Cross-border service intelligent interaction method and device, equipment and storage medium

Abstract

The invention discloses a cross-border service intelligent interaction method, device and equipment and a storage medium, and relates to the technical field of cross-border businesses, and the method comprises the steps: receiving cross-border service request information inputted by a user; analyzing the cross-border service request information through a multistage intention recognition mechanism to obtain a service intention type corresponding to the cross-border service request information; determining a corresponding cross-border business processing module according to the business intention type, wherein the cross-border business processing module comprises at least one of a business question and answer module, a deposit inquiry module, a foreign exchange inquiry module and a shortcut tool calling module; and executing a corresponding service processing operation on the cross-border service request information through a cross-border service processing module, and generating a service processing result fed back to the user. By integrating cross-border service functions, user operation paths are reduced, and the learning cost of green users is reduced; the intention recognition precision is improved by means of a multi-level intention recognition mechanism, the accuracy of a service processing result is improved, and the cross-border service processing efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of cross-border business technology, and in particular to a cross-border business intelligent interaction method, apparatus, device and storage medium. Background Technology

[0002] Cross-border business scenarios typically involve many functions that are scattered. During the process, users need to access multiple professional systems to operate, which not only leads to high learning costs and difficulty for novice users to get started, making it easy to make mistakes, but also makes the business inquiry or processing process time-consuming and laborious due to the lengthy operation path.

[0003] Therefore, improving the efficiency and accuracy of cross-border business processing has become an urgent problem to be solved. Summary of the Invention

[0004] The main objective of this application is to provide a cross-border business intelligent interaction method, apparatus, device, and storage medium, aiming to solve the technical problem of how to improve the processing efficiency and accuracy of cross-border business.

[0005] To achieve the above objectives, this application proposes a cross-border business intelligent interaction method, which includes: Receive cross-border business request information input by the user; The cross-border business request information is parsed through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; The corresponding cross-border business processing module is determined according to the business intent type. The cross-border business processing module includes at least one of the following: a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool call module. The cross-border business processing module performs corresponding business processing operations on the cross-border business request information and generates business processing results that are fed back to the user.

[0006] In one embodiment, the step of parsing the cross-border business request information through a multi-level intent recognition mechanism includes: The cross-border business request information is input into the rule engine for preliminary intent matching to obtain preliminary matching results; If the preliminary matching result meets the first preset threshold, the business intent type corresponding to the cross-border business request information is determined based on the preliminary matching result.

[0007] In one embodiment, before the step of inputting the cross-border business request information into the rule engine for preliminary intent matching, the method further includes: Configure the matching rules of the rule engine, including the regular expression, keyword set and fuzzy search keywords corresponding to the shortcut tool calling module; The matching rules are hot-updated to ensure they are adapted to real-time cross-border business needs.

[0008] In one embodiment, after the step of inputting the cross-border business request information into the rule engine for preliminary intent matching, the method further includes: If the preliminary matching result does not meet the first preset threshold, the cross-border business request information is input into the semantic recognition model for vector similarity calculation to obtain the semantic matching result; If the semantic matching result meets the second preset threshold, the business intent type corresponding to the cross-border business request information is determined based on the semantic matching result; If the semantic matching result does not meet the second preset threshold, the cross-border business request information is input into a large language model for contextual semantic parsing, and the business intent type corresponding to the cross-border business request information is determined based on the model parsing result.

[0009] In one embodiment, the step of performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module includes: When the cross-border business processing module is the business question and answer module, the cross-border business request information is converted into a summary query vector; The summary query vector is input into a preset business knowledge base to perform parallel vector retrieval and obtain multi-path recall results; the preset business knowledge base includes a cross-border document library, a cross-border question and answer library, and a fallback knowledge base. The cross-border business request information and the multi-path recall results that have undergone preset validity verification are input into the large language model for relevance scoring, and a business processing result is generated and fed back to the user based on the scoring result.

[0010] In one embodiment, the step of performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module includes: When the cross-border business processing module is the deposit inquiry module, the key deposit information corresponding to the cross-border business request information is obtained. The key deposit information includes the amount, currency, deposit type and account type. The key deposit information is format-validated to determine whether it conforms to the preset interface format requirements. If the key deposit information meets the preset interface format requirements, the foreign currency deposit quotation interface is called based on the key deposit information to obtain deposit quotation information, and a business processing result is generated and fed back to the user based on the deposit quotation information.

[0011] In one embodiment, the step of performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module includes: When the cross-border business processing module is the foreign exchange query module, the key currency information and currency information tags corresponding to the cross-border business request information are obtained; The exchange rate query interface is invoked based on the key currency information to obtain the corresponding real-time foreign currency exchange rate information; Call the information query interface to obtain the corresponding foreign exchange information based on the currency information tags; Based on the real-time foreign exchange rate information and the foreign exchange information, a business processing result is generated and fed back to the user.

[0012] Furthermore, to achieve the above objectives, this application also proposes a cross-border business intelligent interaction device, the device comprising: The data receiving module is used to receive cross-border business request information input by the user; The intent parsing module is used to parse the cross-border business request information through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; A business processing module is used to determine the corresponding cross-border business processing module according to the business intent type. The cross-border business processing module includes at least one of a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool call module. The interactive feedback module is used to perform corresponding business processing operations on the cross-border business request information through the cross-border business processing module, and generate business processing results to be fed back to the user.

[0013] In addition, to achieve the above objectives, this application also proposes a cross-border business intelligent interaction device, the device comprising: a memory, a processor, and a cross-border business intelligent interaction program stored on the memory and executable on the processor, the cross-border business intelligent interaction program being configured to implement the steps of the cross-border business intelligent interaction method mentioned above.

[0014] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a cross-border business intelligent interaction program is stored. When the cross-border business intelligent interaction program is executed by a processor, it implements the steps of the cross-border business intelligent interaction method mentioned above.

[0015] This application provides a cross-border business intelligent interaction method, apparatus, device, and storage medium. The method includes: receiving cross-border business request information input by a user; parsing the cross-border business request information through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; determining the corresponding cross-border business processing module based on the business intent type, the cross-border business processing module including at least one of a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool invocation module; and performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module to generate a business processing result fed back to the user. This application reduces user operation paths and lowers the learning cost for novice users by integrating cross-border business functions such as business Q&A, deposit inquiry, foreign exchange inquiry, and quick tool invocation; improves intent recognition accuracy through a multi-level intent recognition mechanism to ensure accurate invocation of the corresponding business processing module; and improves the accuracy and comprehensiveness of business processing results through the professional processing logic of each module, thereby improving the efficiency of cross-border business processing. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the first embodiment of the intelligent interaction method for cross-border business in this application; Figure 2 This is a schematic diagram of the intent recognition process in the first embodiment of the cross-border business intelligent interaction method of this application; Figure 3 This is a schematic diagram of the interaction process in the first embodiment of the intelligent interaction method for cross-border business of this application; Figure 4 This is a flowchart illustrating the second embodiment of the intelligent interaction method for cross-border business in this application; Figure 5 This is a schematic diagram of the question-and-answer processing process in the second embodiment of the intelligent interaction method for cross-border business in this application; Figure 6 This is a schematic diagram of the deposit inquiry processing procedure in the second embodiment of the intelligent interaction method for cross-border business of this application; Figure 7 This is a schematic diagram of the foreign exchange query processing process in the second embodiment of the intelligent interaction method for cross-border business of this application. Figure 8This is a schematic diagram of the module structure of the intelligent interaction device for cross-border business according to an embodiment of this application; Figure 9 This is a schematic diagram of the device structure of the hardware operating environment involved in the intelligent interaction method for cross-border business in this application embodiment.

