Information prediction method, information query method, equipment, storage medium and program product

Abstract

The embodiment of the invention provides an information prediction method, an information query method, a time prediction method, computing equipment, a computer readable storage medium and a computer program product. The information prediction method comprises the following steps: in response to a query request, determining a query parameter and a target data flow record hit by the query parameter; querying at least one first sample data flow record matched with the query parameter from a flow knowledge base; under the condition that the query request meets the calculation requirement, according to the circulation information corresponding to the at least one first sample data circulation record, determining predicted circulation information corresponding to the target data circulation record according to a prediction rule; and otherwise, predicting the predicted flow information corresponding to the target data flow record by using the first prediction model according to the historical behavior data of the data transfer-out party hit by the query parameter. According to the technical scheme provided by the embodiment of the invention, the validity of data flow query is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an information prediction method, an information query method, a computing device, a computer-readable storage medium, and a computer program product. Background Technology

[0002] With the development of computer and network technologies, information, bandwidth, storage space, and funds can all be considered as resource data. In practical applications, there are data flow scenarios where resource data is transferred from one data system to another via the network. For example, when the resource data is funds, it involves remittance scenarios between different banking systems. During the data flow process, between the data forwarding system and the data receiving system, one or more intermediary data systems (also known as relay systems) may be used for relaying.

[0003] To ensure secure data transfer, different data systems can typically join the same security organization and transmit data according to that organization's communication protocol. The data sender and receiver can choose their respective data systems for data transfer. When the data sender initiates a data transfer operation using a particular data system, a corresponding data transfer record is generated. To facilitate data tracking, different data systems can use enhanced services provided by the security organization to report transfer information, making it easier for the data sender or receiver to query the data transfer record.

[0004] In practical applications, in order to facilitate data querying, query services are usually provided by query systems associated with data systems or security agencies. These systems can match data flow records based on user-provided data flow credentials, unique identifiers of data flow records generated by security agencies, or several query fields, and obtain the corresponding flow information from the data system or security agency.

[0005] However, because the transfer information may not be updated in a timely manner or may not be searchable, effective results may not be obtained, affecting the search results. Summary of the Invention

[0006] This application provides an information prediction method, an information query method, a computing device, a computer-readable storage medium, and a computer program product.

[0007] In a first aspect, embodiments of this application provide an information prediction method, including: In response to a query request, determine the query parameters and the target data flow record to which the query parameters are matched; The system queries at least one first sample data flow record that matches the query parameters from the flow knowledge base; wherein the flow knowledge base is constructed based on multiple historical data flow records. If the query request meets the calculation requirements, the predicted flow information corresponding to the target data flow record is determined according to the flow information corresponding to the at least one first sample data flow record, in accordance with the prediction rules; otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data transferor hit by the query parameters; the predicted flow information is used as the query result of the query request.

[0008] Secondly, this application provides an information query method, including: Obtain a query request; the query request includes query parameters; Query whether there is target flow information corresponding to the target data flow record that matches the query parameters; If so, output the target flow information; If not, query the knowledge base for multiple first sample data flow records that match the query parameters; the knowledge base is constructed based on multiple historical data flow records. If the query request meets the calculation requirements, the predicted flow information corresponding to the target data flow record is calculated according to the prediction rules based on the flow information corresponding to the multiple first sample data flow records respectively; otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data transferor hit by the query parameters. Output the predicted flow information.

[0009] Thirdly, this application provides an information prediction method, including: In response to the query request, determine the query parameters and the target remittance record matched by the query parameters; Search the remittance knowledge base for at least one sample remittance record that matches the query parameters; the remittance knowledge base is built and generated based on multiple historical remittance records; If the query request meets the calculation requirements, based on the remittance time data corresponding to the at least one sample remittance record, the predicted arrival time corresponding to the target remittance record is determined according to the prediction rules; otherwise, the first prediction model is used to predict the predicted arrival time corresponding to the target remittance record based on the historical behavior data of the remitter matched by the query parameters; the predicted arrival time is used as the query result of the query request.

[0010] Fourthly, this application provides an information query method, including: Obtain a query request; the query request includes query parameters; Check if there is a corresponding arrival time for a target remittance record that matches the query parameters; If so, output the arrival time. If not, search the remittance knowledge base for at least one first sample remittance record that matches the query parameters; the remittance knowledge base is built and generated based on multiple historical remittance records. If the query request meets the calculation requirements, based on the remittance time data corresponding to the at least one first sample remittance record, the predicted arrival time corresponding to the target remittance record is determined according to the prediction rules; otherwise, the first prediction model is used to predict the predicted arrival time corresponding to the target remittance record based on the historical behavior data of the remitter matched by the query parameters. Output the predicted arrival time.

[0011] Fifthly, this application provides a computing device, including a processing component and a storage component; The storage component stores a computer program; the computer program is invoked and executed by the processing component to implement the information prediction method as described in the first aspect above, or the information query method as described in the second aspect above, or the information prediction method as described in the third aspect above, or the information query method as described in the fourth aspect above.

[0012] Sixthly, this application provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processing component, it implements the information prediction method as described in the first aspect above, or the information query method as described in the second aspect above, or the information prediction method as described in the third aspect above, or the information query method as described in the fourth aspect above.

[0013] In a seventh aspect, this application provides a computer program product, including a computer program or instructions, which, when executed by a processing component, implement the information prediction method as described in the first aspect above, or the information query method as described in the second aspect above, or the information prediction method as described in the third aspect above, or the information query method as described in the fourth aspect above.

[0014] This embodiment of the application, after determining the query parameters and target data flow record in response to a query request, first queries at least one first sample data flow record matching the query parameters from the flow knowledge base. If the query request meets the calculation requirements, based on the flow information corresponding to each of the at least one first sample data flow record, predictive flow information is determined according to prediction rules. Otherwise, a first prediction model is used to predict based on the historical behavior data of the data transferor matched by the query parameters. Finally, the generated predicted flow information is output as the query result of the query request. This achieves the prediction of flow information. When the query information cannot be obtained, the predicted flow information can be output as the query result, thereby achieving effective querying, ensuring query results, and improving the user query experience. Furthermore, this application's embodiments employ a hybrid prediction mechanism combining rules and artificial intelligence models. In scenarios where at least one first sample data flow record can be matched from the flow knowledge base and the query request meets the computational requirements, rules are used to quickly provide statistically based flow information predictions. In scenarios where at least one first sample data flow record cannot be matched from the flow knowledge base and the query request does not meet the computational requirements, the first prediction model can be used to deeply mine the historical behavioral data of the data transferor for intelligent prediction. This effectively solves the technical problem of data flow records being unable to obtain effective flow information due to outdated flow information, significantly improving query coverage and query results. It enables scenarios that previously could not provide results to obtain accurate predicted flow information, reduces users' reliance on manual verification, and ensures the effectiveness and practicality of data flow queries.

[0015] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A system architecture diagram illustrating a technical solution of an embodiment of this application that can be applied thereto is shown; Figure 2 A flowchart of an embodiment of the information prediction method provided in this application; Figure 3 A schematic diagram of the structure of one embodiment of the first prediction model provided in this application; Figure 4 A flowchart of an embodiment of an information retrieval method provided in this application; Figure 5 A flowchart of yet another embodiment of the information prediction method provided in this application; Figure 6A flowchart illustrating yet another embodiment of the information retrieval method provided in this application; Figure 7 This is a schematic diagram of the time prediction process in a practical application according to an embodiment of this application; Figure 8 A schematic diagram of the structure of an embodiment of an information prediction device provided in this application; Figure 9 A schematic diagram of the structure of an embodiment of an information query device provided in this application; Figure 10 A schematic diagram of another embodiment of the information prediction device provided in this application; Figure 11 A schematic diagram of another embodiment of the information query device provided in this application; Figure 12 This is a schematic diagram of the structure of one embodiment of a computing device provided in this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] It should be noted that, in the cases involving user information in the embodiments of this application, the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the embodiments of this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use, and processing of related data must comply with the relevant laws, regulations, and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse. In addition, the various models involved in this application (including but not limited to language models or multimodal models) comply with relevant laws and standards.

[0019] To address the technical problem of untimely or unqueried data transfer information leading to ineffective results and impacting query performance, this application provides a solution. The basic idea is as follows: After determining the query parameters and target data transfer record in response to a query request, firstly, at least one first sample data transfer record matching the query parameters is retrieved from the data transfer knowledge base. If the query request meets the computational requirements, predicted data transfer information is determined according to prediction rules based on the transfer information corresponding to each of the at least one first sample data transfer record. Otherwise, a first prediction model is used to predict based on the historical behavior data of the data transferor matched by the query parameters. Finally, the generated predicted data transfer information is output as the query result of the query request. This achieves the prediction of data transfer information. When query information is unavailable, the predicted data transfer information can be output as the query result, thus achieving effective querying, ensuring query performance, and improving the user query experience. Furthermore, this application's embodiments employ a hybrid prediction mechanism combining rules and AI (Artificial Intelligence) models. In scenarios where at least one first sample data flow record can be matched from the flow knowledge base and the query request meets the computational requirements, rules are used to quickly provide statistically based flow information predictions. In scenarios where at least one first sample data flow record cannot be matched from the flow knowledge base and the query request does not meet the computational requirements, the first prediction model can be used to deeply mine the historical behavioral data of the data transferor for intelligent prediction. This effectively solves the technical problem of data flow records being unable to obtain effective flow information due to outdated flow information, significantly improving query coverage and query results. It enables scenarios that previously could not provide results to obtain accurate predicted flow information, reduces users' reliance on manual verification, and ensures the effectiveness and practicality of data flow queries.

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] Figure 1 The diagram illustrates a system architecture diagram in which the technical solution of an embodiment of this application can be applied. In practical applications, this system architecture can be a query system, which can be connected to different data systems or security agencies.

[0022] The system may include a client 101 and a server 102.

[0023] In this system, the user terminal 101 and the server terminal 102 can establish a connection via a network. The network provides a communication link between the user terminal 101 and the server terminal 102. The network can include various connection types, such as wired, wireless, or fiber optic cables. The user terminal 101 can interact with the server terminal 102 through the network to receive or send messages, etc.

[0024] The user terminal 101 can be a browser, an app (application), a web application such as an H5 (HyperText Markup Language 5) application, a lightweight application (also known as a mini-program), or a cloud application. The user terminal 101 can be deployed on electronic devices and depends on the device or certain apps on the device to run. Electronic devices can have displays and support information browsing, such as personal mobile terminals like mobile phones, tablets, personal computers, desktop computers, smart speakers, smartwatches, etc. For ease of understanding... Figure 1 The user end is primarily represented by the image of a device. Various other types of applications can also be configured in electronic devices, such as human-computer interaction applications, model training applications, text processing applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, and social media platform software. Electronic devices can refer to devices used by users that have the computing, internet access, and communication functions required by the user, such as mobile phones, tablets, personal computers, and wearable devices. Electronic devices typically include at least one processing component and at least one storage component. Electronic devices may also include basic configurations such as network interface cards (NICs), I / O (input / output) buses, and audio / video components; this application does not limit their inclusion of these components. Optionally, depending on the implementation of the electronic device, it may also include some peripheral devices, such as keyboards, mice, input pens, and printers; this application does not limit their inclusion of these components.

[0025] Server 102 may include servers that provide various services, such as servers that support model training, or servers that process interactive information sent by users.

[0026] It should be noted that server 102 can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. The server can also be a server in a distributed system, or a server integrated with blockchain. The server can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.

