0007] Determine the business scenario corresponding to the candidate dialogue text, and based on the business scenario and the identity type of the target salesperson, perform anonymization and filtering on the candidate dialogue text to obtain the target dialogue text;

[0008] Based on the target dialogue text and the business data corresponding to the business scenario, the target analysis text is determined, and a large language model is used to generate a structured summary corresponding to the target analysis text to assist relevant business personnel in their business processing work.

[0009] According to another aspect of the present invention, a work assistance device is provided, comprising:

[0010] The preprocessing module is used to respond to the work assistance request of the target salesperson, determine the candidate dialogue text corresponding to the target salesperson, and perform text preprocessing on the candidate dialogue text using a large language model.

[0011] The module is used to determine the business scenario corresponding to the candidate dialogue text, and to perform desensitization and filtering on the candidate dialogue text according to the business scenario and the identity type of the target salesperson, so as to obtain the target dialogue text.

[0012] The auxiliary module is used to determine the target analysis text based on the target dialogue text and the business data corresponding to the business scenario, and to generate a structured summary of the target analysis text using a large language model to assist relevant business personnel in their business processing work.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] At least one processor; and

[0015] A memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the work assistance method described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the work assistance method described in any embodiment of the present invention.

[0018] According to another aspect of the present invention, a computer program product is also provided, the computer program product including a computer program that, when executed by a processor, implements the working assistance method of any embodiment of the present invention.

[0019] The technical solution of this invention, in response to a work assistance request from a target salesperson, determines candidate dialogue text corresponding to the target salesperson and preprocesses the candidate dialogue text using a large language model; it determines the business scenario corresponding to the candidate dialogue text and, based on the business scenario and the identity type of the target salesperson, performs de-identification and filtering processing on the candidate dialogue text to obtain the target dialogue text; based on the target dialogue text and the business data corresponding to the business scenario, it determines the target analysis text and, using a large language model, generates a structured summary corresponding to the target analysis text to assist the relevant salesperson in business processing. By comprehensively analyzing the salesperson's work dialogue text using a large language model and further combining it with business data to provide a structured summary, it can assist the salesperson in business processing and improve business processing efficiency.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, 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 the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a work assistance method provided in Embodiment 1 of the present invention;

[0023] Figure 2 This is a flowchart of a work assistance method provided in Embodiment 2 of the present invention;

[0024] Figure 3 This is a structural block diagram of a work assistance device provided in Embodiment 3 of the present invention;

[0025] Figure 4 This is a schematic diagram of the structure of the electronic device provided in Embodiment 4 of the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first," "second," "target," "candidate," and "alternative," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the invention described herein can be practiced in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The acquisition, storage, use, and processing of data in the technical solutions of this application comply with relevant laws and regulations.

[0028] It should be noted that the user information collected in this invention is information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of this data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, and necessary confidentiality measures have been taken. This process does not violate public order and good morals, and corresponding operation entry points are provided for users to choose to authorize or reject automated decision-making results. If the user chooses to reject, the process proceeds to the expert decision-making process. In other words, the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for display, data used for analysis, etc.) involved in this invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of this data comply with the relevant laws, regulations, and standards of the relevant regions.

[0029] Example 1

[0030] Figure 1 This is a flowchart of a work assistance method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where a comprehensive analysis of a salesperson's work dialogue text is performed using a large language model, and a structured summary is further provided in conjunction with business data to assist the salesperson in business processing. This method can be executed by a work assistance device, which can be implemented in hardware and / or software. The work assistance device can be configured in an electronic device, such as a banking business system. Figure 1 As shown, this work assistance method includes:

[0031] S101. In response to the work assistance request for the target salesperson, determine the candidate dialogue text corresponding to the target salesperson, and use a large language model to preprocess the candidate dialogue text.

[0032] Among them, the target salesperson refers to the person who conducts business processing in the banking business system. The work assistance request refers to the request for comprehensively analyzing the conversation text and relevant business data of the target salesperson during the business processing process to generate a structured summary to assist the target salesperson in handling business.

[0033] The candidate conversation text refers to the conversation text generated by the target salesperson using office software in the banking business system. The large language model refers to a preset natural language processing model. The text preprocessing includes at least one of the following: Chinese word segmentation, stop word filtering, and synonym normalization.