[0019] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0020] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0021] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0022] The main solution of this application is as follows: receiving cross-border business request information input by the user; parsing the cross-border business request information through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; determining the corresponding cross-border business processing module according to the business intent type, the cross-border business processing module including at least one of a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool call module; and performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module to generate business processing results fed back to the user.

[0023] Currently, cross-border business suffers from problems such as fragmented functions (users need to access multiple professional systems to operate), inaccurate and incomplete knowledge queries (relying on rule engines and preset QA answers, resulting in low matching accuracy), and low intent recognition accuracy (traditional single recognition methods have poor fault tolerance).

[0024] This application integrates cross-border business functions such as business Q&A, deposit inquiry, foreign exchange inquiry, and quick tool access, reducing user operation paths and lowering the learning cost for novice users; it improves intent recognition accuracy by leveraging a multi-level intent recognition mechanism to ensure accurate invocation of the corresponding business processing module, significantly improving the accuracy and robustness of user intent recognition; and it enhances the accuracy and comprehensiveness of business processing results through the professional processing logic of each module, thereby improving the efficiency of cross-border business processing.

[0025] It should be noted that the executing entity in this embodiment can be a cross-border business intelligent interaction system, or a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone. It can also be a cross-border business intelligent assistant device (such as a dedicated intelligent assistant system deployed on a bank's internal server or terminal device), serving cross-border business personnel such as account managers and product managers at financial institutions like China Merchants Bank, or a cross-border business intelligent interaction device capable of achieving the aforementioned functions. This embodiment does not specifically limit this. The following description uses a cross-border business intelligent interaction device (hereinafter referred to as the interaction device) as the executing entity to illustrate this embodiment and the subsequent embodiments.

[0026] Based on this, the embodiments of this application provide a cross-border business intelligent interaction method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the intelligent interaction method for cross-border business in this application.

[0027] In this embodiment, the intelligent interaction method for cross-border business includes steps S10 to S40: Step S10: Receive cross-border business request information input by the user; It is easy to understand that interactive devices can obtain cross-border business request information input by users through input interfaces (such as keyboard input, voice input modules, touch input interfaces, etc.). The cross-border business request information can be demand content related to cross-border business, and can cover cross-border business scenario demands such as list query, exchange rate query, deposit quotation query, and business knowledge query.

[0028] Step S20: The cross-border business request information is parsed through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; Step S30: Determine the corresponding cross-border business processing module according to the business intent type. The cross-border business processing module includes at least one of the following: business Q&A module, deposit inquiry module, foreign exchange inquiry module, and quick tool call module. Understandably, the aforementioned business intent types can be user demand categories categorized according to cross-border business scenarios, including business Q&A intents (such as querying knowledge related to cross-border accounts), deposit quotation intents (such as inquiring about foreign currency time deposit quotes), foreign exchange query intents (such as querying exchange rates for specific currencies), and quick tool call intents (such as calling a revenue calculator). To avoid the problems of ambiguity and poor fault tolerance caused by a single recognition method, this embodiment proposes a hierarchical recognition mechanism that integrates three recognition methods: a rule engine, a semantic recognition model, and a large language model. This multi-level intent recognition mechanism accurately determines the business intent corresponding to the user's request information through a progressively advancing recognition process.

[0029] Therefore, this embodiment can employ a multi-level intent recognition mechanism to process and semantically analyze the user-input request information, extracting key content (such as business type, query object, etc.) from the request information to determine the user's core needs. In a first feasible implementation, step S20 may include steps S21-S22: Step S21: Input the cross-border business request information into the rule engine for preliminary intent matching to obtain preliminary matching results; Step S22: If the preliminary matching result meets the first preset threshold, determine the business intent type corresponding to the cross-border business request information based on the preliminary matching result.

[0030] It is important to understand that the aforementioned rule engine can be a program component that quickly matches and judges input data through preset matching rules. It can use algorithms such as regular expression matching, keyword matching, and fuzzy search, and is suitable for identifying simple and clear business intents, such as request information containing clear keywords, like "query foreign exchange rates".

[0031] Therefore, in this embodiment, cross-border business request information can be input into the rule engine, and the rule engine can perform preliminary screening and matching of the cross-border business request information according to the preset matching rules to determine whether the request information meets the simple intent feature. The matching similarity value (such as a score of 0-100) or matching result identifier (such as "matched quick tool call intent" or "did not match explicit intent") output by the rule engine after matching is the above-mentioned preliminary matching result.

[0032] The aforementioned first preset threshold can refer to a pre-set critical value (such as 80 points) used to determine whether the preliminary matching result is valid. It can be set by technical personnel based on historical matching data and the accuracy requirements of business needs. If the similarity value of the preliminary matching result is higher than the threshold, the matching is considered valid and the business intent type can be directly determined.

[0033] For example, in this embodiment, the business intent type that meets the first preset threshold can be a shortcut tool call intent. It can be assumed that the user inputs "call revenue calculator," and the interactive device can input this request information into the rule engine. Since the rule engine has pre-configured matching rules corresponding to the shortcut tool call module, such as regular expressions containing the keywords "revenue calculator" and "call + tool name," the rule engine can use a keyword matching algorithm to identify the keywords "call" and "revenue calculator" in the request information "call revenue calculator," calculating a matching similarity score of 95. The interactive device determines that the preliminary matching result (95 points) meets the first preset threshold (e.g., 80 points), thus confirming that the business intent type corresponding to the cross-border business request information is a shortcut tool call intent. Therefore, this embodiment can use the rule engine to perform preliminary intent matching on simple request information, quickly identify clear intents, reduce subsequent complex identification process calls, improve the overall intent identification speed, and ensure a rapid response to simple cross-border business requests.

[0034] Furthermore, in one feasible implementation, in this embodiment, steps A1 to A2 may be included before step A1: Step A1: Configure the matching rules of the rule engine. The matching rules include the regular expression, keyword set, and fuzzy search keywords corresponding to the shortcut tool calling module. Step A2 involves hot-updating the matching rules to ensure they are adapted to real-time cross-border business needs.