[0027] It should be noted that the information prediction method and information query method provided in this application embodiment are generally executed by the server 102, and the corresponding information prediction device and information query device are generally set in the server 102. However, it is understood that the user terminal 101 may also have similar functions to the server 102, thereby executing the information prediction method and information query method provided in this application embodiment. In addition, the information prediction method and information query method provided in this application embodiment may also be jointly executed by the user terminal 101 and the server 102.

[0028] It should be understood that Figure 1 The number of client and server instances shown is merely illustrative. Depending on implementation needs, there can be any number of client and server instances.

[0029] The implementation details of the technical solutions in the embodiments of this application are described in detail below.

[0030] Figure 2 A flowchart illustrating an embodiment of an information prediction method provided in this application, which can be executed by a server. Figure 2 The information prediction method shown may include the following steps: 201: In response to a query request, determine the query parameters and the target data flow record to which the query parameters are matched.

[0031] In the embodiments of this application, the query request can be initiated by either the data transferee or the data transferor.

[0032] By parsing the query request, query parameters can be obtained. These query parameters may include at least one query factor that can uniquely identify or help locate data flow records, such as query factors representing the data flow subject, flow object, flow characteristics, etc.

[0033] After determining the query parameters, a target data flow record matching the query parameters can be queried. In one embodiment of this application, the query parameters can be compared with the record fields of multiple currently generated data flow records. When a record field of a certain data flow record matches the query parameters, that data flow record can be identified as the target data flow record matched by the query parameters. This target data flow record may contain missing information; therefore, the technical solution of this embodiment can be used for information prediction.

[0034] In a remittance scenario, data flow can be understood as the transfer of funds between different processing entities, such as financial institutions. In this case, the data flow record can be a remittance record, used to document a remittance transaction. In this remittance scenario, the data sender can correspond to the remittance initiator or the financial institution corresponding to the initiator, and the data receiver can correspond to the remittance recipient or the financial institution corresponding to the recipient. Corresponding query parameters can include, for example, the payer's financial institution identifier, the receiver's financial institution identifier, the payer's account information, and the receiver's account information, used to represent the flow relationship of funds between different entities. The aforementioned flow objects in a remittance scenario can correspond to remittance funds or objects related to remittance funds, such as remittance amount, currency, transaction number, and order identifier. Flow characteristics in a remittance scenario can include, for example, the remittance initiation time.

[0035] After determining the query parameters and retrieving the target data flow record, at least one first sample data flow record matching the query parameters can be retrieved from the pre-built flow knowledge base.

[0036] In another embodiment of this application, after determining the query parameters and retrieving the target data flow record, it is possible to further retrieve at least one first sample data flow record that matches the query parameters from a pre-built flow knowledge base if it is determined that the target data flow record has not yet generated a flow result.

[0037] 202: Query at least one first sample data flow record that matches the query parameters from the flow knowledge base.

[0038] The knowledge base for data flow can be constructed based on multiple historical data flow records. Based on these historical data flow records, corresponding sample data flow records can be generated and saved in the knowledge base for data flow.

[0039] The sample data flow record may include information such as the flow path, data recipient information, data sender information, data sender time, and data recipient time.

[0040] Historical data transfer records can be obtained from multiple data systems or from security agencies, and this application does not limit this. In remittance scenarios, such as cross-border remittance scenarios, the security agency may be, for example, GPI (Global Payments Innovation) provided by SWIFT (Society for Worldwide Interbank Financial Telecommunication), and the data transfer records may refer to GPI data.

[0041] In the embodiments of this application, the query parameters can be compared with multiple sample data flow records in the flow knowledge base. If one or more historical data flow records match the query parameters, then the one or more historical data flow records can be identified as the first sample data flow record that matches the query parameters.

[0042] 203: Determine whether the query request meets the calculation requirements. If yes, proceed to step 204; otherwise, proceed to step 205.

[0043] 204: Based on the flow information corresponding to at least one first sample data flow record, determine the predicted flow information corresponding to the target data flow record according to the prediction rules.

[0044] 205: Using the first prediction model, based on the historical behavior data of the data transfer party matched by the query parameters, predict the predicted flow information corresponding to the target data flow record; the predicted flow information is used as the query result of the query request.

[0045] Among them, whether the calculation requirements are met can be used to characterize whether the current query request has the conditions for regular calculation based on at least one first sample data flow record, so that the prediction method of the flow information can be determined according to the judgment result.

[0046] In one embodiment of this application, when the query request meets the calculation requirements, determining the predicted flow information corresponding to the target data flow record according to the prediction rules based on the flow information corresponding to at least one first sample data flow record can be specifically implemented as follows: If a predetermined number of first sample data flow records matching the query parameters exist in the flow knowledge base, the predicted flow information corresponding to the target data flow record is determined according to the flow information corresponding to each of the at least one first sample data flow record and the prediction rule.

[0047] In this case, the calculation requirement ensures that the flow record of the first sample data involved in the calculation has sufficient statistical significance, thereby supporting the stability and reliability of the calculation results based on the prediction rules.

[0048] In another embodiment of this application, the query request is deemed to meet the computational requirements when the first sample data flow record obtained from the query shows a high degree of consistency with the target data flow record in terms of key query parameter dimensions. Key query parameter dimensions may include parameter information used to characterize the data flow subject, data flow object, or data flow characteristics. When the sample data maintains consistency with the target data flow record in the aforementioned dimensions or meets preset similarity conditions, it indicates that the sample data is highly representative and suitable for computational processing based on prediction rules.

[0049] If the query request meets the calculation requirements, the flow information corresponding to each first sample data flow record can be extracted. This flow information can include information related to the prediction target during the historical data flow process. After obtaining the flow information, it can be processed according to preset rules to determine the prediction flow information corresponding to the target data flow record.

[0050] In the embodiments of this application, when there is sufficient historical sample flow data and computational conditions, existing flow information and prediction rules can be used to predict the flow information of the target data flow record.

[0051] If the query request does not meet the computational requirements, a first prediction model can be invoked to obtain the predicted flow information using a model-based prediction method. In the embodiments of this application, historical behavior data corresponding to the data transferor matched by the query parameters can be obtained and provided as input to the first prediction model. The first prediction model can then analyze the historical behavior data of the data transferor to predict the predicted flow information corresponding to the target data flow record.

[0052] In remittance scenarios, historical behavioral data can include long-tail features such as the remittance financial institution's clearing habits, commonly used transit routes, and processing preferences in past remittances.

[0053] By introducing the first prediction model, corresponding prediction results can still be generated even when the query request does not meet the calculation requirements, thereby improving the coverage of information flow prediction.

[0054] The first prediction model can be an artificial intelligence-based language model (LM) or multimodal model (MM), etc. This application does not limit the number of model parameters supported by the model, aiming to meet actual needs. In an optional implementation, such as... Figure 3 As shown, the first prediction model may include, for example, an input layer 301, an encoder 302, a decoder 303, and an output layer 304. It may also include a self-attention layer and a feed-forward neural network, etc., and this application does not impose any limitations on this. The input layer 301 is used to receive data input, including text, images, or multimodal data. The encoder 302 is mainly used to convert the input data (usually in sequence form) into a vector representation. This process can capture the semantic features of the input data. The decoder 303 is responsible for converting the intermediate representation generated by the encoder into output data (usually in sequence form). The output layer 304 is used to output data, such as prediction flow information in JSO format. A self-attention layer is a mechanism that allows a model to focus on other positions in a sequence to better encode information about the current position. Feedforward neural networks can perform nonlinear transformations on the output of the self-attention layer to enhance the model's expressive power. The various parts work together, enabling models built on them to perform well in a variety of complex processing tasks, such as natural language processing, computer vision, speech recognition, machine translation, text summarization, and intelligent question answering.

[0055] This embodiment of the application, after determining the query parameters and target data flow record in response to a query request, first queries at least one first sample data flow record matching the query parameters from the flow knowledge base. If the query request meets the calculation requirements, based on the flow information corresponding to each of the at least one first sample data flow record, predictive flow information is determined according to prediction rules. Otherwise, a first prediction model is used to predict based on the historical behavior data of the data transferor matched by the query parameters. Finally, the generated predicted flow information is output as the query result of the query request. This achieves the prediction of flow information. When the query information cannot be obtained, the predicted flow information can be output as the query result, thereby achieving effective querying, ensuring query results, and improving the user query experience. Furthermore, this application's embodiments employ a hybrid prediction mechanism combining rules and AI models. In scenarios where at least one first sample data flow record can be matched from the flow knowledge base and the query request meets the computational requirements, rules are used to quickly provide statistically based flow information predictions. In scenarios where at least one first sample data flow record cannot be matched from the flow knowledge base and the query request does not meet the computational requirements, the first prediction model can be used to deeply mine the historical behavioral data of the data transferor for intelligent prediction. This effectively solves the technical problem of data flow records being unable to obtain effective flow information due to outdated flow information, significantly improving query coverage and query results. It enables scenarios that previously could not provide results to obtain accurate predicted flow information, reduces users' reliance on manual verification, and ensures the effectiveness and practicality of data flow queries.

[0056] The prediction rule may include a calculation method for determining the predicted flow information based on at least one first sample data flow information. For example, the prediction rule may be: the average, median or flow information whose proportion exceeds a predetermined ratio of at least one flow information corresponding to the at least one first sample data flow record shall be used as the predicted flow information.

[0057] In a practical application, the flow information may include the data transfer time; The calculation method could be as follows: based on the data transfer-in time and data transfer-out time in each first sample data transfer record, calculate the data transfer time corresponding to each first sample data transfer record; based on the data transfer time corresponding to multiple first sample data transfer records, determine the first data transfer time that belongs to the median or the second data transfer time that exceeds a predetermined proportion; and based on the data transfer-out time in the target data transfer record, as well as the first data transfer time and / or the second data transfer time, calculate the predicted transfer-in time.

[0058] Therefore, in some embodiments, at least one first sample data flow record may include multiple first sample data flow records.

[0059] The above-mentioned determination of the predicted flow information corresponding to the target data flow record based on the flow information corresponding to at least one first sample data flow record, according to the prediction rule, can be specifically implemented as follows: Determine the data flow time corresponding to each of the multiple first sample data flow records to obtain multiple data flow times; Statistical analysis of multiple data processing times is conducted to determine the first data processing time that falls within the median or the second data processing time that exceeds a predetermined proportion. The predicted transfer-in time is calculated based on the data transfer-out time in the target data transfer record, as well as the first data transfer time and / or the second data transfer time.

[0060] In the embodiments of this application, firstly, for multiple first sample data flow records, the data flow time corresponding to each first sample data flow record in the historical flow process can be determined, thereby obtaining multiple data flow times. Data flow time can be used to reflect the time interval between data flow from the data sender to the data receiver. For example, in a remittance scenario, data flow time can be the actual number of processing days or hours required from the remittance financial institution sending funds to the receiving financial institution receiving the funds. By determining the data flow time corresponding to each first sample data flow record in the historical flow process, multiple data flow times reflecting the historical performance of similar flow scenarios can be obtained.

[0061] After obtaining the time taken for multiple data streams, statistical processing can be performed on these multiple data stream times to extract representative time-consuming features.

[0062] For example, statistical analysis can be used to determine the median of multiple data flow times, and this median can be used as the first data flow time. Furthermore, a second flow time exceeding a predetermined proportion can be determined from multiple data flow times, such as the time that accounts for 90% of the statistical time distribution (i.e., the RT90 index). By using the median or the time exceeding a predetermined proportion as the statistical result, the impact of extreme values ​​on the prediction results can be effectively reduced, and the stability of the predicted flow information can be improved.