[0034] Optionally, a word segmentation tool of the banking business custom dictionary can be used to perform Chinese word segmentation on the candidate conversation text to ensure the complete segmentation of professional vocabulary.

[0035] Exemplarily, general function words (such as "de", "le") can be removed, but business sensitive words (such as "urgent", "high risk") are retained for stop word filtering.

[0036] Optionally, synonym normalization can be performed based on the banking business synonym library. For example, if the word "lending" appears in the candidate conversation text, it can be normalized to "credit approval" based on the banking business synonym library. If the word "system downtime" appears in the candidate conversation text, it can be normalized to "system failure" based on the banking business synonym library.

[0037] S102. Determine the business scenario corresponding to the candidate conversation text, and perform desensitization and screening processing on the candidate conversation text according to the business scenario and the identity type of the target salesperson to obtain the target conversation text.

[0038] Among them, the business scenario includes at least one of the following: system failure type, anti-money laundering monitoring type, credit approval type, and requirement change type; the identity type is customer manager, credit approval specialist, anti-money laundering specialist, system administrator, or external auditor; the target conversation text refers to the text obtained after performing desensitization and screening processing on the candidate conversation text.

[0039] Optionally, determining the business scenario corresponding to the candidate conversation text includes at least one of the following:

[0040] 1) Use the large language model to process the candidate conversation text to determine the business scenario corresponding to the candidate conversation text.

[0041] Optionally, the candidate conversation text can be directly input into the pre-trained large language model to output the business scenario corresponding to the candidate conversation text.

[0042] 2) Generate regular expressions based on the high-frequency business words in the candidate business scenarios and the structural relationships between them, and use the regular expressions to match the candidate dialogue text to determine the business scenario corresponding to the candidate dialogue text.

[0043] Optionally, based on the candidate business scenarios, common high-frequency business terms and their structural relationships in subsequent business scenarios can be determined to construct regular expressions. These high-frequency business terms can include terms like "credit," "approval," "expedited processing," and "process," with the order of these terms representing the structural relationships.

[0044] 3) Based on the candidate dialogue text, match it in the business terminology database, and combine the scenario weights of each proper noun in the business terminology database under different business scenarios to determine the business scenario corresponding to the candidate dialogue text.

[0045] The business terminology library can pre-store the scenario weights of different proper nouns in different business scenarios. For example, the banking term "loan value ratio" has a weight of >0.9 in credit approval business scenarios and a weight of <0.3 in payment scenarios.

[0046] For example, if the proper noun "loan value ratio" is matched in the business terminology library in the candidate dialogue text, it can be preliminarily determined that the business scenario corresponding to the candidate dialogue text is credit approval.

[0047] Optionally, for candidate dialogue text, the business scenario corresponding to the candidate dialogue text can be determined by the three methods mentioned above. If the business scenario determined by the three methods is the same, the business scenario corresponding to the candidate dialogue text is finally determined to be the business scenario determined by the three methods mentioned above, so as to ensure the accuracy of business scenario determination. Alternatively, at least one of the three methods mentioned above can be used to determine the business scenario in order to improve the efficiency of business scenario determination. This invention does not limit this.

[0048] Optionally, when different business scenarios are used as described above, the final business scenario can be determined based on a preset scenario optimization level.

[0049] For example, if the candidate dialogue text is "Cryptocurrency payment module error, error code 205, transaction volume surge causes service unavailability", then based on the business scenario determination method described in step 3) above, combined with the proper noun "cryptocurrency", the business scenario can be initially determined to be anti-money laundering monitoring. Further, by identifying "error code 205" through the large language model, the business scenario can be initially determined to be system failure. Finally, combined with the preset scenario optimization level, the final business scenario can be determined to be anti-money laundering monitoring.

[0050] It should be noted that the above-mentioned technical solution of the present invention can perform a more comprehensive analysis of the dialogue text by using a large model, analyzing the relationship between business terms, or combining the weights of different proper nouns in the business terminology library to determine the business scenario, thereby quickly and accurately determining the business scenario corresponding to the dialogue text.

[0051] Optionally, based on the business scenario and the target salesperson's identity type, the candidate dialogue text is anonymized and filtered to obtain the target dialogue text, including:

[0052] If the business scenario is a requirement change type and the target salesperson's identity type is account manager, then the customer information in the candidate dialogue text is anonymized, and statements related to requirement change in the candidate dialogue text are filtered out to obtain the target dialogue text.