[0035] In simple terms, technical personnel can import, edit, and delete matching rules into the rule engine through the system management interface, thus pre-configuring the rule engine. Regular expressions can be special character sequences used to match text patterns, which can match simple request information such as "calling the cross-border whitelist maintenance tool". Keyword sets can be a group of core words related to a specific intent, such as the keyword set corresponding to the intent of calling a shortcut tool, which may include "revenue calculator", "pricing calculator", "appointment remote assistant", etc. Fuzzy search keywords can be keywords that allow for matching even if there are some differences in characters, such as "cross-border whitelist maintenance" which can fuzzily match request information such as "cross-border whitelist maintenance tool" or "cross-border whitelist management".

[0036] It is important to understand that the aforementioned hot update processing refers to real-time updates of the matching rules of the rule engine without interrupting the interactive process. This embodiment can achieve this through remote push of update packages or automatic replacement of rule files in the background, avoiding service interruptions caused by system restarts. The aforementioned real-time cross-border business requirements can refer to new or adjusted cross-border business requirements from financial institutions based on market changes, policy adjustments, and user feedback, such as adding a quick access requirement for "cross-border e-commerce settlement tools" or adjusting keywords for "foreign exchange derivative quote query."

[0037] For example, when a quick access function for "cross-border e-commerce settlement tools" needs to be added, the matching rules can be re-edited, adding the keyword "cross-border e-commerce settlement tools" and the corresponding regular expression to generate a rule update package. This update package can then be received through the backend communication interface, automatically replacing the old rules in the rule engine without interrupting the current service, thus completing a hot update of the matching rules and enabling the rule engine to recognize new requests such as "calling cross-border e-commerce settlement tools".

[0038] Therefore, existing rule engines have fixed matching rules that cannot be updated in a timely manner to meet changing cross-border business needs, resulting in the inability to recognize new business intents and affecting system applicability. This embodiment addresses this issue by pre-configuring targeted matching rules to ensure the rule engine accurately recognizes the intents of existing shortcut tool calls; and by using hot update processing, it ensures that the matching rules are synchronized with real-time cross-border business needs without interrupting system services, thereby improving the flexibility and timeliness of the rule engine and expanding the system's support capabilities for new business scenarios.

[0039] In a second feasible implementation, step S10 may further include steps S23 to S25: Step S23: If the preliminary matching result does not meet the first preset threshold, the cross-border business request information is input into the semantic recognition model for vector similarity calculation to obtain the semantic matching result. Step S24: If the semantic matching result meets the second preset threshold, determine the business intent type corresponding to the cross-border business request information based on the semantic matching result; It should be noted that if the initial matching result does not meet the first preset threshold, this embodiment can further perform intent recognition based on a semantic recognition model. For ease of understanding, please refer to... Figure 2 To illustrate, Figure 2This is a schematic diagram of the intent recognition process in the first embodiment of the intelligent interaction method for cross-border business of this application. The aforementioned semantic recognition model can refer to a model built based on natural language processing (NLP) technology, such as the BERT model, RoBERTa model, etc. It can segment and semantically encode the text of cross-border business request information, convert the text of cross-border business request information into vector form, and then calculate the similarity between the input request information and the preset business intent vector (such as the business question and answer intent vector, deposit inquiry intent vector), thereby achieving semantic intent matching. It is suitable for recognizing request information that has certain differences in expression but similar semantics.

[0040] It is easy to understand, such as Figure 2 As shown, the matching rules in the rule engine and the preset business intent vectors in semantic matching within the interactive device can both be the latest data determined through hot updates. Therefore, when the rule engine cannot determine the specific recognition result, the user-input cross-border business request information can be further converted into a semantic vector, and then cosine similarity calculation, Euclidean distance calculation, and other operations can be performed with the preset business intent vectors in the semantic recognition model. This is the aforementioned vector similarity operation, which yields a vector similarity value. This value is used to determine the degree of semantic association between the request information and each intent, and the semantic matching result is generated based on the vector similarity value and the corresponding candidate business intent type, such as "70% similarity with the business question and answer intent vector and 30% similarity with the deposit inquiry intent vector". The second preset threshold can be a critical value for judging whether the semantic matching result is valid (e.g., 75%), which can be set by technical personnel based on the training accuracy of the semantic recognition model and the needs of the business scenario. If the highest similarity value of the semantic matching result is higher than this threshold, the semantic matching is considered valid.

[0041] Step S25: If the semantic matching result does not meet the second preset threshold, the cross-border business request information is input into a large language model for contextual semantic parsing, and the business intent type corresponding to the cross-border business request information is determined based on the model parsing result.

[0042] Understandably, the aforementioned large language models can be large language models with deep semantic understanding capabilities (such as the GPT series models and LLaMA models). After being trained on large-scale text data, they can parse complex, implicit, and ambiguous request information, capture subtle semantic differences in the text and potential user needs, and are suitable for handling rare or complex requests that are difficult for rule engines and semantic recognition models to cover.

[0043] Therefore, this embodiment can use a large language model combined with the textual context of the request information (such as the user's historical dialogue content and the current business scenario background) to perform in-depth analysis of the text, extract implicit key information (such as the account type that the user did not explicitly mention but can infer), and determine the user's true intent, that is, to perform the aforementioned contextual semantic parsing. At this time, the intent judgment result output by the large language model, that is, the above-mentioned model parsing result, can include a clear business intent type and the key information extraction result that supports the judgment, such as "deposit inquiry intent, key information: amount 1 million, currency A".

[0044] For example, suppose a user enters "I want to know what documents I need to prepare to open an account with Branch A". The interactive device can first perform a preliminary matching of the request information through the rule engine. Since it does not contain preset simple keywords (such as "call" or "exchange rate"), the preliminary matching result has a similarity score of 40, which does not meet the first preset threshold (80 points).

[0045] Then, the interactive device can input the request information into the semantic recognition model. The semantic recognition model performs word segmentation (e.g., "want", "understand", "branch A", "corporate", "open account", "prepare", "materials") on the text, semantically encodes it, converts it into a semantic vector, and then performs cosine similarity calculation with preset "cross-border business Q&A intent vector" and "foreign exchange query intent vector". The similarity value with the cross-border business Q&A intent is 72 points, and the similarity value with other intents is less than 50 points. The semantic matching result is then output.

[0046] At this point, since the semantic matching result (72 points) of the interactive device does not meet the second preset threshold (75 points), the request information can be further input into the large language model. The large language model, combined with the cross-border business scenario background, performs contextual semantic analysis on the request "I want to know what documents I need to prepare for opening a corporate account with Branch A," identifying key information such as "Branch A," "corporate account," "account opening," and "documents." It determines that the user's core need is to query knowledge related to opening a cross-border account and outputs the model's analysis result, i.e., the business intent type is cross-border business question-and-answer intent. Subsequently, the interactive device can determine the business intent type corresponding to this request information as a cross-border business question-and-answer intent based on the model's analysis result.

[0047] In summary, this embodiment utilizes a hierarchical and progressive approach involving a rule engine, a semantic recognition model, and a large language model to cover intent recognition scenarios for simple, medium-complexity, and complex request information, thus overcoming the limitations of traditional single recognition methods. For requests that fail to match using the rule engine and semantic recognition model, accurate recognition is further achieved through deep analysis using the large language model, improving the robustness and accuracy of intent recognition and ensuring that all types of cross-border business requests can be matched with the correct intent type.