[0063] The predicted transition time can be a specific moment or a time interval.

[0064] The predicted transition time can be calculated according to the time calculation method in the prediction rules.

[0065] The calculation of the predicted transfer-in time based on the data transfer-out time in the target data transfer record, and the first data transfer time and / or the second data transfer time may include: adding the first data transfer time to the data transfer-out time in the target data transfer record to obtain the first transfer-in time; adding the second data transfer time to the data transfer-out time in the target data transfer record to obtain the second transfer-in time; and using the first transfer-in time, or the second transfer-in time, or the time interval formed by the first transfer-in time and the second transfer-in time, or the time interval obtained by subtracting and superimposing a predetermined error time from the first transfer-in time, or the time interval obtained by subtracting and superimposing a predetermined error time from the second transfer-in time, as the predicted transfer-in time.

[0066] Once the predicted transfer time is determined, it can be used as the predicted transfer information. In cross-border remittance scenarios, this predicted transfer information can be reflected as the predicted arrival time, allowing users to obtain a reliable estimate of the arrival time based on statistics of similar historical remittance records, even when real-time remittance tracking is unavailable.

[0067] The embodiments of this application improve the robustness and accuracy of prediction by performing median and / or percentile statistics on the transfer time of multiple first sample data transfer records.

[0068] In embodiments of this application, after initially determining the predicted flow information of the target data flow record according to the prediction rules based on the flow information of the first sample data flow record, it is further possible to determine at least one influencing factor hit by the target data flow record, and update the initially obtained predicted flow information according to the change conditions corresponding to each of the at least one influencing factor, so that the final output prediction result is closer to the actual flow situation, thereby improving the accuracy and reliability of the prediction. In some embodiments, the method may further include: Identify at least one influencing factor that is present in the target data flow record, and update the predicted flow information according to the change conditions corresponding to each of the at least one influencing factor.

[0069] Among them, the influencing factors may include factors that may affect the time taken for data transfer. In some embodiments, at least one influencing factor may include one or more of the following: the data transfer-out time is a holiday, the data transfer-in time is a holiday, the transferred data meets predetermined conditions, the data transfer-out region is a target type region, and the data transfer-in region is a target type region.

[0070] Here, "holiday period" can refer to statutory holidays or non-working days in the regions where the data sender and receiver are located; "pre-determined conditions" can refer to data flow characteristics that may trigger additional processing or review, such as large-amount flows exceeding a certain threshold; "target type region" can refer to specific regions that may lead to additional verification or delays based on historical experience or regulatory requirements, such as countries or regions with strict foreign exchange controls, or countries or regions listed as high-risk. "Change conditions" can refer to specific rules pre-set for each influencing factor to adjust the predicted flow information, such as extending / shortening the duration or correction factor of the initially calculated data flow time, for example, adding or shortening several working days or hours.

[0071] After determining the predicted data flow information, it can be determined whether the target data flow record matches any one or more influencing factors. If the target data flow record matches any one or more influencing factors, the predicted data flow information can be updated according to the change conditions corresponding to the influencing factors. In a specific example, for instance, in a cross-border remittance scenario, suppose the remitting financial institution of the target data flow record is located in region A, the receiving financial institution is located in region B, the remittance date is December 22, 2025, the currency is USD (United States Dollar), and the amount is $500,000. Initially, based on the first matched sample data flow record, the expected remittance date can be preliminarily calculated to be December 25, 2025. At this point, the following influencing factors can be further identified: First, the expected remittance date of December 25th falls on Christmas Day, which is a holiday period in region B where the receiving bank is located. Therefore, the remittance can be extended by one business day based on the corresponding change conditions. Second, the remittance amount exceeds the preset large-amount threshold of US$300,000, meeting the predetermined conditions. Historical data shows that large remittances often trigger additional compliance reviews by financial institutions. Therefore, the remittance can be extended by another 0.5 to 1 business day based on the change conditions. In addition, if the transit system involved in the remittance route is located in a target type of region with strict foreign exchange controls (such as countries with a high probability of delays according to historical statistics), further adjustments can be made accordingly. After identifying these influencing factors one by one and updating them based on the change conditions, the initial expected remittance time can be adjusted to the range of December 27th to 30th, 2025. This makes the final output forecast flow information more accurately reflect the actual possible arrival of funds and avoids forecast deviations caused by ignoring dynamic factors such as holidays, large-amount reviews, or regional risks.

[0072] In some embodiments, determining at least one influencing factor that the target data flow record hits, and updating the predicted flow information according to the change conditions corresponding to each of the at least one influencing factor, can be specifically implemented as follows: Using the third prediction model, at least one influencing factor that is hit in the target data flow record is identified, as well as the change conditions corresponding to each of the at least one influencing factor, and the predicted flow information is updated according to the change conditions corresponding to each of the at least one influencing factor.

[0073] In the embodiments of this application, a third prediction model can be invoked to determine at least one influencing factor that the target data flow record hits, and the predicted flow information can be updated according to the change conditions corresponding to each of the at least one influencing factor.

[0074] Similar to the first prediction model, the third prediction model can also be an artificial intelligence-based language model (LM) or multimodal model (MM), etc. The embodiments of this application do not limit the number of model parameters supported by the model, aiming to meet actual needs. The first and third prediction models can be implemented as the same model, but are not limited to this, and can also be different models.

[0075] In some embodiments, the query parameters include at least one query factor.

[0076] Among them, query factors can refer to the elements that constitute query parameters, such as data transferor identifier, data transferee identifier, currency, amount range, data transfer time period, etc.

[0077] In some embodiments, querying at least one first sample data flow record that matches the query parameters from the flow knowledge base can be specifically implemented as follows: with a predetermined number as a constraint, according to the combination requirements, one or more query factors are used as query conditions in sequence to search for at least one first sample data flow record that matches the query conditions in the flow knowledge base until a predetermined number of first sample data flow records are obtained or the query conditions are empty.

[0078] In the embodiments of this application, when querying at least one first sample data flow record matching the query parameters from the flow knowledge base, a progressive retrieval strategy can be adopted. Using a predetermined number as a constraint, one or more query factors are sequentially used as query conditions according to combination requirements. Matching first sample data flow records are searched layer by layer in the flow knowledge base until the predetermined number of first sample data flow records is accumulated or all reasonable combinations of query factors are exhausted. The predetermined number can be a pre-set threshold used to determine whether there are enough samples to support subsequent rule prediction paths, and can be determined based on historical data distribution and prediction accuracy requirements. Combination requirements can be the priority order, relaxation rules, etc., that need to be followed when combining at least one query factor. For example, it can include a predefined sequence that gradually relaxes from the most stringent factor combinations to more lenient factor combinations. This predefined sequence can be sorted according to the importance of each query factor. For example, the query factors that have the greatest impact on finding the first sample data flow record (such as the data sender and data receiver) are retained first, while query factors with relatively minor impact are placed in a later, more relaxed position. Combination requirements may include, for example, using the full combination of all query factors as the initial condition, then removing one or more query factors with the lowest importance to form a second strict combination, then removing the next lowest importance query factors, until only the most important single or a few core query factors are retained, or further introducing fuzzy matching of factor ranges (such as replacing the exact amount with an amount range) as a supplementary combination.

[0079] In one implementation, the core query factor combination that best identifies the first sample data flow record can be used as the initial query condition. For example, four query factors—data sender, data receiver, data type, and data volume—can be used simultaneously for exact matching to attempt to retrieve the first sample data flow record that meets all the conditions from the flow knowledge base. If the matching results under these strict conditions reach or exceed a predetermined number, these first sample data flow records are used directly. If the number of matches is less than the predetermined number, the query conditions can be relaxed, gradually reducing the number of query factors or converting some factors to range matching. For example, it can first degenerate to searching only using the data sender and data receiver as query conditions; if it is still insufficient, it can further degenerate to searching only using the data sender and data type as conditions, and continuously accumulate the matched first sample data flow records until the predetermined number is reached or all reasonable query factor combinations are exhausted.

[0080] A progressive retrieval strategy can effectively balance the relationship between sample similarity and sample quantity, avoiding frequent switching to the model prediction path due to insufficient samples caused by overly strict query conditions. At the same time, it can also ensure that the first sample data flow record retrieved has a high correlation with the target data flow record.

[0081] In embodiments of this application, if the number of first sample data flow records matched from the flow knowledge base is still lower than a preset number after adopting a progressive retrieval strategy, at least one first sample data flow record corresponding to the data transferor matched by the query parameters can be further determined from the flow knowledge base to provide necessary supplementary data for subsequent prediction paths based on the first prediction model. In some embodiments, when the number of first sample data flow records matched from the flow knowledge base is lower than a preset number, the method may further include: determining at least one first sample data flow record corresponding to the data transferor matched by the query parameters from the flow knowledge base.

[0082] In some embodiments, using the first prediction model to predict the predicted flow information corresponding to the target data flow record based on the historical behavior data of the data transferor matched by the query parameters can be specifically implemented as follows: Determine at least one second sample data flow record corresponding to the data transferor matched by the query parameters from the flow knowledge base; determine at least one influencing factor matched by the target data flow record; and use a first prediction model to predict the predicted flow information corresponding to the target data flow record based on the query parameters, at least one second sample data flow record, and the change conditions corresponding to different influencing factors.

[0083] At least one second sample data flow record may include data flow records initiated by the data transferor in the past, from the data flow knowledge base. This at least one second sample data flow record can characterize the behavioral patterns and historical habits of the data transferor.

[0084] After determining at least one second sample data flow record corresponding to the data transferor whose query parameters match from the flow knowledge base, the query parameters, at least one second sample data flow record, and the change conditions corresponding to different influencing factors can be input into the first prediction model. The first prediction model can then directly generate the predicted flow information corresponding to the target data flow record based on the query parameters, combined with the long-term behavior patterns reflected in at least one second sample data flow record of the data transferor, as well as the predefined different influencing factors and their change conditions.

[0085] In some embodiments, the predicted flow information corresponding to the target data flow record can be specifically implemented using a first prediction model based on query parameters, at least one second sample data flow record, and change conditions corresponding to different influencing factors: Based on a preset prompt word template, query parameters, and at least one second sample data flow record, the first prompt word is constructed according to the change conditions corresponding to different influencing factors. The first prompt word is then input into the first prediction model so that the first prediction model can determine the prediction flow information according to the instructions of the first prompt word.

[0086] The preset prompt word template defines the task flow, core focus, calculation logic, and output format requirements that the first prediction model must follow during prediction. In the actual prediction process, based on the current query parameters, at least one second sample data flow record related to the data sender obtained from the flow knowledge base, and the change conditions corresponding to different influencing factors, this specific information can be filled into or integrated into the corresponding positions in the prompt word template, thereby dynamically constructing a complete first prompt word. This first prompt word may include, for example, model role positioning, detailed task decomposition (such as intermediate flow paths and calculation of processing time for each transit system, calculation of data transfer time windows, and calculation method identification), core focus points (such as processing principles for holidays and weekends, and priority of referencing historical samples), and JSON output format requirements to ensure that the model output is standardized and easy to parse subsequently.

[0087] Then, the constructed first prompt word can be input into the first prediction model. The first prediction model can strictly follow the instructions defined in the first prompt word, combine the provided query parameters, historical sample data flow records, and change conditions, perform comprehensive reasoning and calculation, and finally output the predicted flow information.