[0053] For example, the target dialogue text for a salesperson with the identity type of account manager may include information such as the loan approval progress and financial needs records of their clients. In this case, only the client information associated with the target salesperson can be filtered out and bound using a client identity identifier. At the same time, the client information is anonymized, and finally, redundant information such as client assessment information that is irrelevant to this business process is discarded.

[0054] Optionally, if the business scenario is credit approval and the target salesperson's identity type is credit approval specialist, the corresponding target dialogue text may include loan application materials, collateral appraisal reports, credit inquiry results, etc. In this case, the corresponding filtering rules may be: retaining and displaying statements with the guarantee validity field, hiding the internal notes of the third-party appraisal agency, and automatically archiving the keyword "repayment source" in the associated chat history.

[0055] If the business scenario is anti-money laundering monitoring and the target salesperson's identity type is anti-money laundering specialist, then the transaction amount anonymization operation will not be performed, and the associated text of low-risk ordinary transactions in the candidate dialogue text will be removed for filtering to obtain the target dialogue text.

[0056] For example, in a business scenario involving anti-money laundering monitoring, and where the target salesperson's identity type is anti-money laundering specialist, the target dialogue text may include: suspicious transaction reports, cross-border payment logs, high-risk customer monitoring lists, etc. In this case, the corresponding filtering rules may be: removing the anonymization of cryptocurrency transaction amounts (unique permissions), associating multimodal data (transaction logs + dialogue text), and filtering low-risk ordinary transaction records.

[0057] For example, in a business scenario involving system failures, and where the target salesperson's identity is that of a system administrator, the target dialogue text may include server monitoring metrics, interface logs, infrastructure warnings, etc. In this case, the corresponding filtering rules may be: open all technical operation and maintenance data, anonymize business content (such as "customer service disabled" replacing real names), and discard business discussion details (such as credit policy adjustments).

[0058] For example, in the case of an external audit scenario, where the target salesperson's identity type is an external auditor, the target dialogue text may include operation logs during the audit period, anonymized transaction samples, etc. In this case, the corresponding filtering rules may be: timeliness restrictions (such as data only visible within the audit period), secondary anonymization: amount → range (such as 100,000-500,000) and prohibition of access to chat content.

[0059] It should be noted that the above-mentioned technical solution of the present invention provides specific methods for desensitization and screening under different business scenarios and identity types, which can more accurately screen the dialogue text and extract the effective text in the dialogue text that is applicable to the business scenario and conforms to the scope of the salesperson's work, thus helping to improve the efficiency and accuracy of subsequent text processing.

[0060] S103. Based on the target dialogue text and the business data corresponding to the business scenario, determine the target analysis text, and use a large language model to generate a structured summary corresponding to the target analysis text to assist relevant business personnel in their business processing work.

[0061] The target analysis text is the text obtained by integrating the target dialogue text and business data. Business data refers to other collected data related to the business scenario. Business data includes at least one of the following: log data, transaction record data, and server monitoring data. The structured summary contains the text topic corresponding to the target dialogue text, the responsible party corresponding to the text topic, the time node, and the action instructions. The relevant salesperson refers to the salesperson involved in the responsible party of the structured summary, which may or may not include the target salesperson.

[0062] Optionally, the target analysis text is determined based on the target dialogue text and the business data corresponding to the business scenario, including: determining the business data corresponding to the business scenario; and integrating the target dialogue text and the business data corresponding to the business scenario to determine the target analysis text.

[0063] Specifically, when the business scenario is anti-money laundering monitoring, the corresponding business data is transaction record data and log data; when the business scenario is system failure, the corresponding business data is server monitoring data.

[0064] It should be noted that the above-mentioned technical solution of the present invention can quickly determine the business data corresponding to different business scenarios by providing the business data that needs to be further collected and determined under different business scenarios, and obtain the final target analysis text. This can realize the association between business scenarios and business data, which helps to improve the accuracy of subsequent text analysis.

[0065] Optionally, after determining the target analysis text, the process further includes: verifying the integrity of the target analysis text based on the business rule base corresponding to the business scenario, and analyzing whether any key steps in the business scenario are missing during the business processing based on the verification results; if so, a large language model is used to generate a structured summary corresponding to the target analysis text.