[0048] The above are only three feasible implementation methods of step S20 provided in this embodiment. This embodiment does not specifically limit the specific implementation method of step S20.

[0049] Step S40: The cross-border business processing module performs corresponding business processing operations on the cross-border business request information and generates a business processing result that is fed back to the user.

[0050] It is easy to understand that the aforementioned cross-border business processing module can be a collection of program modules pre-configured in the interactive device to implement specific cross-border business functions. This may include at least one of the following: a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool access module. For specific details, please refer to... Figure 3 , Figure 3 This is a schematic diagram of the interaction process in the first embodiment of the intelligent interaction method for cross-border business of this application. Figure 3 The "Business Q&A Module" shown can be used to handle cross-border business knowledge query needs, the "Deposit Inquiry Module" can be used to handle foreign currency deposit quotation query needs, the "Foreign Exchange Inquiry Module" can be used to handle foreign currency exchange rate query needs, and the "Quick Tool Call Module" can provide quick tool call functions such as revenue calculator and cross-border whitelist maintenance tool.

[0051] Furthermore, such as Figure 3 As shown, after preliminary judgment based on rule matching, and intent recognition corresponding to the semantic recognition model and the large language model, the cross-border business processing module to be invoked can be accurately determined. Figure 3 After the cross-border business knowledge Q&A corresponds to the aforementioned business Q&A module, the foreign currency deposit inquiry corresponds to the aforementioned deposit inquiry module, the foreign exchange inquiry corresponds to the aforementioned foreign exchange inquiry module, and the quick tool entry integration corresponds to the aforementioned quick tool call module, each cross-border business processing module in the interactive device can perform data retrieval, calculation, analysis, and other processing on the cross-border business request information input by the user according to its own functional logic, that is, execute the corresponding business processing operation. After the operation is completed, the output content that meets the user's needs, such as Q&A results, quotation information, exchange rate data, tool jump links, and other business processing results, can be transmitted to the user through output interfaces (such as display screen, voice broadcast, etc.), completing the interactive loop.

[0052] This implementation addresses the problems of fragmented cross-border business functions (requiring users to access multiple specialized systems), inaccurate and incomplete knowledge queries (relying on rule engines and preset QA answers with low matching accuracy), and low intent recognition accuracy (traditional single recognition methods have poor fault tolerance). This implementation integrates cross-border business functions such as business Q&A, deposit price inquiries, foreign exchange inquiries, and quick tool access, reducing user operation paths and lowering the learning cost for novice users. It improves intent recognition accuracy through a multi-level intent recognition mechanism, ensuring accurate invocation of the corresponding business processing module. Furthermore, it enhances the accuracy and comprehensiveness of business processing results through the professional processing logic of each module, thereby improving the efficiency of cross-border business processing.

[0053] This embodiment provides a cross-border business intelligent interaction method, which includes: receiving cross-border business request information input by a user; configuring matching rules for a rule engine, including regular expressions, keyword sets, and fuzzy search keywords corresponding to a shortcut tool call module; and performing hot update processing on the matching rules to ensure that the matching rules are adapted to real-time cross-border business needs. The cross-border business request information is input into the rule engine for preliminary intent matching to obtain a preliminary matching result. If the preliminary matching result meets a first preset threshold, the business intent type corresponding to the cross-border business request information is determined based on the preliminary matching result. If the preliminary matching result does not meet the first preset threshold, the cross-border business request information is input into a semantic recognition model for vector similarity calculation to obtain a semantic matching result. If the semantic matching result meets a second preset threshold, the business intent type corresponding to the cross-border business request information is determined based on the semantic matching result. If the semantic matching result does not meet the second preset threshold, the cross-border business request information is input into a large language model for contextual semantic parsing, and the business intent type corresponding to the cross-border business request information is determined based on the model parsing result. The corresponding cross-border business processing module is determined based on the type of business intent. This module includes at least one of the following: a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool invocation module. The cross-border business processing module executes corresponding business processing operations on the cross-border business request information, generating business processing results that are fed back to the user. This addresses the problems of existing cross-border business functions being fragmented (requiring users to access multiple professional systems), inaccurate and incomplete knowledge queries (relying on rule engines and preset QA answers, resulting in low matching accuracy), and low intent recognition accuracy (traditional single recognition methods have poor fault tolerance). This embodiment integrates cross-border business functions such as business Q&A, deposit inquiry, foreign exchange inquiry, and quick tool invocation, reducing user operation paths and lowering the learning cost for novice users; it improves intent recognition accuracy through a multi-level intent recognition mechanism, ensuring accurate invocation of the corresponding business processing module; and it enhances the accuracy and comprehensiveness of business processing results through the professional processing logic of each module, thereby improving the efficiency of cross-border business processing.

[0054] Based on the first embodiment of this application, in the second embodiment of this application, the same or similar content as the first embodiment described above can be referred to the above description, and will not be repeated hereafter.

[0055] It is easy to understand that, based on the first embodiment, please refer to... Figure 4 , Figure 4 This is a flowchart illustrating the second embodiment of the intelligent interaction method for cross-border business of this application. In this embodiment, step S40 includes steps B1 to B3: Step B1: When the cross-border business processing module is the business question and answer module, the cross-border business request information is converted into a summary query vector. Step B2: Input the summary query vector into a preset business knowledge base to perform parallel vector retrieval and obtain multi-path recall results; the preset business knowledge base includes a cross-border document library, a cross-border question and answer library, and a fallback knowledge base; Step B3: Input the cross-border business request information and the multi-path recall results that have passed the preset validity verification into the large language model for relevance scoring, and generate a business processing result to be fed back to the user based on the scoring result.

[0056] Understandably, the aforementioned summary query vector can refer to the vector obtained by extracting the summary of cross-border business request information and converting it through semantic encoding. In this process, the interactive device can use text summarization algorithms (such as TextRank algorithm, BERT-SUM algorithm, etc.) to extract the core content (such as "Branch A's account opening information") from the cross-border business request information, remove redundant expressions, obtain summary information, and then convert the summary information into a vector through a semantic recognition model for subsequent knowledge base retrieval.

[0057] The aforementioned pre-configured business knowledge base can refer to a collection of databases storing cross-border business-related knowledge data. Among them, the cross-border document library can store cross-border business documents in formats such as DOC, PDF, PPT, and EXCEL, such as the "Measures for the Administration of Cross-border Accounts" and the "Trade Finance Business Guide"; the cross-border Q&A library can store pre-configured business Q&A pairs, such as "Q: Conditions for opening an account at a branch in city N; A: Materials such as a business license and the legal representative's ID card are required"; and the "fallback knowledge base" can be a pre-configured basic knowledge base or a user-customized knowledge base, which can store cross-border business knowledge not covered by the above two databases for supplementary retrieval.