[0088] In some embodiments, the method may further include: Identify at least one path node in the target data flow record; based on the at least one path node, predict the target flow path corresponding to the target data flow record; The above-mentioned determination of the predicted flow information corresponding to the target data flow record based on the flow information corresponding to at least one first sample data flow record, according to the prediction rule, may include: From at least one first sample data flow record, determine at least one target sample data flow record that the target flow path hits; based on at least one target sample data flow record, determine the predicted transfer time of the target flow path corresponding to the target data flow record according to the prediction rules.

[0089] In the embodiments of this application, at least one path node can be determined from the target data flow record. The path node can refer to the participants or systems in the data flow process, such as the data sender, the data receiver, and at least one transit system.

[0090] Then, based on at least one path node and combined with historical data flow patterns in the flow knowledge base, the target flow path that the target data flow record may take can be predicted. The target flow path can refer to the flow sequence from the data sender to the data receiver, including at least one possible transit system. In the embodiments of this application, the prediction of the target flow path can be achieved, for example, by statistically analyzing the sequence of transit nodes that appear most frequently in at least one first sample data flow record. For example, in a cross-border remittance scenario, if historical data shows that when a remitting bank sends a remittance to a receiving bank, there is a high probability that it will be cleared through a specific transit bank, then this transit bank can be added as a missing node to form one or more complete expected target flow paths.

[0091] After determining the target flow path, at least one target sample data flow record that matches the target flow path can be further filtered from the retrieved first sample data flow record. That is, first sample data flow records whose historical flow paths are completely consistent with or highly similar to the predicted target flow path can be filtered. Subsequently, based on at least one target sample data flow record, the predicted transfer time corresponding to the target flow path can be determined according to the prediction rules described in the above embodiments.

[0092] In some embodiments, using the first prediction model to predict the predicted flow information corresponding to the target data flow record based on the historical behavior data of the data transferor matched by the query parameters can be specifically implemented as follows: Using the first prediction model, determine at least one historical data transfer record corresponding to the target transfer path from the historical behavior data of the data transferor that is matched by the query parameters, and predict the transfer time of the target transfer path corresponding to the target data transfer record based on the at least one historical data transfer record.

[0093] In the embodiments of this application, the historical behavior data may include long-tail features such as the clearing habits, common transit paths, processing time preferences, and path selection tendencies of the data transferor in a large number of past data transfer records.

[0094] During the prediction process, the first prediction model can first combine known path nodes (such as data sender and data receiver) in the query parameters, and through pattern mining and similarity analysis of the historical behavior data of the data sender, predict the most likely target flow path for the current target data flow record. For example, it can supplement the prediction of possible intermediate transit systems or clearing nodes, thereby forming one or more complete expected target flow paths. Subsequently, the first prediction model can retrieve and determine at least one historical data flow record that highly matches the target flow path from the historical behavior data of the data sender or the associated flow knowledge base. Preferably, the at least one historical data flow record is a complete flow case in which the data sender has actually used the path under similar conditions (such as the same or similar currency, amount, and receiver).

[0095] After obtaining these corresponding historical data flow records, the first prediction model can perform statistical aggregation or pattern inference based on the actual flow time recorded for each record to predict the transfer time of the target data flow record under the target flow path. For example, it can analyze the time distribution of matching historical records, calculate the median time and confidence interval, and combine the specific context of the current query (such as the transfer-out time) to infer the expected predicted transfer time. If multiple possible target flow paths are predicted, the first prediction model can further evaluate the historical probability of each path and prioritize outputting the predicted transfer time corresponding to the path with the highest probability, or provide alternative prediction results for multiple paths simultaneously.

[0096] In some embodiments, the method may further include: Statistical analysis is performed on multiple historical data flows to identify at least one data system whose data flow loss meets the loss criteria; based on at least one data system, exchange loss warning information is output to the data transfer recipient and / or data transferor.

[0097] In the embodiments of this application, information related to data transfer losses can be extracted from multiple historical data transfer records and statistically processed. Data transfer loss can refer to data loss caused by data type conversion or other operations forced by intermediate systems during the transfer process. For example, in cross-border remittance scenarios, this manifests as exchange losses incurred by a transit bank when converting funds in the original currency to another currency. By analyzing the currency changes, amount differences, and specific data systems involved in each transfer in historical records (including transit systems, sending-out systems, or receiving-out systems), the frequency, average loss magnitude, or loss probability of data transfer losses caused by each data system is statistically analyzed and compared with preset loss conditions. Loss conditions can be empirical thresholds, such as a loss probability higher than 30% or an average loss amount exceeding a certain proportion, thereby filtering out at least one data system that meets the conditions, typically a transit system whose historical data shows a high probability of exchange loss risk.

[0098] Once at least one data system is identified as meeting the loss criteria, corresponding data transfer loss alerts can be output to the data recipient and / or data sender. These alerts may include identification of the high-risk data system, a summary of historical loss statistics, and / or avoidance suggestions. For example, users may be advised to specify a low-risk transit system when initiating a transfer, add a note indicating the original currency for receipt, or choose an alternative transfer path to avoid data loss. In cross-border remittance scenarios, this function can be implemented by proactively alerting the remitter and / or recipient that "this path may involve exchange rate loss risk; it is recommended to specify the New York clearing path or add a 'original currency receipt' note" when the predicted return path involves a specific transit bank that historically shows frequently forcibly converts currency A to currency B. This effectively reduces the probability and amount of actual exchange rate loss.

[0099] By statistically analyzing historical data flow records, the risk of data system-level loss was quantitatively identified, and the risk information was promptly fed back to relevant parties. This not only enriched the practical value of predictive flow information but also provided users with actionable optimization suggestions, significantly reducing economic losses caused by intermediate system processing habits in practical applications.

[0100] In the embodiments of this application, the process of querying at least one first sample data flow record matching the query parameters from the flow knowledge base, and subsequently determining the selection of different prediction paths to predict flow information based on whether the query request meets the calculation requirements, can be implemented using an intelligent processing module. This makes the entire hybrid prediction process more efficient and intelligent. In some embodiments, querying at least one first sample data flow record matching the query parameters from the flow knowledge base can specifically be implemented as follows: using the intelligent processing module to search for at least one first sample data flow record matching the query parameters in the flow knowledge base; If the query request meets the computational requirements, the predicted flow information corresponding to the target data flow record is determined according to the prediction rules based on the flow information corresponding to at least one first sample data flow record. Otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data sender hit by the query parameters. Specifically, this can be achieved by using an intelligent processing module to determine the predicted flow information corresponding to the target data flow record according to the prediction rules based on the flow information corresponding to at least one first sample data flow record if the query request meets the computational requirements; otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data sender hit by the query parameters.

[0101] The intelligent processing module functions as an AI-based agent, possessing powerful data retrieval, analysis, and decision-making capabilities. During the query phase, it searches the knowledge base for multiple first-sample data flow records that match the query parameters. This module employs a progressive retrieval strategy based on multiple query factors in the query parameters to efficiently filter highly similar first-sample data flow records from the knowledge base, quickly returning reliable matching results even with a large knowledge base.

[0102] After completing the retrieval of the first sample data flow record, the intelligent processing module can further predict the flow information based on the found first sample data flow record, or call the first prediction model to predict the predicted flow information corresponding to the target data flow record.

[0103] In some embodiments, the method may further include: Obtain multiple historical data flow records; generate multiple sample data flow records based on these historical records; and construct a flow knowledge base from these sample data flow records.

[0104] In some embodiments, constructing a flow knowledge base from multiple sample data flow records can be specifically achieved by: grouping multiple sample data flow records according to the different field values ​​of predetermined record fields; and storing multiple sample data flow records into the flow knowledge base according to their respective groups.

[0105] In the embodiments of this application, historical data transfer records can first be obtained from different data processing systems. In a remittance scenario, these different data processing systems may include multiple financial institutions and clearing institutions participating in the data transfer. These systems generate detailed transfer trajectory information when processing actual transfer tasks. In the typical data transfer scenario of cross-border remittances, historical data transfer records can be generated by the Global Payments Innovation Service (GPI). GPI provides real-time or near real-time cross-border remittance tracking data through standardized messages on the SWIFT network (such as MT103). Each record can contain rich information such as the remitting bank, receiving bank, intermediary bank (if any), currency, amount, processing time at each node, and actual arrival time.

[0106] After obtaining the original historical data flow records, preprocessing can be performed on these records to generate sample data flow records. Preprocessing steps may include data cleaning, outlier removal, missing field completion, format standardization, and feature extraction. For example, this involves removing duplicate records, correcting obvious timeliness anomalies, normalizing records based on currency and amount, and extracting key flow information such as the complete path, total time consumption, and processing time of each transit system. This transforms the original heterogeneous data into structurally uniform and quality-controlled sample data flow records, meeting the needs of subsequent knowledge base storage and retrieval.

[0107] Subsequently, a knowledge base can be built from multiple processed sample data flow records. This knowledge base can be grouped and indexed according to predefined record fields (such as remitting bank, receiving bank, currency, amount range, country / region, etc.) to support fast, multi-dimensional matching queries. During the construction process, sample data flow records can be grouped and stored according to different field values; for example, records from the same remitting bank to the same receiving bank or of the same currency can be grouped into the same group. Simultaneously, statistical indicators within each group (such as path distribution probability, median time consumption, common transit systems, etc.) can be calculated and stored, thus forming a knowledge base that supports incremental updates and dynamic expansion.

[0108] By completing the entire process from acquiring historical data flow records generated from the original GPI, to generating standardized samples, and then to building a structured flow knowledge base, a basic closed loop for data-driven prediction is achieved. This ensures that the flow knowledge base can continuously reflect the real characteristics of data flow, thereby improving the robustness, accuracy, and practical application value of flow information prediction.

[0109] In some embodiments, retrieving at least one sample data flow record from the flow knowledge base that matches the query parameters can be specifically implemented as follows: Based on the parameter values ​​of the corresponding predefined record fields in the query parameters, determine the target group; from the data records corresponding to the target group in the flow knowledge base, find at least one sample data flow record that matches the query parameters.

[0110] As shown above, during the construction of the knowledge base, multiple sample data flow records are grouped and stored according to the different field values ​​of the predefined record fields, so that sample data flow records with the same field values ​​are grouped into the same group.

[0111] In actual querying, the query parameters can be parsed first to extract the parameter values ​​corresponding to the predetermined record fields, and then used as grouping keys to directly locate the target group in the knowledge base. If the query parameters are range values ​​(such as a monetary range) on certain fields, they may correspond to multiple adjacent groups. In this case, these groups can be considered together as the target group range. Subsequently, further fine-grained matching and searching are performed only on the data records corresponding to the target group, without traversing the entire knowledge base, thereby significantly reducing the retrieval scope and computational overhead.

[0112] In some embodiments, the method may further include: In response to a query request, determine the data transfer certificate provided by the data transferee; identify the query parameters in the data transfer certificate; if the query parameters have at least one missing factor, identify at least one missing factor from the historical behavior data corresponding to the data transferee; update the query parameters based on at least one missing factor.

[0113] Among them, data transfer vouchers can be electronic or image vouchers that prove that data transfer has been initiated. For example, in cross-border remittance scenarios, data transfer vouchers can be provided as user-uploaded bank slip images, PDF files, bank transfer confirmation screenshots, or text information containing key fields.

[0114] After obtaining the data transfer voucher, it can be processed to extract query parameters. For example, Optical Character Recognition (OCR) technology can be used to parse the text content of the data transfer voucher, identifying fields such as the printed or handwritten remittance bank name, payee bank information, currency, amount, remittance date, and reference number, and using these recognition results as query parameters. If the voucher is in plain text format, query parameters can be obtained directly through regular expressions or entity extraction.