[0066] The business rule base corresponding to the business scenario can contain a list of necessary steps to complete the business under that scenario. Completeness verification is used to verify whether the target analysis text fully incorporates the list of necessary steps corresponding to the business rule base.

[0067] Optionally, based on the business rule base corresponding to the business scenario, the completeness verification of the target analysis text is performed, including: if the business scenario is credit approval, then the business rule base corresponding to the business scenario is determined to be the credit approval rule base; the credit approval rule base includes at least one of the following key steps: customer qualification verification, collateral assessment, risk scoring, repayment ability verification, and guarantee validity check; and whether the target analysis text contains keywords of the key steps in the credit approval rule base is determined to verify the completeness of the target analysis text.

[0068] For example, when the business scenario is credit approval, the corresponding business rule base can be a credit approval rule base. The list of necessary steps to complete the credit approval business stored in the credit approval rule base can be: customer qualification verification (verification of income certificate and credit record); collateral appraisal (asset valuation document); risk scoring (output of internal rating system); repayment ability verification (salary slip / tax certificate); guarantee validity check (guarantor qualification or ownership of collateral).

[0069] Optionally, for each key step, a matching process can be performed in the target analysis text using preset keywords corresponding to the key steps to determine whether the target analysis text contains keywords of key steps in the credit approval rule base. If there are more than a preset number of key step keywords in the target analysis text, the integrity check of the target analysis text is considered to have passed.

[0070] It should be noted that the above-mentioned technical solution of the present invention provides a feasible method for verifying the integrity of target analysis text by giving the mandatory steps that the corresponding business rule base can include in the credit approval scenario, which helps to quickly and accurately generate structured summaries in the credit approval scenario.

[0071] For example, if the target dialogue text is "The customer says the mortgage application materials have been submitted, contract number 2023-089, customer service replies please provide the property certificate and income certificate, and the qualification verification needs to be completed before the end of today's workday," and the risk score is determined to be B based on business data, indicating a repayment ability warning, then by evaluating keywords such as "mortgage application," "contract number," and "collateral appraisal," the business scenario can be determined to be credit approval. Further analysis can identify the mandatory steps in the credit approval rule base, and then use a large language model for analysis to obtain the following statement position weights: the customer service's first sentence "Please provide..." (weight 0.9), the system's last sentence (weight 0.8), and the customer's first sentence (weight 0.6). Key entities identified are contract number, property certificate, income certificate, collateral appraisal, and risk score. The time-sensitive phrase "before the end of today's workday" triggers an emergency flag.

[0072] For example, the final integrity verification result can be: (1) Existing steps (high probability of model output): Customer qualification verification (probability 0.92): Customer service requests proof of income; Collateral valuation (probability 0.88): Risk control system outputs valuation value; Risk score (probability 0.95): Explicitly mentions "Grade B" rating. (2) Missing steps (low probability of model output): Repayment ability verification (probability 0.35): Only prompts "early warning" without providing proof such as salary slips; Guarantee validity check (probability 0.1): Not mentioned at all.

[0073] It should be noted that the above-mentioned technical solution of the present invention utilizes a business rule base to perform integrity verification on the analyzed text, which can quickly identify the key steps missing in completing the business in a business scenario, and help to provide targeted structured summaries to instruct business personnel to carry out further work.

[0074] Optionally, a large language model is used to generate a structured summary corresponding to the target analysis text, including: performing topic clustering on the target analysis text based on dependency parsing strategy to determine the text topic and key sentences corresponding to the target analysis text; and using a large language model to parse the text topic and key sentences to determine the responsible party, time node and action instructions of the target analysis text, so as to generate a structured summary of the missing key steps in the business scenario.

[0075] The text subject can be customer qualification verification, product / solution details inquiry, or application progress inquiry, etc. The responsible party is the person responsible for performing the missing critical steps, the time point is the time when the missing critical steps need to be performed, and the action instructions are the action instructions that need to be performed to perform the missing critical steps.

[0076] Optionally, the keywords involved in the target analysis text can be statistically classified to determine the text theme corresponding to the target analysis text, and the text sentences containing the keywords can be identified as key sentences. For example, the keywords "application status, under review, approved, rejected, materials to be supplemented, review time, estimated completion time, progress query, notification method and SMS reminder" can indicate that the corresponding text theme is application progress query.