[0058] At this point, the interactive device can simultaneously input the summary query vector into the retrieval modules of the cross-border document library, the cross-border question and answer library, and the fallback knowledge base for parallel RAG vector retrieval. Each library retrieval module can calculate the similarity between the summary query vector and the knowledge vectors within the library, executing the retrieval operation in parallel and reducing the overall retrieval time. Since RAG supports training on multiple document types, it can effectively solve the problems of time-consuming and fixed formatting in organizing QA materials, thus broadening the boundaries of business question and answer.

[0059] It is important to understand that the above-mentioned multi-path recall results can be knowledge fragments with high similarity to the summary query vector returned after retrieval from different knowledge bases. Each base can return a preset number of results (such as 5). For example, the cross-border document base returns 5 text fragments about account opening information in the "Branch A Corporate Business Handling Manual", and the cross-border question and answer base returns 5 sets of related question and answer pairs.

[0060] Furthermore, the interactive device can perform preset validity checks on multiple recall results to identify invalid results (such as expired account opening information requirements or knowledge unrelated to the request). During this process, the timeliness (such as whether it is an expired business rule), relevance (such as whether the semantic similarity with the request information is higher than 60 points) and completeness (such as whether it is a complete knowledge fragment) of the knowledge fragment can be determined.

[0061] It should be noted that the aforementioned large language model can be an intelligent processing component integrating a large language model. This component can receive request information and valid recall results, score the recall results for relevance (e.g., based on the degree of matching between knowledge fragments and request information, knowledge authority, etc., using a 100-point scoring system), and integrate and generate logically coherent and complete business processing results based on the scoring results (e.g., selecting the top 10 high-scoring results). During this process, the interactive device can also segment and number candidate knowledge fragments to obtain numbered knowledge sub-fragments. Then, the text request and the numbered knowledge sub-fragments are organized into a prompt text according to a preset template. The prompt text includes instruction fields, knowledge fields, and question fields. Finally, the prompt text is sent to the large language model, enabling the large language model to generate a response text with a reference number based on the knowledge fields. This process can solve the problem of lack of traceability in the generated answers; that is, this implementation can further improve the interpretability and credibility of the answer through a numbered reference mechanism.

[0062] For ease of understanding, please refer to Figure 5 The knowledge question-and-answer retrieval process is illustrated with an example. Figure 5 This is a schematic diagram of the question-and-answer processing process in the second embodiment of the intelligent interaction method for cross-border business of this application. Figure 5 As shown, when the interactive device (i.e. Figure 5 The cross-border business intelligent assistant in the system determines the call to the business Q&A module (i.e., Figure 5When processing results in an intelligent agent, the business question-and-answer module can first perform a summary extraction operation on the user's request information "I want to know what materials are needed to open a business account with Branch A", and then use a preset knowledge base (i.e., Figure 5 (Managing user-pre-built knowledge bases) and the TextRank algorithm (corresponding to) Figure 5 The input information is parsed according to the configuration processing rules shown, and the core summary "Branch A's Account Information" is extracted.

[0063] The business question-and-answer module then inputs the summary into the semantic recognition model, performs text slicing and vectorization to convert it into a summary query vector, and simultaneously inputs this vector into the cross-border document library, the cross-border question-and-answer library, and the fallback knowledge base (i.e., Figure 5 The system manages user-pre-configured service parameters in a search database, and each database performs parallel vector searches. Multiple search results can be obtained at this point. For example, the cross-border document database can retrieve 5 relevant text fragments from the "Branch A Corporate Account Opening Operation Guidelines," the cross-border Q&A database can retrieve 5 sets of Q&A pairs related to "Branch A Account Opening Information," and the fallback knowledge base can retrieve 3 supplementary knowledge entries (the other 2 can be filtered out due to insufficient relevance). Then, a pre-defined validity check can be performed on the above 13 retrieved results: removing 1 expired account opening information requirement and 2 personal account opening knowledge entries unrelated to "corporate banking," resulting in 10 valid retrieved results.

[0064] Then, the user request information is compared with 10 valid recall results for relevance scoring (e.g., the latest account opening document list fragment in the "Branch A Account Opening Operation Guide" scores 95 points, and a certain Q&A gets 88 points). Based on the scoring results, the information is integrated to generate a business processing result (Q&A result) containing "a copy of the business license, the original of the legal representative's ID card, the company's articles of association, and the account opening application form." Finally, the business Q&A module feeds back the business processing result to the interactive device, which then outputs it to the user.

[0065] In this embodiment, existing business knowledge QA systems rely on manually compiled preset QA materials, resulting in limited material and inaccurate or incomplete answers to complex questions, failing to cover diverse cross-border business knowledge query needs. This embodiment addresses these issues by pre-setting multiple types of business knowledge bases to enrich the sources of QA materials, covering most cross-border business knowledge scenarios. Parallel vector retrieval is employed to improve retrieval efficiency, and preset validity checks ensure the quality of retrieved results.

[0066] In addition, the scoring and integration functions can generate accurate, comprehensive, and logically coherent question and answer results, which can overcome the limitations of traditional QA systems and improve the reliability and practicality of business knowledge Q&A.

[0067] In one feasible implementation, step S40 includes steps C1 to C3: Step C1: When the cross-border business processing module is the deposit inquiry module, obtain the key deposit information corresponding to the cross-border business request information. The key deposit information includes the amount, currency, deposit type and account type. Step C2: Perform format verification on the key deposit information to determine whether the key deposit information meets the preset interface format requirements; Step C3: If the key deposit information meets the preset interface format requirements, the foreign currency deposit quotation interface is called based on the key deposit information to obtain deposit quotation information, and a business processing result is generated and fed back to the user based on the deposit quotation information.

[0068] It is important to understand that the aforementioned key deposit information can be core data related to foreign currency deposit quotes. Among these, the amount can refer to the deposit amount (e.g., 1 million); the currency can refer to the corresponding foreign currency (e.g., currency A, euro, pound sterling); the deposit type can refer to the deposit term (e.g., time deposit, demand deposit, notice deposit); and the account type can refer to the type of deposit account (e.g., domestic foreign currency account, overseas foreign currency account). This information is a necessary parameter for calling the foreign currency deposit quote interface.

[0069] It's easy to understand that the interactive device can use scripts (such as Python scripts or Java programs) to check the data format of the extracted key deposit information to determine whether the information meets the input requirements of the foreign currency deposit quotation interface. This includes checking if the amount is in numeric format, if the currency is a supported currency code (e.g., currency A corresponds to "USD"), and if the deposit type is within the interface's preset options (e.g., "fixed-term" or "current account"). The aforementioned foreign currency deposit quotation interface can be a specific API interface connected to a database. When key deposit information that meets the format requirements is passed in, this interface can return real-time deposit quotation information for the corresponding deposit product, including interest rate, interest calculation method, and minimum deposit amount.