[0115] However, in real-world scenarios, user-provided data transfer documents are often incomplete. For example, bank slip images may be blurry, making certain fields unrecognizable, or the data transfer document itself may only contain partial information. In such cases, if at least one missing factor is detected in the query parameters, it can be further identified from the historical behavioral data of the data transferor. Historical behavioral data refers to the habitual characteristics of the data transferor in past transfer records, such as frequently used remittance banks, typical currencies, amount distribution, or cooperating payers. This data can be obtained from the transfer knowledge base or user historical query records. By matching the identifier of the data transferor, its historical sample data transfer records are retrieved, and the most likely missing factor value is inferred based on statistical frequency or pattern matching. For example, if more than 90% of the recipient's historical remittances come from a specific remittance bank and use USD as the currency, then that remittance bank and USD are used as the completion value for the missing factor.

[0116] Finally, based on at least one identified missing factor, the original query parameters are updated to form a complete set of query parameters. These updated parameters can then be used to retrieve matching samples from the data transfer knowledge base or drive predictive models, ensuring accurate predictions of data transfer information even when data transfer documentation is incomplete. In cross-border remittance scenarios, this feature allows users to simply upload a bank slip image; even if some fields in the image are blurry or missing, the system can intelligently complete the information based on the recipient's historical remittance behavior, reducing the burden of manual input for users.

[0117] In some embodiments, the method may further include: determining whether there is an anomaly at the data receiving end or the data sending end; if so, outputting an anomaly message.

[0118] In the embodiments of this application, after responding to a query request and determining the query parameters, or during the process of generating predicted data transfer information, anomaly detection can be performed on the data transferor and / or data transferor involved in the target data transfer record. Anomaly detection can be based on predefined anomaly detection rules and historical data analysis. For example, it can check whether the data transferor or data transferee belongs to a risky country or region, whether it involves entities on a sanctions list, whether there are abnormally frequent large-amount transfers, whether it deviates significantly from historical behavior patterns, or whether the account status is frozen or restricted. Anomalies can be comprehensively assessed by accessing external compliance databases, internal blacklist databases, or by performing statistical deviation analysis based on the historical behavior data of the transferor / transferor in the data transfer knowledge base.

[0119] If the assessment results indicate an anomaly at either the data recipient or the data sender, an anomaly alert can be output. This alert can include a clear description of the anomaly type, the risk level, potential impacts (e.g., delays, rejections, or triggering of additional checks), and suggested actions (e.g., contacting the bank for confirmation, changing the remittance route, or providing supplementary materials). In cross-border remittance scenarios, for example, when query parameters show that the remitting or receiving bank is located in a country deemed high-risk, or when the receiving bank's historical remittance records show an abnormal pattern, the user initiating the query (usually the data recipient) can be prompted with "The receiving bank's location is in a region with strict foreign exchange controls, which may lead to additional audits and delays in fund arrival. Please confirm the remittance information or consult the bank," thus helping the user promptly avoid potential risks of fund retention or loss.

[0120] This application embodiment integrates a proactive risk control mechanism while providing predictive transfer information, enhancing users' confidence in the transfer process and effectively reducing uncontrollable timeliness or fund security issues caused by abnormalities of the transferee or transferor.

[0121] In some embodiments, before determining the query parameters and the target data flow record to which the query parameters are matched in response to a query request, the method may further include: Obtain the query request; check if there is target flow information corresponding to the target data flow record that matches the query parameters; if yes, output the target flow information; if no, respond to the query request to determine the query parameters and the target data flow record that the query parameters match.

[0122] In the embodiments of this application, before determining the query parameters and the target data flow record matched by the query parameters in response to the query request, the method may further query whether there is target flow information corresponding to the target data flow record that matches the query parameters; if so, the target flow information is directly output; otherwise, the subsequent determination of query parameters and target data flow record continues, thereby achieving priority processing of completed or real-time trackable flow records, avoiding unnecessary prediction calculations, and improving the system's response efficiency and user experience.

[0123] In some embodiments, the method may further include: Obtain historical data flow records; use the second prediction model to generate prediction rules based on the historical data flow records.

[0124] In the embodiments of this application, the second prediction model can automatically discover and extract patterns that affect the timeliness of data flow from a large number of historical data flow records. For example, it can learn high-probability data flow paths under specific combinations of data senders and data receivers, typical processing time distributions of various transit systems, and time-extending patterns during holidays, thereby generating a set of structured prediction rules.

[0125] Similar to the first prediction model, the second prediction model can also be an artificial intelligence-based language model (LM) or multimodal model (MM), etc. The embodiments of this application do not limit the number of model parameters supported by the model, aiming to meet actual needs. The first and second prediction models can be implemented as the same model, but are not limited to this, and can also be different models.

[0126] Figure 4 A flowchart illustrating an embodiment of an information query method provided in this application, wherein the method can be executed by a server, such as... Figure 4 As shown, the method may specifically include the following steps: 401: Query request received; the query request includes query parameters.

[0127] 402: Query whether there is target flow information corresponding to the target data flow record that matches the query parameters.

[0128] 403: If so, output the target flow information.

[0129] 404: If not, query the flow knowledge base for multiple first sample data flow records that match the query parameters; the flow knowledge base is constructed based on multiple historical data flow records.

[0130] 405: Determine if the query request meets the calculation requirements. If yes, proceed to step 406; otherwise, proceed to step 407.

[0131] 406: Based on the flow information corresponding to multiple first sample data flow records, calculate the predicted flow information corresponding to the target data flow record according to the prediction rules.

[0132] 407: Using the first prediction model, based on the historical behavior data of the data transfer party matched by the query parameters, predict the predicted flow information corresponding to the target data flow record.

[0133] 408: Output predicted flow information.

[0134] This embodiment of the application, after determining the query parameters and target data flow record in response to a query request, first queries at least one first sample data flow record matching the query parameters from the flow knowledge base. If the query request meets the calculation requirements, based on the flow information corresponding to each of the at least one first sample data flow record, predictive flow information is determined according to prediction rules. Otherwise, a first prediction model is used to predict based on the historical behavior data of the data transferor matched by the query parameters. Finally, the generated predicted flow information is output as the query result of the query request. This achieves the prediction of flow information. When the query information cannot be obtained, the predicted flow information can be output as the query result, thereby achieving effective querying, ensuring query results, and improving the user query experience. Furthermore, this application's embodiments employ a hybrid prediction mechanism combining rules and AI models. In scenarios where at least one first sample data flow record can be matched from the flow knowledge base and the query request meets the computational requirements, rules are used to quickly provide statistically based flow information predictions. In scenarios where at least one first sample data flow record cannot be matched from the flow knowledge base and the query request does not meet the computational requirements, the first prediction model can be used to deeply mine the historical behavioral data of the data transferor for intelligent prediction. This effectively solves the technical problem of data flow records being unable to obtain effective flow information due to outdated flow information, significantly improving query coverage and query results. It enables scenarios that previously could not provide results to obtain accurate predicted flow information, reduces users' reliance on manual verification, and ensures the effectiveness and practicality of data flow queries.

[0135] The detailed implementation methods and beneficial effects of each step in this embodiment have been described in detail in the foregoing embodiments, and will not be elaborated here.

[0136] Figure 5 This is a flowchart illustrating another embodiment of the information prediction method provided in this application. This embodiment uses a remittance scenario as an example to introduce the technical solution of this application. The method can be executed by a server, such as... Figure 5 As shown, the method may specifically include the following steps: 501: Responds to a query request, specifying the query parameters and the target remittance record matched by those parameters; 502: Find at least one first-sample remittance record in the Huilu Knowledge Base that matches the query parameters; the Huilu Knowledge Base is built and generated based on multiple historical remittance records; 503: Determine if the query request meets the calculation requirements. If yes, proceed to step 504; otherwise, proceed to step 505.

[0137] 504: Based on the remittance time data corresponding to at least one first sample remittance record, determine the predicted arrival time corresponding to the target remittance record according to the prediction rules.

[0138] 505: Using the first prediction model, based on the historical behavior data of the remitter corresponding to the query parameters, predict the expected arrival time of the target remittance record; the predicted arrival time is used as the query result of the query request.

[0139] This embodiment of the application, after determining the query parameters and target remittance record in response to a query request, first queries the remittance knowledge base for at least one first sample remittance record matching the query parameters. If the query request meets the calculation requirements, based on the remittance time data corresponding to each of the at least one first sample remittance record, the predicted arrival time is determined according to prediction rules. Otherwise, a first prediction model is used to predict the arrival time based on the historical behavior data of the remitter matched by the query parameters. Finally, the generated predicted arrival time is output as the query result of the query request. This achieves the prediction of arrival time, and when query information is unavailable, the predicted arrival time can be output as the query result, thus achieving effective querying, ensuring query results, and improving the user query experience. Furthermore, this application's embodiments employ a hybrid prediction mechanism combining rules and AI models. In scenarios where at least one first sample remittance record can be matched from the remittance knowledge base and the query request meets the computational requirements, rules are used to quickly provide statistically based arrival time predictions. In scenarios where at least one first sample remittance record cannot be matched from the remittance knowledge base and the query request does not meet the computational requirements, the first prediction model can be used to deeply mine the remitter's historical behavioral data for intelligent prediction. This effectively solves the technical problem of remittance records not being able to obtain valid arrival times due to untimely updates of remittance information, significantly improving query coverage and query results. It enables scenarios that previously could not provide results to obtain accurate predicted arrival times, reduces users' reliance on manual verification, and ensures the effectiveness and practicality of remittance arrival time queries.

[0140] In some embodiments, at least one first sample remittance record includes multiple first sample remittance records; The above-mentioned method of determining the predicted arrival time of the target remittance record according to the prediction rules can be implemented as follows: determine the remittance time corresponding to each of the multiple first sample remittance records to obtain multiple remittance times; count the multiple remittance times to determine the first arrival time that belongs to the median or the second arrival time that exceeds the predetermined proportion; and calculate the predicted arrival time based on the remittance time corresponding to the target remittance record, as well as the first arrival time and / or the second arrival time.

[0141] In some embodiments, determining the predicted arrival time of the target remittance record based on the remittance time data corresponding to at least one first sample remittance record, according to the prediction rule, can be specifically implemented as follows: Based on the remittance time corresponding to at least one first sample remittance record, determine the candidate arrival time corresponding to the target remittance record according to the prediction rules; determine at least one influencing factor that the target remittance record hits, and update the candidate arrival time according to the change conditions corresponding to at least one influencing factor to obtain the predicted arrival time; at least one influencing factor includes one or more of the following: data transfer time is during a holiday, data transfer time is during a holiday, the transferred data meets the predetermined conditions, the data transfer region is the target type region, and the data transfer region is the target type region.

[0142] In some embodiments, when the query request meets the calculation requirements, the predicted arrival time of the target remittance record is determined based on the remittance time data corresponding to at least one first sample remittance record and according to the prediction rules. Specifically, if there are a predetermined number of first sample remittance records in the remittance knowledge base that match the query parameters, the predicted arrival time of the target remittance record is determined according to the remittance time corresponding to each of the at least one first sample remittance record and according to the prediction rules.

[0143] In some embodiments, the query parameters include at least one query factor; In some embodiments, finding at least one first sample remittance record matching the query parameters in the remittance knowledge base can be specifically implemented as follows: using a predetermined quantity as a constraint, according to the combination requirements, one or more query factors are used as query conditions in sequence to find at least one sample remittance record matching the query conditions in the remittance knowledge base until a predetermined number of sample exchange rate records are obtained or the query conditions are empty.