[0077] For example, after associating the "risk score B" in the target analysis text with regulatory rules, since a B score requires an additional guarantor, the action instructions in the target analysis text can be determined as adding a guarantor, verifying repayment ability, and checking the validity of the guarantee.

[0078] For example, the process of clustering the target analysis text into topics based on dependency parsing strategies can specifically include: (1) analyzing the grammatical structure of sentences to identify verbs and their directly governed elements such as subjects and objects. Sentences containing core business actions (such as "apply", "query", "submit", "confirm", "reject", "calculate", "inform") and their objects (such as "interest rate", "materials", "credit report", "contract") are usually key sentences. (2) identifying modifiers such as negation words, modal verbs (such as "must", "should", "may"), and degree adverbs, which may change the degree of importance of the statement (such as emphasizing obligation or possibility). (3) introduction of new information: the first mention of important business entities (such as specific product names, amounts, and time points), key facts (such as reasons for rejection, special requirements), and core needs (such as "I need a low interest rate"). (4) status changes: statements describing changes in business status (such as "Your application has been approved" and "Material review has not passed").

[0079] Optionally, text topics and key statements can be input into a large language model to output a structured summary containing the responsible parties, time points, and action instructions of the target analysis text.

[0080] For example, using a large language model, the responsible party can be identified based on dependency analysis, role lexicon, and topic context. For instance, in the sentence "Customer service requests you to provide proof of income," the subject "customer service" is the responsible party (the requesting party), and the implicit object "you" is also the responsible party (the executing party). That is, for each topic, the key action instructions are clearly marked to indicate "who" (the responsible party) needs to perform or be responsible for them.

[0081] For example, using a large language model, time nodes can be determined based on time entity recognition, logical reasoning, and topic association. First, the explicit time in key statements, such as absolute time, relative time, and time descriptive words, is determined. Then, the logical temporal relationship between the actions described in the key statements is analyzed. Finally, for each key action or state change under a topic, "when" (time node / requirement / deadline) is marked as occurring or needing to be completed.

[0082] For example, using a large language model, action instructions can be determined based on core predicate extraction, semantic induction, and topic aggregation. Specifically, the predicate verbs (and necessary objects / complements) expressing the core operation can be extracted from key sentences, and the "what to do" (specific, standardized, and unambiguous action instructions) under each topic can be clearly extracted.

[0083] For example, large language models can be used to effectively filter out purely repetitive or low-information-content statements based on semantic similarity calculation, information freshness / completeness comparison, and topic consistency, ensuring that every piece of information in the final summary is necessary and non-redundant.

[0084] Optionally, a large language model is used to parse key sentences corresponding to each text topic. Dependency analysis, semantic recognition, and reasoning are employed to determine the responsible party, time point, and action instructions in the target analysis text. These elements are then logically assembled into a clear and concise declarative sentence or structured entry. For example: "[The customer] needs to [upload proof of income and bank statements] within [3 working days]".

[0085] It should be noted that the above-mentioned technical solution of the present invention generates a structured summary of the missing key steps in the business scenario through topic clustering and parsing of a large language model. It can provide the responsible party, time node and action instructions for the missing key steps, which helps business people quickly understand the tasks to be performed and assists their work.

[0086] Optionally, relevant sales personnel can be identified based on the responsible party, and a structured summary can be displayed to them to assist them in processing business based on the structured summary.

[0087] The technical solution of this invention, in response to a work assistance request from a target salesperson, determines candidate dialogue text corresponding to the target salesperson and preprocesses the candidate dialogue text using a large language model; it determines the business scenario corresponding to the candidate dialogue text and, based on the business scenario and the identity type of the target salesperson, performs de-identification and filtering processing on the candidate dialogue text to obtain the target dialogue text; based on the target dialogue text and the business data corresponding to the business scenario, it determines the target analysis text and, using a large language model, generates a structured summary corresponding to the target analysis text to assist the relevant salesperson in business processing. By comprehensively analyzing the salesperson's work dialogue text using a large language model and further combining it with business data to provide a structured summary, it can assist the salesperson in business processing and improve business processing efficiency.