[0070] For ease of understanding, please refer to Figure 6 The deposit inquiry process is illustrated with an example. Figure 6 This is a schematic diagram of the deposit inquiry processing procedure in the second embodiment of the intelligent interaction method for cross-border business of this application. Figure 6As shown, assuming a user inputs a cross-border business request as "1 million RMB in Currency A, 1-month fixed-term domestic foreign currency deposit quote", the device can call the deposit quote module. The large language model corresponding to the deposit quote module can extract key information (slot extraction) from the request information based on preset prompts (such as "accurately parse the user's input deposit quote text, extract amount, currency, deposit type, and account type"), obtaining the key deposit information: amount "1 million", currency "currency A", deposit type "fixed-term (1 month)", and account type "domestic foreign currency". The deposit quote module can call a script to perform format validation on the key deposit information, and when the validation result meets the preset interface format requirements, it uses the format-converted key deposit information as input parameters to call the foreign currency deposit quote interface of the database.

[0071] After receiving the input parameters, the foreign currency deposit quotation interface queries the latest deposit quotation stored in the database and returns the quotation information for a one-month domestic currency A time deposit (such as an annual interest rate of 2.5%, a minimum deposit of 50,000 currency A, and quarterly interest calculation). The deposit inquiry module organizes this quotation information into business processing results, feeds them back to the device, and outputs them to the user.

[0072] Therefore, existing methods for obtaining foreign currency deposit quotes require manual queries across multiple systems, and suffer from errors in key information extraction and a lack of format validation, resulting in low efficiency and inaccurate results. This embodiment addresses these issues by using a large language model to accurately extract key deposit information, reducing manual intervention and the error rate in information extraction; format validation ensures that input parameters meet interface requirements, preventing interface call failures due to incorrect parameters; and it automatically calls the foreign currency deposit quote interface and processes the results, improving deposit quote efficiency, ensuring the real-time nature and accuracy of quote information, and meeting the needs of business personnel to quickly obtain deposit quotes.

[0073] In one feasible implementation, in this embodiment, step S40 includes steps D1 to D4: Step D1: When the cross-border business processing module is the foreign exchange query module, obtain the key currency information and currency information tags corresponding to the cross-border business request information; Step D2: Based on the key currency information, call the exchange rate query interface to obtain the corresponding real-time foreign currency exchange rate information; Step D3: Call the information query interface to obtain the corresponding foreign exchange information based on the currency information tags; Step D4: Generate a business processing result that is fed back to the user based on the real-time foreign exchange rate information and the foreign exchange information.

[0074] It's easy to understand that the aforementioned key currency information refers to core data related to foreign exchange inquiries. This may include the currency pair being queried (e.g., "currency A to currency B", "currency B to currency A"), the time range for the query (e.g., "real-time", "today", "last week"), etc., and are key parameters for calling exchange rate query interfaces and information query interfaces. The aforementioned exchange rate query interface can refer to an API interface provided by a bank or third-party financial data platform for querying foreign exchange rates. It supports inputting key currency information and returns real-time exchange rates for the corresponding currency pair (e.g., spot buying rate, spot selling rate, mid-rate), exchange rate fluctuation data, etc., ensuring the timeliness of exchange rate information.

[0075] At this time, the interactive device can receive the current or most recently updated foreign exchange rate data returned by the exchange rate query interface after calling it. This is the real-time foreign exchange rate information mentioned above. It can include exchange rate values ​​for different transaction types (such as the bank's spot buying rate and spot selling rate for customers) and exchange rate update time. It can be directly used by business personnel to judge exchange rate trends and handle cross-border settlement business.

[0076] It's easy to understand that the aforementioned information query interface can be an API interface used to obtain news, policy interpretations, market analysis reports, and other content related to foreign exchange. It supports filtering relevant information based on currency consultation tags (i.e., currency information and consultation scope included in cross-border business request information), ensuring that the information content matches the user's query needs. Foreign exchange information can be information related to foreign exchange returned by the information query interface, including analysis of the impact of international economic events on exchange rates (such as the impact of the Federal Reserve's interest rate hike on the exchange rate of currency A), notifications of changes in foreign exchange policies (such as adjustments to foreign exchange controls in a certain country), market forecast reports, etc., providing business personnel with market background information behind exchange rates.

[0077] For ease of understanding, please refer to Figure 7 The following example illustrates the foreign exchange inquiry process. Figure 7 This is a schematic diagram of the foreign exchange query processing procedure in the second embodiment of the intelligent interaction method for cross-border business of this application. Figure 7 As shown, assuming a user inputs "query the real-time exchange rate of currency A against currency B and related information", the device calls the foreign exchange query module to perform content recognition.

[0078] At this point, the foreign exchange query module can analyze the request information based on the corresponding large language model and the preset prompt ("parse the foreign exchange query request and extract key currency information, including currency pairs and query time range"), and extract key currency information: currency pairs, such as "currency A to currency B"; query time range, such as "real-time".

[0079] Then, the key currency information is passed to the exchange rate query interface to query the currency exchange rate. After receiving the key currency information, the exchange rate query interface queries the real-time exchange rate database and returns the latest exchange rate information of currency A to currency B (such as spot buying rate 7.1234, spot selling rate 7.1245, mid-rate 7.1239, update time 2024-05-20 14:30:00), which is the above-mentioned real-time foreign exchange rate information.

[0080] Simultaneously, the foreign exchange query module can convert cross-border business request information into a request token (i.e., the aforementioned currency information tags, such as currency information (currency A, currency B), time information (within the last month), etc.) and pass it to the information query interface to query the consultation list. The information query interface can filter relevant foreign exchange information based on the request token and return three latest news items (such as "Country A announced its latest interest rate decision last month, and the exchange rate of currency A fluctuated in the short term," "Country B conducted foreign exchange market operations this week to stabilize the exchange rate of currency B," "Analysis report on the exchange rate trend of currency A against currency B this month," etc.). Finally, the foreign exchange query module can integrate real-time foreign exchange rate information with foreign exchange information to generate a business processing result containing exchange rate data tables, news titles, and summaries, which is transmitted to the interactive device and returned to the user by the interactive device.

[0081] This implementation addresses the problem that existing foreign exchange queries only provide exchange rate data without corresponding market information support, hindering business personnel's comprehensive understanding of the background of exchange rate fluctuations and impacting cross-border business decisions. This embodiment provides real-time and accurate foreign exchange rate information through an exchange rate query interface, meeting the basic query needs of business personnel; it also provides relevant foreign exchange information through an information query interface, offering background interpretation and market analysis of the exchange rate data to help business personnel understand the reasons for exchange rate changes and predict trends; and by integrating exchange rate and informational data, it enhances the comprehensiveness and practicality of foreign exchange query results, assisting business personnel in making more informed cross-border business decisions.

[0082] In summary, this embodiment constructs an intelligent closed-loop system for cross-border business, innovatively designing and implementing an end-to-end business integration closed loop encompassing "knowledge query - corporate customer information query - intelligent business decision-making - business tool invocation." This closed-loop system seamlessly connects all key aspects of cross-border business, fundamentally solving the inefficiency problems caused by fragmented business functions and process breakpoints in existing technologies.