[0144] In some embodiments, using a first prediction model to predict the predicted arrival time of a target remittance record based on the historical behavior data of the remitter corresponding to the remitter matched by the query parameters includes: determining at least one second sample remittance record corresponding to the remitter matched by the query parameters from the remittance knowledge base; determining at least one influencing factor matched by the target remittance record; and using the first prediction model to predict the predicted arrival time of the target remittance record based on the query parameters, at least one second sample remittance record, and the change conditions corresponding to different influencing factors.

[0145] In some embodiments, the method may further include: Identify at least one path node in the target transfer record; based on the at least one path node, predict the target remittance path corresponding to the target transfer record; In some embodiments, determining the predicted arrival time of the target remittance record based on remittance time data corresponding to at least one first sample remittance record, according to prediction rules, includes: From at least one first sample remittance record, identify at least one target sample remittance record that the target remittance path matches; based on at least one target sample remittance record, determine the predicted arrival time of the target remittance path corresponding to the target remittance record according to the prediction rules.

[0146] In some embodiments, the predicted arrival time of the target remittance record is predicted using a first prediction model based on the historical behavior data of the remitter matched by the query parameters, including: Using the first prediction model, at least one historical remittance record corresponding to the target remittance path is determined from the historical behavior data of the remitter whose query parameters match. Based on the at least one historical remittance record, the predicted arrival time of the target remittance path corresponding to the target remittance record is predicted.

[0147] In some embodiments, the method further includes: By analyzing multiple historical remittance records, at least one financial institution with exchange losses that meets the loss criteria can be identified. Based on at least one financial institution that provides exchange loss warnings, the remitter and / or recipient are provided with exchange loss alert information.

[0148] In some embodiments, searching the remittance knowledge base for at least one sample remittance record that matches the query parameters includes: The intelligent processing module searches the remittance knowledge base for at least one first-sample remittance record that matches the query parameters. If the query request meets the computational requirements, based on the remittance time data corresponding to at least one first-sample remittance record, the predicted arrival time of the target remittance record is determined according to the prediction rules. Otherwise, the first prediction model is used to predict the predicted arrival time of the target remittance record based on the historical behavior data of the remitter matched by the query parameters. Specifically, this can be implemented as follows: Using the intelligent processing module, if the query request meets the calculation requirements, the predicted arrival time of the target remittance record is determined according to the prediction rules based on the remittance time data corresponding to at least one first sample remittance record. Otherwise, the predicted arrival time of the target remittance record is predicted based on the historical behavior data of the remitter corresponding to the remitter matched by the query parameters using the first prediction model.

[0149] In some embodiments, the method may further include: Obtain multiple historical remittance records; the record information in the historical remittance records is generated by the Global Payments Innovation Service (GPI); based on the multiple historical remittance records, generate multiple sample remittance records; construct a remittance knowledge base from the multiple sample remittance records.

[0150] In some embodiments, constructing a remittance knowledge base from multiple sample remittance records can be specifically implemented as follows: Multiple sample remittance records are grouped according to the different field values ​​of the predefined record fields; the multiple sample remittance records are stored in the remittance knowledge base according to their respective groups.

[0151] In some embodiments, searching the remittance knowledge base for at least one sample remittance record that matches the query parameters includes: Based on the parameter values ​​of the corresponding predefined record fields in the query parameters, determine the target group; from the data records corresponding to the target group in the Huilu knowledge base, find at least one sample remittance record that matches the query parameters.

[0152] In some embodiments, the method may further include: In response to a query request, determine the remittance voucher provided by the payee; identify the query parameters in the remittance voucher; if the query parameters have at least one missing factor, identify at least one missing factor from the historical behavior data corresponding to the payee; update the query parameters based on at least one missing factor.

[0153] In some embodiments, before responding to a query request and determining the query parameters and the target remittance record to which the query parameters are matched, the method may further include: Get the query request; check if there is an arrival time corresponding to the target remittance record that matches the query parameters; if yes, output the arrival time; if no, respond to the query request, determine the query parameters and the target remittance record that the query parameters match.

[0154] The detailed implementation methods and beneficial effects of each step in this embodiment have been described in detail in the foregoing embodiments, and will not be elaborated here.

[0155] Figure 6 A flowchart illustrating another embodiment of the information query method provided in this application, which can be executed by a server, such as... Figure 6 As shown, the method may specifically include the following steps: 601: Query request received; the query request includes query parameters; 602: Check if there is a corresponding arrival time for the target remittance record that matches the query parameters; 603: If yes, output the arrival time; 604: If not, search the Huilu Knowledge Base for at least one first-sample remittance record that matches the query parameters; the Huilu Knowledge Base is built based on multiple historical remittance records. 605: Determine whether the query request meets the calculation requirements. If yes, proceed to step 606; otherwise, proceed to step 607.

[0156] 606: Based on the remittance time data corresponding to at least one first sample remittance record, determine the predicted arrival time of the target remittance record according to the prediction rules.

[0157] 607: Using the first prediction model, based on the historical behavior data of the remitter corresponding to the query parameters, predict the arrival time of the target remittance record; 608: Output the predicted arrival time of funds.

[0158] This embodiment of the application, after determining the query parameters and target remittance record in response to a query request, first queries the remittance knowledge base for at least one first sample remittance record matching the query parameters. If the query request meets the calculation requirements, based on the remittance time data corresponding to each of the at least one first sample remittance record, the predicted arrival time is determined according to prediction rules. Otherwise, a first prediction model is used to predict the arrival time based on the historical behavior data of the remitter matched by the query parameters. Finally, the generated predicted arrival time is output as the query result of the query request. This achieves the prediction of arrival time, and when query information is unavailable, the predicted arrival time can be output as the query result, thus achieving effective querying, ensuring query results, and improving the user query experience. Furthermore, this application's embodiments employ a hybrid prediction mechanism combining rules and AI models. In scenarios where at least one first sample remittance record can be matched from the remittance knowledge base and the query request meets the computational requirements, rules are used to quickly provide statistically based arrival time predictions. In scenarios where at least one first sample remittance record cannot be matched from the remittance knowledge base and the query request does not meet the computational requirements, the first prediction model can be used to deeply mine the remitter's historical behavioral data for intelligent prediction. This effectively solves the technical problem of remittance records not being able to obtain valid arrival times due to untimely updates of remittance information, significantly improving query coverage and query results. It enables scenarios that previously could not provide results to obtain accurate predicted arrival times, reduces users' reliance on manual verification, and ensures the effectiveness and practicality of remittance arrival time queries.

[0159] The detailed implementation methods and beneficial effects of each step in this embodiment have been described in detail in the foregoing embodiments, and will not be elaborated here.

[0160] To facilitate understanding, the technical solution of this application will be introduced using a remittance scenario as an example to predict the arrival time of funds. Figure 7 The diagram shown is a signaling flowchart of the time prediction process in a practical application according to an embodiment of this application.

[0161] The user initiates a query request through the user terminal 101 (step 701). This query request can use various input methods, including manually filling in remittance information, uploading a remittance slip image or PDF, etc. After receiving the query request, the server 102 can first check whether there is real-time GPI tracking information corresponding to the query. If real-time GPI tracking exists and the funds have been received or are being tracked, the actual arrival time or current transfer status is directly returned as the query result; if no real-time GPI tracking information exists, the prediction process begins.

[0162] In response to the query request, server 102 first identifies the remittance voucher (such as a bank slip image) provided by the payee. It then uses OCR technology combined with natural language processing to identify the query parameters in the voucher (step 702), including the remitting bank, receiving bank, currency, amount, and remittance date. If the identification result contains missing factors (such as a blurred bank slip preventing the remitting bank from being identified), the most likely missing factor is inferred from the payee's historical remittance data, and the query parameters are updated accordingly (step 703). Simultaneously, server 102 also checks for any anomalies in the remitting or receiving party (step 704). If an anomaly is found, an anomaly alert is immediately sent to the user, highlighting potential delays or compliance risks (step 705).

[0163] After the query parameters are complete, the server 102 identifies the remittance corresponding to the current query as the target remittance record and uses the intelligent processing module (AI Agent) to search for matching sample remittance records in the pre-built remittance knowledge base 103 (step 705). The construction process of the remittance knowledge base 103 can be as follows: obtain a large number of historical remittance records from multiple financial institutions participating in GPI, generate sample remittance records after data cleaning, and then group and store them according to predetermined record fields to form a grouped remittance knowledge base 103 that supports efficient retrieval.

[0164] The intelligent processing module can employ a progressive retrieval strategy, using a predetermined number (e.g., 100 records) as a constraint. It sequentially uses the combination of query factors (e.g., prioritizing remittance financial institutions + receiving financial institutions + currency, then relaxing to remittance financial institutions + receiving financial institutions, and finally only remittance financial institutions) as query conditions. First, it locates the target group based on the values ​​of key fields in the query parameters. Then, it searches for at least one matching sample remittance record within that group. If more than the predetermined number of sample remittance records are ultimately retrieved, the query request is deemed to meet the calculation requirements, and the system proceeds to the rule-based prediction path; otherwise, it enters the AI ​​prediction path.

[0165] In high deterministic scenarios that meet computational requirements, the intelligent processing module determines the predicted arrival time of funds based on multiple matched sample remittance records and in accordance with prediction rules (step 707). First, historical remittance paths are extracted from sample remittance records to determine one or more high-probability remittance paths most likely to be used by the target remittance (such as direct remittances within the same bank or common transit paths). Then, for each remittance path, the financial institutions involved (including transit banks) are identified, and the processing time of each financial institution is predicted based on historical statistics and accumulated to obtain the total path time. Next, the remittance times of multiple matching samples are statistically analyzed, and the median time (first arrival time) and RT90 time (second arrival time) are calculated. Then, the first arrival time and the second arrival time are added to the remittance time respectively to obtain the first arrival time and the second arrival time. The time interval formed by the first arrival time and the second arrival time is used as the preliminary predicted arrival time. Finally, the influencing factors of the target remittance are identified (such as the remittance time or expected arrival time involving holidays, large amounts triggering review, remittance / receiver regions being risky regions, etc.), and the predicted arrival time is updated according to the corresponding change conditions (such as holidays being extended by 1-2 working days). Finally, the arrival time window is output and sent to the query user terminal 101.

[0166] In low-determinism scenarios that do not meet computational requirements, even if the number of matched sample remittance records is insufficient, the intelligent processing module can still retrieve all historical remittance records of the remitter (remitting bank) as supplementary behavioral data, and then call the first prediction model to predict the arrival time (step 708). By dynamically filling query parameters, remitter historical samples, and changes in influencing factors through preset prompt word templates, a structured prompt word input model is constructed. This allows the first prediction model to comprehensively analyze the remitter's clearing habits, long-tail path preferences, and the influence of dynamic factors, output the predicted arrival time, and send it to the query user terminal 101 (step 709).

[0167] In addition, the server 102 will also statistically analyze historical remittance records, identify financial institutions that meet the loss conditions for exchange loss probability or magnitude, and output exchange loss warning information to the remitter and / or recipient in the prediction results, recommending to designate a low-risk intermediary bank or add the "original currency deposit" note.

[0168] Prediction rules can be automatically generated and optimized periodically based on historical remittance records through a second prediction model, ensuring that the rules adapt to changes in bank clearing practices.