[0088] Example 2

[0089] Figure 2 This is a flowchart of a work assistance method provided in Embodiment 2 of the present invention; based on the above embodiments, this embodiment provides a preferred example of analyzing work texts and business data to assist business personnel in business processing, such as... Figure 2 As shown, the method includes:

[0090] S201. In response to the work assistance request for the target salesperson, determine the candidate dialogue text corresponding to the target salesperson, and use a large language model to preprocess the candidate dialogue text.

[0091] S202. Based on the high-frequency business words in the candidate business scenarios and the structural relationships between the high-frequency business words, generate regular expressions and use regular expressions to match the candidate dialogue text to determine the business scenario corresponding to the candidate dialogue text.

[0092] S203. If the business scenario is a requirement change, and the target salesperson's identity type is account manager, then the customer information in the candidate dialogue text is anonymized, and statements related to requirement change are filtered out from the candidate dialogue text to obtain the target dialogue text.

[0093] S204. If the business scenario is anti-money laundering monitoring and the target salesperson's identity type is anti-money laundering specialist, then the transaction amount anonymization operation will not be performed, and the associated text of low-risk ordinary transactions in the candidate dialogue text will be removed for filtering to obtain the target dialogue text.

[0094] S205. Determine the business data corresponding to the business scenario, and integrate the target dialogue text and the business data corresponding to the business scenario to determine the target analysis text.

[0095] S206. Based on the business rule base corresponding to the business scenario, perform integrity verification on the target analysis text, and based on the verification results, analyze whether any key steps in the business scenario are missing during the business processing.

[0096] S207. If so, then based on the dependency parsing strategy, the target analysis text is clustered into topics to determine the text topics and key sentences corresponding to the target analysis text.

[0097] S208. Use a large language model to parse the text topic and key sentences, determine the responsible party, time node and action instructions of the target analysis text, so as to generate a structured summary of the missing key steps in the business scenario.

[0098] S209. Use structured summaries to assist relevant business personnel in their business processing work.

[0099] Example 3

[0100] Figure 3 This is a structural block diagram of a work assistance device provided in Embodiment 3 of the present invention. This embodiment is applicable to situations where a large language model is used to comprehensively analyze the work dialogue text of a salesperson, and further combined with business data to provide a structured summary, in order to assist the salesperson in performing business processing work. The work assistance device provided in this embodiment of the present invention can execute the work assistance method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method. This work assistance device can be implemented in hardware and / or software and configured in an electronic device with work assistance functions, such as a banking business system. Figure 3 As shown, the work assistance device may specifically include:

[0101] The preprocessing module 301 is used to respond to the work assistance request of the target salesperson, determine the candidate dialogue text corresponding to the target salesperson, and perform text preprocessing on the candidate dialogue text using a large language model.

[0102] The module 302 is used to determine the business scenario corresponding to the candidate dialogue text, and to perform desensitization and filtering on the candidate dialogue text according to the business scenario and the identity type of the target salesperson, so as to obtain the target dialogue text.

[0103] The auxiliary module 303 is used to determine the target analysis text based on the target dialogue text and the business data corresponding to the business scenario, and to generate a structured summary corresponding to the target analysis text using a large language model, so as to assist relevant business personnel in carrying out business processing work.

[0104] The technical solution of this invention, in response to a work assistance request from a target salesperson, determines candidate dialogue text corresponding to the target salesperson and preprocesses the candidate dialogue text using a large language model; it determines the business scenario corresponding to the candidate dialogue text and, based on the business scenario and the identity type of the target salesperson, performs de-identification and filtering processing on the candidate dialogue text to obtain the target dialogue text; based on the target dialogue text and the business data corresponding to the business scenario, it determines the target analysis text and, using a large language model, generates a structured summary corresponding to the target analysis text to assist the relevant salesperson in business processing. By comprehensively analyzing the salesperson's work dialogue text using a large language model and further combining it with business data to provide a structured summary, it can assist the salesperson in business processing and improve business processing efficiency.

[0105] Furthermore, the business scenarios include at least one of the following: system failure, anti-money laundering monitoring, credit approval, and requirement change.

[0106] Module 302 is specifically used to perform at least one of the following operations:

[0107] A large language model is used to process candidate dialogue texts to determine the business scenarios corresponding to the candidate dialogue texts.