[0083] Furthermore, this embodiment utilizes a RAG-enhanced cross-border knowledge question-answering engine, applying the Retrieval Enhanced Generation (RAG) technology framework and combining it with a Large Language Model (LLM) to construct a new generation of cross-border business knowledge question-answering system. This engine can dynamically retrieve the latest document knowledge base and generate accurate, comprehensive, and context-relevant professional answers, completely resolving the core pain points of existing question-answering systems: "inaccurate answers" and "incomplete answers" (insufficient information coverage).

[0084] This embodiment discloses that when the cross-border business processing module is a business question-and-answer module, the cross-border business request information is converted into a summary query vector; the summary query vector is input into a preset business knowledge base to perform parallel vector retrieval and obtain multi-path recall results; the preset business knowledge base includes a cross-border document library, a cross-border question-and-answer library, and a fallback knowledge base; the cross-border business request information and the multi-path recall results that have undergone preset validity verification are input into a large language model for relevance scoring, and a business processing result is generated and fed back to the user based on the scoring result. When the cross-border business processing module is a deposit inquiry module, the key deposit information corresponding to the cross-border business request information is obtained. The key deposit information includes amount, currency, deposit type, and account type; the key deposit information is format-validated to determine whether it conforms to the preset interface format requirements; if the key deposit information conforms to the preset interface format requirements, the foreign currency deposit quotation interface is called based on the key deposit information to obtain deposit quotation information, and a business processing result is generated and fed back to the user based on the deposit quotation information. When the cross-border business processing module is a foreign exchange query module, it obtains key currency information and currency information tags corresponding to the cross-border business request information; it calls the exchange rate query interface based on the key currency information to obtain the corresponding real-time foreign exchange rate information; it calls the information query interface to obtain the corresponding foreign exchange information based on the currency information tags; and it generates business processing results that are fed back to the user based on the real-time foreign exchange rate information and foreign exchange information. This embodiment constructs an intelligent closed-loop system for cross-border business, innovatively designing and implementing an end-to-end business integration closed loop of "knowledge query - corporate customer information query - intelligent business decision-making - business tool call". This closed-loop system seamlessly connects all key links of cross-border business, fundamentally solving the inefficiency problems caused by fragmented business functions and process breakpoints in existing technologies. In addition, this embodiment can build a new generation of cross-border business knowledge question answering system based on the RAG-enhanced cross-border knowledge question answering engine, applying the Retrieval Enhanced Generation (RAG) technology framework, combined with the Large Language Model (LLM). This engine can dynamically retrieve the latest document knowledge base and generate accurate, comprehensive, and context-relevant professional answers, completely solving the core pain points of existing question answering systems that are "inaccurate" (low accuracy) and "incomplete" (insufficient information coverage).

[0085] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the cross-border business intelligent interaction method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0086] This application also provides a smart interactive device for cross-border business; please refer to... Figure 8 , Figure 8 This is a schematic diagram of the module structure of the intelligent interaction device for cross-border business according to an embodiment of this application. In this embodiment, the intelligent interaction device for cross-border business includes: The data receiving module 801 is used to receive cross-border business request information input by the user; The intent parsing module 802 is used to parse the cross-border business request information through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; The business processing module 803 is used to determine the corresponding cross-border business processing module according to the business intent type. The cross-border business processing module includes at least one of the following: a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool call module. The interactive feedback module 804 is used to perform corresponding business processing operations on the cross-border business request information through the cross-border business processing module, and generate business processing results to be fed back to the user.

[0087] As one possible implementation, in this embodiment, the intent parsing module 802 is further configured to input the cross-border business request information into a rule engine for preliminary intent matching to obtain a preliminary matching result; and, if the preliminary matching result meets a first preset threshold, determine the business intent type corresponding to the cross-border business request information based on the preliminary matching result.

[0088] As one possible implementation, in this embodiment, the intent parsing module 802 is further configured to configure the matching rules of the rule engine, the matching rules including the regular expression, keyword set and fuzzy search keywords corresponding to the shortcut tool calling module; and to perform hot update processing on the matching rules to ensure that the matching rules are adapted to real-time cross-border business needs.

[0089] As one possible implementation, in this embodiment, the intent parsing module 802 is further configured to: input the cross-border business request information into a semantic recognition model for vector similarity calculation to obtain a semantic matching result if the preliminary matching result does not meet a first preset threshold; determine the business intent type corresponding to the cross-border business request information based on the semantic matching result if the semantic matching result meets a second preset threshold; input the cross-border business request information into a large language model for contextual semantic parsing to obtain a model parsing result if the semantic matching result does not meet the second preset threshold; and determine the business intent type corresponding to the cross-border business request information based on the model parsing result.

[0090] As one possible implementation, in this embodiment, the business processing module 803 is further configured to, when the cross-border business processing module is the business question-and-answer module, convert the cross-border business request information into a summary query vector; input the summary query vector into a preset business knowledge base to perform parallel vector retrieval and obtain multi-path recall results; the preset business knowledge base includes a cross-border document library, a cross-border question-and-answer library, and a fallback knowledge base; input the cross-border business request information and the multi-path recall results that have undergone preset validity verification into the large language model for relevance scoring, and generate business processing results that are fed back to the user based on the scoring results.

[0091] As one possible implementation, in this embodiment, the business processing module 803 is further configured to, when the cross-border business processing module is the deposit inquiry module, obtain key deposit information corresponding to the cross-border business request information, the key deposit information including amount, currency, deposit type, and account type; perform format validation on the key deposit information to determine whether the key deposit information conforms to the preset interface format requirements; if the key deposit information conforms to the preset interface format requirements, call the foreign currency deposit quotation interface based on the key deposit information to obtain deposit quotation information, and generate a business processing result fed back to the user based on the deposit quotation information.

[0092] As one possible implementation, in this embodiment, the business processing module 803 is further configured to, when the cross-border business processing module is the foreign exchange query module, obtain key currency information and currency information tags corresponding to the cross-border business request information; call the exchange rate query interface based on the key currency information to obtain the corresponding real-time foreign exchange rate information; call the information query interface to obtain the corresponding foreign exchange information information based on the currency information tags; and generate a business processing result that is fed back to the user based on the real-time foreign exchange rate information and the foreign exchange information information.

[0093] The cross-border business intelligent interaction device provided in this application, employing the cross-border business intelligent interaction method in the above embodiments, can solve the problems of low processing efficiency caused by the dispersion of cross-border business functions and inaccurate intent recognition. Compared with the prior art, the beneficial effects of the cross-border business intelligent interaction device provided in this application are the same as those of the cross-border business intelligent interaction method provided in the above embodiments, and other technical features in the device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0094] This application provides a cross-border business intelligent interaction device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the cross-border business intelligent interaction method in the first embodiment described above.

[0095] The following is for reference. Figure 9 The diagram illustrates a structural schematic suitable for implementing cross-border business intelligent interactive devices in the embodiments of this application. The cross-border business intelligent interactive devices in the embodiments of this application may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 9 The cross-border business intelligent interaction device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0096] like Figure 9 As shown, the cross-border business intelligent interactive device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the cross-border business intelligent interactive device. The processing unit 1001, the read-only memory 1002, and the RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the cross-border business intelligent interaction device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a cross-border business intelligent interaction device with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented or possessed alternatively.