[0169] This embodiment of the application, after determining the query parameters and target remittance record in response to a query request, first queries the remittance knowledge base for at least one sample remittance record matching the query parameters. If the query request meets the calculation requirements, the predicted arrival time is determined according to the prediction rules based on the remittance time data corresponding to the matching sample remittance records. Otherwise, a first prediction model is used to predict the arrival time based on the historical behavior data of the remitter matched by the query parameters. Finally, the generated predicted arrival time is output as the query result of the query request. This achieves the prediction of arrival time. When query information is unavailable, the predicted arrival time can be output as the query result, thus achieving effective querying, ensuring query results, and improving the user query experience. Furthermore, this application's embodiments employ a hybrid prediction mechanism combining rules and AI models. In scenarios where at least one first sample remittance record can be matched from the remittance knowledge base and the query request meets the computational requirements, rules are used to quickly provide statistically based arrival time predictions. In scenarios where at least one first sample remittance record cannot be matched from the remittance knowledge base and the query request does not meet the computational requirements, the first prediction model can be used to deeply mine the remitter's historical behavioral data for intelligent prediction. This effectively solves the technical problem of remittance records not being able to obtain valid arrival times due to untimely updates of remittance information, significantly improving query coverage and query results. It enables scenarios that previously could not provide results to obtain accurate predicted arrival times, reduces users' reliance on manual verification, and ensures the effectiveness and practicality of remittance arrival time queries.

[0170] Figure 8 This is a schematic diagram of one embodiment of an information prediction device provided in this application. The device can be deployed on a server, such as... Figure 8 As shown, the device may include: The first determining module 801 is used to respond to a query request and determine the query parameters and the target data flow record to which the query parameters are matched. The first query module 802 is used to query at least one first sample data flow record that matches the query parameters from the flow knowledge base; wherein, the flow knowledge base is constructed based on multiple historical data flow records; The first prediction module 803 is used to determine the predicted flow information corresponding to the target data flow record according to the prediction rules based on the flow information corresponding to at least one first sample data flow record when the query request meets the calculation requirements; otherwise, it uses the first prediction model to predict the predicted flow information corresponding to the target data flow record based on the historical behavior data of the data transferor hit by the query parameters; the predicted flow information is used as the query result of the query request.

[0171] In some embodiments, at least one first sample data flow record includes multiple first sample data flow records; In some embodiments, the first prediction module 803 is specifically used for: Determine the data flow time corresponding to each of the multiple first sample data flow records to obtain multiple data flow times; Statistical analysis of multiple data processing times is conducted to determine the first data processing time that falls within the median or the second data processing time that exceeds a predetermined proportion. The predicted transfer-in time is calculated based on the data transfer-out time in the target data transfer record, as well as the first data transfer time and / or the second data transfer time.

[0172] In some embodiments, the first prediction module 803 is specifically used for: Based on the flow information corresponding to at least one first sample data flow record, determine the candidate transfer time corresponding to the target data flow record according to the prediction rule; Identify at least one influencing factor that the target data transfer record hits, and update the candidate transfer-in time according to the change conditions corresponding to each of the at least one influencing factor to obtain the predicted transfer-in time; the at least one influencing factor includes one or more of the following: data transfer-out time is a holiday, data transfer-in time is a holiday, the transferred data meets the predetermined conditions, the data transfer-out region is a target type region, and the data transfer-in region is a target type region.

[0173] In some embodiments, the first prediction module 803 is specifically used for: If there are a predetermined number of first sample data flow records in the flow knowledge base that match the query parameters, the predicted flow information corresponding to the target data flow record is determined according to the prediction rules based on the flow information corresponding to at least one first sample data flow record.

[0174] In some embodiments, the query parameters include at least one query factor; In some embodiments, the first query module 802 is specifically used for: With a predetermined quantity as a constraint, and according to the combination requirements, one or more query factors are used as query conditions in sequence to find at least one first sample data flow record that matches the query conditions in the flow knowledge base, until the predetermined number of first sample data flow records are obtained.

[0175] In some embodiments, the first prediction module 803 is specifically used for: Determine at least one second sample data flow record corresponding to the data transferor whose query parameters match from the flow knowledge base; determine at least one influencing factor whose target data flow record matches. Using the first prediction model, based on the query parameters, at least one second sample data flow record, and the change conditions corresponding to different influencing factors, the predicted flow information corresponding to the target data flow record is predicted.

[0176] In some embodiments, the device may further include: The path node determination module is used to determine at least one path node in the target data flow record; The path determination module is used to predict the target flow path corresponding to the target data flow record based on at least one path node. In some embodiments, the first prediction module 803 is specifically used for: From at least one first sample data flow record, determine at least one target sample data flow record that the target flow path hits; Based on at least one target sample data flow record, determine the predicted transfer time of the target flow path corresponding to the target data flow record according to the prediction rules.

[0177] In some embodiments, the first prediction module 803 is specifically used for: Using the first prediction model, determine at least one historical data transfer record corresponding to the target transfer path from the historical behavior data of the data transferor that is matched by the query parameters, and predict the transfer time of the target transfer path corresponding to the target data transfer record based on the at least one historical data transfer record.

[0178] In some embodiments, the device may further include: The loss statistics module is used to statistically analyze multiple historical data flow records and determine at least one data system whose data flow loss meets the loss conditions. The loss alert module is used to output data loss alert information to the data transfer recipient and / or data transferor based on at least one data system.

[0179] In some embodiments, the first query module 802 is specifically used for: The intelligent processing module is used to search for multiple first sample data flow records in the flow knowledge base that match the query parameters; In some embodiments, the first prediction module 803 is specifically used for: Using the intelligent processing module, if the query request meets the calculation requirements, the predicted flow information corresponding to the target data flow record is determined according to the prediction rules based on the flow information corresponding to at least one first sample data flow record; otherwise, the predicted flow information corresponding to the target data flow record is predicted based on the historical behavior data of the data transferor hit by the query parameters using the first prediction model.

[0180] In some embodiments, the device may further include: The historical record determination module is used to obtain multiple historical data flow records; The sample generation module is used to generate multiple sample data flow records based on multiple historical data flow records; The knowledge base construction module is used to build a circulation knowledge base based on multiple sample data circulation records.

[0181] In some embodiments, the knowledge base construction module is specifically used for: Group multiple sample data flow records according to the different field values ​​of the predefined record fields; Multiple sample data transfer records are stored in the transfer knowledge base according to their respective groups.

[0182] In some embodiments, the first query module 802 is specifically used for: Determine the target group based on the parameter values ​​of the corresponding predefined record fields in the query parameters; From the data records corresponding to the target group in the knowledge base, find at least one sample data flow record that matches the query parameters.

[0183] In some embodiments, the device may further include: The voucher receiving module is used to respond to query requests and determine the data transfer voucher provided by the data transferee; The parameter recognition module is used to identify query parameters in data transfer vouchers; The missing completion module is used to identify at least one missing factor from the historical behavior data corresponding to the data importer when there is at least one missing factor in the query parameters. The parameter update module is used to update query parameters based on at least one missing factor.

[0184] In some embodiments, the device may further include; The request retrieval module is used to retrieve query requests. The information query module is used to query whether there is target flow information corresponding to the target data flow record that matches the query parameters; The information output module is used to output the target flow information when there is target flow information corresponding to the target data flow record that matches the query parameters; The parameter determination module is used to determine the query parameters and the target data flow record that the query parameters match when there is no target flow information corresponding to the target data flow record that matches the query parameters.

[0185] Figure 8 Information prediction devices can perform Figure 2The implementation principle and technical effects of the information prediction method in the illustrated embodiments will not be repeated here. The specific methods by which each module and unit of the information prediction device in the above embodiments performs its operations have been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0186] Figure 9 This is a schematic diagram of one embodiment of an information query device provided in this application. The device can be deployed on a server, such as... Figure 9 As shown, the device may include: The first acquisition module 901 is used to acquire a query request; the query request includes query parameters. The second query module 902 is used to query whether there is target flow information corresponding to the target data flow record that matches the query parameters; The first output module 903 is used to output the target flow information when there is target flow information corresponding to the target data flow record that matches the query parameters; The third query module 904 is used to query multiple first sample data flow records that match the query parameters from the flow knowledge base when there is no target flow information corresponding to the target data flow record that matches the query parameters; the flow knowledge base is constructed based on multiple historical data flow records. The second prediction module 905 is used to calculate the predicted flow information corresponding to the target data flow record according to the prediction rules based on the flow information corresponding to the multiple first sample data flow records when the query request meets the calculation requirements; otherwise, it uses the first prediction model to predict the predicted flow information corresponding to the target data flow record based on the historical behavior data of the data transferor hit by the query parameters. The second output module 906 is used to output predicted flow information.

[0187] Figure 9 The information query device can perform Figure 4 The implementation principle and technical effects of the information query method in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the information query device in the above embodiments performs its operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0188] Figure 10 This is a schematic diagram of another embodiment of the information prediction device provided in this application. This device can be deployed on a server, such as... Figure 10 As shown, Figure 10 As shown, the device may include: The second determining module 1001 is used to respond to the query request and determine the query parameters and the target remittance record matched by the query parameters. The fourth query module 1002 is used to find at least one sample remittance record in the Huilu knowledge base that matches the query parameters; the Huilu knowledge base is built and generated based on multiple historical remittance records. The third prediction module 1003 is used to determine the predicted arrival time of the target remittance record based on the remittance time data corresponding to at least one sample remittance record and in accordance with the prediction rules, if the query request meets the calculation requirements; otherwise, it uses the first prediction model to predict the predicted arrival time of the target remittance record based on the historical behavior data of the remitter corresponding to the remitter matched by the query parameters; the predicted arrival time is used as the query result of the query request.

[0189] Figure 10 The time prediction device can perform Figure 5 The implementation principle and technical effects of the time prediction method described in the illustrated embodiments will not be repeated here. The specific methods by which each module and unit of the time prediction device in the above embodiments performs its operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0190] Figure 11 This is a schematic diagram of another embodiment of an information query device provided in this application. This device can be deployed on a server, such as... Figure 11 As shown, the device may include: The second acquisition module 1101 is used to acquire a query request; the query request includes query parameters. The fifth query module 1102 is used to query whether there is an arrival time corresponding to the target remittance record that matches the query parameters; The third output module 1103 is used to output the arrival time when there is an arrival time corresponding to the target remittance record that matches the query parameters; The sixth query module 1104 is used to search for at least one first sample remittance record that matches the query parameters in the remittance knowledge base when there is no arrival time corresponding to the target remittance record that matches the query parameters; the remittance knowledge base is built and generated based on multiple historical remittance records; The fourth prediction module 1105 is used to determine the predicted arrival time of the target remittance record based on the remittance time data corresponding to the at least one first sample remittance record, according to the prediction rules, if the query request meets the calculation requirements; otherwise, it uses the first prediction model to predict the predicted arrival time of the target remittance record based on the historical behavior data of the remitter matched by the query parameters. The fourth output module 1106 is used to output the predicted arrival time.

[0191] Figure 11 The time query device can perform... Figure 6 The implementation principle and technical effects of the time query method described in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the time query device in the above embodiments performs its operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0192] It should be noted that some processes described in the above embodiments and accompanying drawings include multiple operations appearing in a specific order. However, it should be clearly understood that these operations may not be executed in the order they appear in this document, or they may be executed in parallel. The operation numbers, such as 201, 202, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should also be noted that the descriptions such as "first" and "second" in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0193] Figure 12 This is a schematic diagram of the structure of one embodiment of a computing device provided in this application. Figure 12 As shown, in practice, the computing device may include a storage component 1201 and a processing component 1202.

[0194] Storage component 1201 is used to store computer programs and can be configured to store various other data to support operation on a computing device. Examples of this data include instructions for any application or method operating on the computing device, data structures, contact data, phone book data, messages, pictures, videos, etc.