[0108] Based on the high-frequency business words in the candidate business scenarios and the structural relationships between the high-frequency business words, regular expressions are generated, and the regular expressions are used to match the candidate dialogue text to determine the business scenario corresponding to the candidate dialogue text.

[0109] Based on the candidate dialogue text, a matching process is performed in the business terminology database. By combining the scenario weights of each proper noun in the business terminology database under different business scenarios, the business scenario corresponding to the candidate dialogue text is determined.

[0110] Furthermore, the identity type is account manager, credit approval specialist, anti-money laundering specialist, system administrator, or external auditor;

[0111] Module 302 is also used for:

[0112] If the business scenario is a requirement change type and the target salesperson's identity type is account manager, then the customer information in the candidate dialogue text is anonymized, and statements related to requirement change in the candidate dialogue text are filtered out to obtain the target dialogue text.

[0113] If the business scenario is anti-money laundering monitoring and the target salesperson's identity type is anti-money laundering specialist, then the transaction amount anonymization operation will not be performed, and the associated text of low-risk ordinary transactions in the candidate dialogue text will be removed for filtering to obtain the target dialogue text.

[0114] Furthermore, the auxiliary module 303 is specifically used for:

[0115] Identify the business data corresponding to the business scenario; wherein, the business data includes at least one of the following: log data, transaction record data, and server monitoring data; when the business scenario is anti-money laundering monitoring, the corresponding business data is transaction record data and log data; when the business scenario is system failure, the corresponding business data is server monitoring data.

[0116] The target dialogue text and the corresponding business data of the business scenario are integrated to determine the target analysis text.

[0117] Furthermore, the above-mentioned device is also used for:

[0118] Based on the business rule library corresponding to the business scenario, the completeness of the target analysis text is verified, and based on the verification results, it is analyzed whether any key steps in the business scenario are missing during the business processing.

[0119] If so, a large language model is used to generate a structured summary corresponding to the target analysis text.

[0120] Furthermore, the auxiliary module 303 is also used for:

[0121] Based on dependency parsing strategies, topic clustering is performed on the target analysis text to determine the text topic and key sentences corresponding to the target analysis text.

[0122] A large language model is used to parse the text topic and key sentences, determine the responsible party, time node and action instructions of the target analysis text, and generate a structured summary of the key steps missing in the business scenario.

[0123] Example 4

[0124] Figure 4 This is a schematic diagram of the structure of the electronic device provided in Embodiment 4 of the present invention. Figure 4 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0125] like Figure 4As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory 12 or a random access memory 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the read-only memory 12 or loaded from storage unit 18 into the random access memory 13. The random access memory 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, read-only memory 12, and random access memory 13 are interconnected via a bus 14. An input / output interface 15 is also connected to the bus 14.

[0126] Multiple components in electronic device 10 are connected to input / output 15, including: input unit 16, such as a keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as a disk, optical disk, etc.; and communication unit 19, such as a network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0127] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, central processing units, graphics processing units, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as job-aiding methods.

[0128] In some embodiments, the job-assistance method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via read-only memory 12 and / or communication unit 19. When the computer program is loaded into random access memory 13 and executed by processor 11, one or more steps of the job-assistance method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the job-assistance method by any other suitable means (e.g., by means of firmware).

[0129] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays, application-specific integrated circuits (ASICs), application-specific standard products (ASICs), system-on-a-chip (SoCs), complex programmable logic devices (PLCs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0130] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0131] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0132] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a cathode ray tube or liquid crystal display) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (e.g., voice input, speech input, or tactile input).

[0133] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0134] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system to address the shortcomings of traditional physical hosts and virtual reality services, such as high management difficulty and weak business scalability.

[0135] In one embodiment, the present invention further includes a computer program product, which includes a computer program that, when executed by a processor, implements the working assistance method of any embodiment of the present invention.

[0136] In the implementation of a computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​as well as conventional procedural programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including local area networks (LANs) or wide area networks (WANs), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0137] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0138] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A work assistance method, characterized in that, include: In response to a work assistance request from a target salesperson, candidate dialogue texts corresponding to the target salesperson are identified, and a large language model is used to preprocess the candidate dialogue texts. Determine the business scenario corresponding to the candidate dialogue text, and based on the business scenario and the identity type of the target salesperson, perform anonymization and filtering on the candidate dialogue text to obtain the target dialogue text; Based on the target dialogue text and the business data corresponding to the business scenario, the target analysis text is determined, and a large language model is used to generate a structured summary corresponding to the target analysis text to assist relevant business personnel in their business processing work.