[0097] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, an embodiment disclosed in this application includes a cross-border business intelligent interaction program product, which includes a cross-border business intelligent interaction program carried on a computer-readable medium, the cross-border business intelligent interaction program containing program code for performing the methods shown in the flowcharts. In such an embodiment, the cross-border business intelligent interaction program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the cross-border business intelligent interaction program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0098] The cross-border business intelligent interaction device provided in this application, employing the cross-border business intelligent interaction method in the above embodiments, can solve the technical problems of cross-border business intelligent interaction. Compared with the prior art, the beneficial effects of the cross-border business intelligent interaction device provided in this application are the same as the beneficial effects of the cross-border business intelligent interaction method provided in the above embodiments, and other technical features in the cross-border business intelligent interaction device are the same as the features disclosed in the method of the previous embodiment, and will not be repeated here.

[0099] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0100] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included 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.

[0101] This application provides a storage medium having computer-readable program instructions (i.e., a cross-border business intelligent interaction program) stored thereon, which are used to execute the cross-border business intelligent interaction method in the above embodiments.

[0102] The storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of the storage medium may include, but are not limited to: electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0103] The aforementioned storage medium may be included in the cross-border business intelligent interaction device; or it may exist independently and not be installed in the cross-border business intelligent interaction device.

[0104] The aforementioned storage medium carries one or more programs. When the aforementioned one or more programs are executed by the cross-border business intelligent interaction device, the cross-border business intelligent interaction device becomes: cross-border business intelligent interaction.

[0105] Cross-border business intelligent interactive program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—as well as conventional procedural programming languages—such as the "C" language or similar 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 a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0106] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and cross-border business intelligent interactive program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0107] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0108] The readable storage medium provided in this application is a storage medium that stores computer-readable program instructions (i.e., a cross-border business intelligent interaction program) for executing the aforementioned cross-border business intelligent interaction method, and is able to solve the technical problems of cross-border business intelligent interaction. Compared with the prior art, the beneficial effects of the storage medium provided in this application are the same as the beneficial effects of the cross-border business intelligent interaction method provided in the above embodiments, and will not be repeated here.

[0109] The above are only some embodiments of this application and do not limit the scope of the solution of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included within the protection scope of this application.

Claims

1. A cross-border business intelligent interaction method, characterized in that, The method includes: Receive cross-border business request information input by the user; The cross-border business request information is parsed through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; The corresponding cross-border business processing module is determined according to the business intent type. The cross-border business processing module includes at least one of the following: a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool call module. The cross-border business processing module performs corresponding business processing operations on the cross-border business request information and generates business processing results that are fed back to the user.

2. The intelligent interaction method for cross-border business as described in claim 1, characterized in that, The step of parsing the cross-border business request information through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information includes: The cross-border business request information is input into the rule engine for preliminary intent matching to obtain preliminary matching results; If the preliminary matching result meets the first preset threshold, the business intent type corresponding to the cross-border business request information is determined based on the preliminary matching result.

3. The intelligent interaction method for cross-border business as described in claim 2, characterized in that, Before inputting the cross-border business request information into the rule engine for preliminary intent matching and obtaining the preliminary matching result, the process further includes: Configure the matching rules of the rule engine, including the regular expression, keyword set and fuzzy search keywords corresponding to the shortcut tool calling module; The matching rules are hot-updated to ensure they are adapted to real-time cross-border business needs.

4. The intelligent interaction method for cross-border business as described in claim 2, characterized in that, After inputting the cross-border business request information into the rule engine for preliminary intent matching and obtaining the preliminary matching result, the process further includes: If the preliminary matching result does not meet the first preset threshold, the cross-border business request information is input into the semantic recognition model for vector similarity calculation to obtain the semantic matching result; If the semantic matching result meets the second preset threshold, the business intent type corresponding to the cross-border business request information is determined based on the semantic matching result; If the semantic matching result does not meet the second preset threshold, the cross-border business request information is input into a large language model for contextual semantic parsing, and the business intent type corresponding to the cross-border business request information is determined based on the model parsing result.

5. The intelligent interaction method for cross-border business as described in claim 4, characterized in that, The step of performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module and generating business processing results fed back to the user includes: When the cross-border business processing module is the business question and answer module, the cross-border business request information is converted into a summary query vector; The summary query vector is input into a preset business knowledge base to perform parallel vector retrieval and obtain multi-path recall results; the preset business knowledge base includes a cross-border document library, a cross-border question and answer library, and a fallback knowledge base. The cross-border business request information and the multi-path recall results that have undergone preset validity verification are input into the large language model for relevance scoring, and a business processing result is generated and fed back to the user based on the scoring result.

6. The intelligent interaction method for cross-border business as described in claim 4, characterized in that, The step of performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module and generating business processing results fed back to the user includes: When the cross-border business processing module is the deposit inquiry module, the key deposit information corresponding to the cross-border business request information is obtained. The key deposit information includes the amount, currency, deposit type and account type. The key deposit information is format-validated to determine whether it conforms to the preset interface format requirements. If the key deposit information meets the preset interface format requirements, the foreign currency deposit quotation interface is called based on the key deposit information to obtain deposit quotation information, and a business processing result is generated and fed back to the user based on the deposit quotation information.

7. The intelligent interaction method for cross-border business as described in claim 4, characterized in that, The step of performing corresponding business processing operations on the cross-border business request information through the cross-border business processing module and generating business processing results fed back to the user includes: When the cross-border business processing module is the foreign exchange query module, the key currency information and currency information tags corresponding to the cross-border business request information are obtained; The exchange rate query interface is invoked based on the key currency information to obtain the corresponding real-time foreign currency exchange rate information; Call the information query interface to obtain the corresponding foreign exchange information based on the currency information tags; Based on the real-time foreign exchange rate information and the foreign exchange information, a business processing result is generated and fed back to the user.

8. A cross-border business intelligent interaction device, characterized in that, The cross-border business intelligent interaction device includes: The data receiving module is used to receive cross-border business request information input by the user; The intent parsing module is used to parse the cross-border business request information through a multi-level intent recognition mechanism to obtain the business intent type corresponding to the cross-border business request information; A business processing module is used to determine the corresponding cross-border business processing module according to the business intent type. The cross-border business processing module includes at least one of a business Q&A module, a deposit inquiry module, a foreign exchange inquiry module, and a quick tool call module. The interactive feedback module is used to perform corresponding business processing operations on the cross-border business request information through the cross-border business processing module, and generate business processing results to be fed back to the user.

9. A cross-border business intelligent interactive device, characterized in that, The device includes: a memory, a processor, and a cross-border business intelligent interaction program stored in the memory and executable on the processor, the cross-border business intelligent interaction program being configured to implement the steps of the cross-border business intelligent interaction method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a cross-border business intelligent interaction program, which, when executed by a processor, implements the steps of the cross-border business intelligent interaction method as described in any one of claims 1 to 7.

Images