[0195] The processing component 1202, coupled to the storage component 1201, is used to execute the computer program in the storage component 1201 to implement the information prediction method, information query method, and time prediction method provided in the embodiments of this application.

[0196] Furthermore, such as Figure 12 As shown, the computing device may also include other components such as a communication component 1203, a display component 1204, a power supply component 1205, and an audio component 1206. Figure 12 The diagram only shows some components and does not mean that the device includes only these components. Figure 12 The components shown. Additionally... Figure 12The components within the dashed box are optional, not mandatory, and their specific requirements depend on the product form of the computing device. The computing device in this embodiment can be a terminal device such as a desktop computer, laptop computer, smartphone, or IoT (Internet of Things) device, or a server-side device such as a conventional server, cloud server, or server array. If the computing device in this embodiment is implemented as a terminal device such as a desktop computer, laptop computer, or smartphone, it may include... Figure 12 The components within the dashed box; if the computing device in this embodiment is implemented as a conventional server, cloud server, or server array, etc., then it may not include... Figure 12 The component within the dashed box.

[0197] The processing component described above includes one or more processors to execute computer instructions to complete all or part of the steps in the method described above. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the method described above.

[0198] The aforementioned storage components can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0199] The aforementioned communication component is configured to facilitate wired or wireless communication between the device housing the communication component and other devices. The device housing the communication component can access wireless networks based on communication standards, such as mobile communication networks, or combinations thereof. In one exemplary embodiment, the communication component receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel.

[0200] The aforementioned display components may include a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touchscreen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors can sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation.

[0201] The aforementioned power supply components provide power to various components within the device in which they reside. These power supply components may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power to the device in which they reside.

[0202] The aforementioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC) configured to receive external audio signals when the device containing the audio component is in an operating mode, such as call mode, recording mode, or voice recognition mode. The received audio signals can be further stored in memory or transmitted via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.

[0203] Accordingly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to implement the steps in the above-described method embodiments. The computer-readable storage medium includes volatile or non-volatile components, or a combination thereof, and can be removable or non-removable. Examples of computer-readable storage media include, but are not limited to, phase-change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random-access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), flash memory or other memory technologies, CD-ROM, Digital Video Disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium. Accordingly, this application also provides a computer program product, which includes a computer program or instructions that, when executed by a processor, cause the processor to implement the steps in the above method embodiments. It should be understood that each step or combination of steps in the above method flow can be implemented by the computer program or instructions. Furthermore, these computer programs or instructions can be applied to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device, enabling the processor of the general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to function as an apparatus for implementing the corresponding functions in the above method embodiments.

[0204] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0205] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0206] Finally, it should be noted that the above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. An information prediction method, characterized in that, include: In response to a query request, determine the query parameters and the target data flow record to which the query parameters are matched; The system queries at least one first sample data flow record that matches the query parameters from the flow knowledge base; wherein the flow knowledge base is constructed based on multiple historical data flow records. If the query request meets the calculation requirements, the predicted flow information corresponding to the target data flow record is determined according to the flow information corresponding to the at least one first sample data flow record, in accordance with the prediction rules; otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data transferor hit by the query parameters; the predicted flow information is used as the query result of the query request.

2. The method according to claim 1, characterized in that, The at least one first sample data flow record includes multiple first sample data flow records; The step of determining the predicted flow information corresponding to the target data flow record according to the prediction rule based on the flow information corresponding to the at least one first sample data flow record includes: Determine the data flow time corresponding to each of the multiple first sample data flow records to obtain multiple data flow times; The time consumed by the multiple data streams is statistically analyzed to determine either the first data stream time that is the median or the second data stream time that exceeds a predetermined proportion. Based on the data transfer-out time in the target data transfer record, and the first data transfer time and / or the second data transfer time, the predicted transfer-in time is calculated.

3. The method according to claim 2, characterized in that, The step of determining the predicted flow information corresponding to the target data flow record according to the prediction rule based on the flow information corresponding to the at least one first sample data flow record includes: Based on the flow information corresponding to each of the at least one first sample data flow records, the candidate transfer time corresponding to the target data flow record is determined according to the prediction rule. Identify at least one influencing factor that the target data transfer record hits, and update the candidate transfer-in time according to the change conditions corresponding to the at least one influencing factor to obtain the predicted transfer-in time; the at least one influencing factor includes one or more of the following: data transfer-out time is a holiday time, data transfer-in time is a holiday time, the transferred data meets predetermined conditions, the data transfer-out region is a target type region, and the data transfer-in region is the target type region.

4. The method according to claim 1, characterized in that, When the query request meets the calculation requirements, determining the predicted flow information corresponding to the target data flow record according to the prediction rule based on the flow information corresponding to each of the at least one first sample data flow record includes: If a predetermined number of first sample data flow records matching the query parameters exist in the flow knowledge base, the predicted flow information corresponding to the target data flow record is determined according to the flow information corresponding to each of the at least one first sample data flow record and the prediction rule.

5. The method according to claim 4, characterized in that, The query parameters include at least one query factor; The step of retrieving at least one first sample data flow record from the flow knowledge base that matches the query parameters includes: Using the predetermined quantity as a constraint, and according to the combination requirements, one or more query factors are sequentially used as query conditions to search for at least one first sample data flow record that matches the query conditions in the flow knowledge base, until the predetermined number of first sample data flow records is obtained or the query conditions are empty.

6. The method according to claim 5, characterized in that, The step of using the first prediction model to predict the predicted flow information corresponding to the target data flow record based on the historical behavior data of the data transferor matched by the query parameters includes: From the flow knowledge base, determine at least one second sample data flow record corresponding to the data sender matched by the query parameters; determine at least one influencing factor matched by the target data flow record; Using the first prediction model, based on the query parameters, the at least one second sample data flow record, and the change conditions corresponding to different influencing factors, the predicted flow information corresponding to the target data flow record is predicted.

7. The method according to claim 1, characterized in that, Also includes: Identify at least one path node in the target data flow record; Based on the at least one path node, predict the target flow path corresponding to the target data flow record; The step of determining the predicted flow information corresponding to the target data flow record according to the prediction rule based on the flow information corresponding to the at least one first sample data flow record includes: From at least one first sample data flow record, determine at least one target sample data flow record that the target flow path hits; Based on the at least one target sample data flow record, determine the predicted transfer time of the target data flow record corresponding to the target flow path according to the prediction rule; The step of using the first prediction model to predict the predicted flow information corresponding to the target data flow record based on the historical behavior data of the data transferor matched by the query parameters includes: Using a first prediction model, at least one historical data transfer record corresponding to the target transfer path is determined from the historical behavior data of the data transferor matched by the query parameters. Based on the at least one historical data transfer record, the predicted transfer time of the target data transfer record corresponding to the target transfer path is predicted.

8. The method according to claim 1, characterized in that, Also includes: Statistical analysis is performed on the multiple historical data flow records to identify at least one data system whose data flow loss meets the loss conditions; Based on the at least one data system, data loss warning information is output to the data receiving party and / or the data sending party.

9. The method according to claim 1, characterized in that, The step of retrieving at least one sample data flow record from the flow knowledge base that matches the query parameters includes: The intelligent processing module is used to search the knowledge base for multiple first sample data flow records that match the query parameters; If the query request meets the calculation requirements, the predicted flow information corresponding to the target data flow record is determined according to the prediction rules based on the flow information corresponding to each of the at least one first sample data flow record; otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data transferor hit by the query parameters, including: Using the intelligent processing module, if the query request meets the calculation requirements, the predicted flow information corresponding to the target data flow record is determined according to the flow information corresponding to the at least one first sample data flow record and the prediction rule; otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data transferor hit by the query parameters.

10. The method according to claim 1, characterized in that, Also includes: Retrieve multiple historical data flow records; Based on the aforementioned multiple historical data flow records, multiple sample data flow records are generated; A knowledge base for data transfer is constructed based on the aforementioned multiple sample data transfer records.

11. The method according to claim 10, characterized in that, The construction of the flow knowledge base based on the multiple sample data flow records includes: The multiple sample data flow records are grouped according to the different field values ​​of the predetermined record fields; The multiple sample data transfer records are stored in the transfer knowledge base according to their respective groups; The step of retrieving at least one sample data flow record from the flow knowledge base that matches the query parameters includes: The target group is determined based on the parameter values ​​corresponding to the predetermined record fields in the query parameters; From the data records corresponding to the target group in the knowledge base, find at least one sample data flow record that matches the query parameters.

12. The method according to claim 1, characterized in that, Also includes: In response to a query request, determine the data transfer credentials provided by the data recipient; Identify the query parameters in the data transfer voucher; If at least one missing factor exists in the query parameters, the at least one missing factor is identified from the historical behavior data corresponding to the data transfer recipient. Update the query parameters based on the at least one missing factor.

13. An information retrieval method, characterized in that, include: Get the query request; The query request includes query parameters; Query whether there is target flow information corresponding to the target data flow record that matches the query parameters; If so, output the target flow information; If not, query the knowledge base for multiple first sample data flow records that match the query parameters; the knowledge base is constructed based on multiple historical data flow records. If the query request meets the calculation requirements, the predicted flow information corresponding to the target data flow record is calculated according to the prediction rules based on the flow information corresponding to the multiple first sample data flow records respectively; otherwise, the predicted flow information corresponding to the target data flow record is predicted using the first prediction model based on the historical behavior data of the data transferor hit by the query parameters. Output the predicted flow information.

14. An information prediction method, characterized in that, include: In response to the query request, determine the query parameters and the target remittance record matched by the query parameters; Search the remittance knowledge base for at least one first sample remittance record that matches the query parameters; the remittance knowledge base is built and generated based on multiple historical remittance records; If the query request meets the calculation requirements, based on the remittance time data corresponding to the at least one first sample remittance record, the predicted arrival time corresponding to the target remittance record is determined according to the prediction rules; otherwise, the first prediction model is used to predict the predicted arrival time corresponding to the target remittance record based on the historical behavior data of the remitter matched by the query parameters; the predicted arrival time is used as the query result of the query request.

15. An information retrieval method, characterized in that, include: Get the query request; The query request includes query parameters; Check if there is a corresponding arrival time for a target remittance record that matches the query parameters; If so, output the arrival time. If not, search the remittance knowledge base for at least one first sample remittance record that matches the query parameters; the remittance knowledge base is built and generated based on multiple historical remittance records. If the query request meets the calculation requirements, based on the remittance time data corresponding to the at least one first sample remittance record, the predicted arrival time corresponding to the target remittance record is determined according to the prediction rules; otherwise, the first prediction model is used to predict the predicted arrival time corresponding to the target remittance record based on the historical behavior data of the remitter matched by the query parameters. Output the predicted arrival time.

16. A computing device, characterized in that, This includes processing components and storage components; The storage component stores a computer program; the computer program is invoked and executed by the processing component to implement the information prediction method as described in any one of claims 1 to 12, or to implement the information query method as described in claim 13, or to implement the information prediction method as described in claim 14, or to implement the information query method as described in claim 15.

17. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by a processing component, implements the information prediction method as described in any one of claims 1 to 12, or the information query method as described in claim 13, or the information prediction method as described in claim 14, or the information query method as described in claim 15.

18. A computer program product, characterized in that, It includes a computer program or instructions that, when executed by a processing component, implement the information prediction method as described in any one of claims 1 to 12, or implement the information query method as described in claim 13, or implement the information prediction method as described in claim 14, or implement the information query method as described in claim 15.

Images