2. The method according to claim 1, characterized in that, The business scenarios include at least one of the following: system failure, anti-money laundering monitoring, credit approval, and requirement change. Identify the business scenario corresponding to the candidate dialogue text, including at least one of the following: A large language model is used to process candidate dialogue texts to determine the business scenarios corresponding to the candidate dialogue texts. Based on the high-frequency business words in the candidate business scenarios and the structural relationships between the high-frequency business words, regular expressions are generated, and the regular expressions are used to match the candidate dialogue text to determine the business scenario corresponding to the candidate dialogue text. Based on the candidate dialogue text, a matching process is performed in the business terminology database. By combining the scenario weights of each proper noun in the business terminology database under different business scenarios, the business scenario corresponding to the candidate dialogue text is determined.

3. The method according to claim 1, characterized in that, in, The identity type is account manager, credit approval specialist, anti-money laundering specialist, system administrator, or external auditor; Based on the business scenario and the target salesperson's identity type, the candidate dialogue text is anonymized and filtered to obtain the target dialogue text, including: If the business scenario is a requirement change type and the target salesperson's identity type is account manager, then the customer information in the candidate dialogue text is anonymized, and statements related to requirement change in the candidate dialogue text are filtered out to obtain the target dialogue text. If the business scenario is anti-money laundering monitoring and the target salesperson's identity type is anti-money laundering specialist, then the transaction amount anonymization operation will not be performed, and the associated text of low-risk ordinary transactions in the candidate dialogue text will be removed for filtering to obtain the target dialogue text.

4. The method according to claim 1, characterized in that, Based on the target dialogue text and the business data corresponding to the business scenario, determine the target analysis text, including: Identify the business data corresponding to the business scenario; wherein, the business data includes at least one of the following: log data, transaction record data, and server monitoring data; when the business scenario is anti-money laundering monitoring, the corresponding business data is transaction record data and log data; when the business scenario is system failure, the corresponding business data is server monitoring data. The target dialogue text and the corresponding business data of the business scenario are integrated to determine the target analysis text.

5. The method according to claim 1, characterized in that, After identifying the target analysis text, the following is also included: Based on the business rule library corresponding to the business scenario, the completeness of the target analysis text is verified, and based on the verification results, it is analyzed whether any key steps in the business scenario are missing during the business processing. If so, a large language model is used to generate a structured summary corresponding to the target analysis text.

6. The method according to claim 5, characterized in that, Based on the business rule base corresponding to the business scenario, perform integrity verification on the target analysis text, including: If the business scenario is credit approval, then the business rule base corresponding to the business scenario is determined to be the credit approval rule base; the credit approval rule base includes at least one of the following key steps: customer qualification verification, collateral valuation, risk scoring, repayment ability verification, and guarantee validity check; To verify the completeness of the target analysis text, we need to determine whether it contains keywords related to key steps in the credit approval rule base.

7. The method according to claim 1, characterized in that, Using a large language model, a structured summary corresponding to the target analysis text is generated, including: Based on dependency parsing strategies, topic clustering is performed on the target analysis text to determine the text topic and key sentences corresponding to the target analysis text. A large language model is used to parse the text topic and key sentences, determine the responsible party, time node and action instructions of the target analysis text, and generate a structured summary of the key steps missing in the business scenario.

8. A work auxiliary device, characterized in that, include: The preprocessing module is used to respond to the work assistance request of the target salesperson, determine the candidate dialogue text corresponding to the target salesperson, and perform text preprocessing on the candidate dialogue text using a large language model. The module is used to determine the business scenario corresponding to the candidate dialogue text, and to perform desensitization and filtering on the candidate dialogue text according to the business scenario and the identity type of the target salesperson, so as to obtain the target dialogue text. The auxiliary module is used to determine the target analysis text based on the target dialogue text and the business data corresponding to the business scenario, and to generate a structured summary of the target analysis text using a large language model to assist relevant business personnel in their business processing work.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the work assistance method according to any one of claims 1-7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the work assistance method according to any one of claims 1-7.

Images