Business state identification method and device and electronic equipment

Abstract

The invention discloses a service state identification method and device and electronic equipment. Relates to the big data field. The method comprises the steps that a trusteeship fund allocation instruction sequence sent by a user is received, a classification variable sequence of the trusteeship fund allocation instruction sequence is generated, and the trusteeship fund allocation instruction sequence comprises M trusteeship fund allocation instructions; target parameters in the probability prediction model are obtained, and the target parameters comprise an initial probability vector representing an initial state probability, a state transition probability matrix representing a state transition probability and an observation probability matrix representing an observation probability; and calculating the target parameter and the classification variable sequence through a Viterbi algorithm to obtain a service state sequence corresponding to the trusteeship fund allocation instruction sequence. According to the method and the device, the problem of relatively low accuracy and efficiency of manually determining the operation state of the business indicated by the instruction in the related technology is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of big data, in particular, to a business state identification method and device and electronic equipment. BACKGROUND

[0002] In the case of handling annuity fund custody business by a custody institution, fund transfer instructions sent by a trust account holder will be received, and accurate processing of the fund transfer instructions is a key link to ensure smooth operation and compliance of the fund.

[0003] When processing the fund transfer instructions, it is necessary to first determine the business state of the instruction indicated business. The instructions sent by the trust account holder usually indicate two main operation states of withdrawal and injection, and in the case of these two states, the custody institution usually identifies them with different codes or types, facilitating clear communication and instruction execution between the custody institution and the fund manager.

[0004] However, in addition to the above two basic states, there are many other fund transfer scenarios, such as interest payment, management fee deduction, investment allocation adjustment, etc., but currently these scenarios are often classified as one operation state in the existing management system, and the specific operation state is determined by manual judgment, and then the corresponding instruction processing is performed, but this process not only consumes time, but also introduces errors due to differences in manual judgment, reducing the overall efficiency and accuracy of the custody business.

[0005] In view of the problem that the accuracy and efficiency of manually determining the operation state of the business indicated by the instruction in the related art are low, no effective solution has been proposed so far. SUMMARY

[0006] The main purpose of the present application is to provide a business state identification method and device and electronic equipment to solve the problem of low accuracy and efficiency of manually determining the operation state of the business indicated by the instruction in the related art.

[0007] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a business state identification method is provided. The method comprises: receiving a sequence of custody fund transfer instructions sent by a user, and generating a sequence of classification variables of the sequence of custody fund transfer instructions, wherein the sequence of custody fund transfer instructions comprises M custody fund transfer instructions, and M is a positive integer; obtaining target parameters in a probability prediction model, wherein the target parameters include: an initial probability vector representing an initial state probability, a state transition probability matrix representing a state transition probability, and an observation probability matrix representing an observation probability; calculating the target parameters and the sequence of classification variables through a Viterbi algorithm to obtain a sequence of business states corresponding to the sequence of custody fund transfer instructions, wherein the sequence of business states is used to indicate the business state of each custody fund transfer instruction.

[0008] Optionally, before obtaining the target parameter in the probability prediction model, the method further comprises: obtaining a plurality of historical escrow fund transfer instruction sequences, and obtaining a historical business state sequence corresponding to each historical escrow fund transfer instruction sequence; determining each historical escrow fund transfer instruction sequence and the corresponding historical business state sequence as a set of sample data, to obtain a plurality of sets of sample data; and training the probability prediction model according to the plurality of sets of sample data by using an expectation maximization algorithm, to obtain the target parameter in the probability prediction model.

[0009] Optionally, the training of the probability prediction model according to the plurality of sets of sample data by using the expectation maximization algorithm to obtain the target parameter in the probability prediction model comprises: randomly generating the target parameter to obtain a random probability parameter; for each set of sample data, estimating a probability distribution of a hidden state sequence by using the random probability parameter; adjusting the random probability parameter according to the probability distribution until the log-likelihood of the probability prediction model under the random probability parameter is maximum, and determining the random probability parameter at the current time as the target parameter.

[0010] Optionally, the calculation of the target parameter and the classification variable sequence by using the Viterbi algorithm to obtain the business state sequence corresponding to the escrow fund transfer instruction sequence comprises: calculating an initial probability vector of each business state corresponding to a first feature element of the classification variable sequence by using an initial probability vector in the target parameter; for each subsequent feature element in the classification variable sequence, calculating a probability of each business state when reaching the next feature element based on a current target path, a transition probability and an observation probability, and adding the business state with the maximum probability to the target path to obtain an updated target path, wherein the target path is composed of the business state with the maximum probability; and in the case that the last feature element in the classification variable sequence is processed, determining the business state contained in the target path as the business state sequence.

[0011] Optionally, the generation of the classification variable sequence of the escrow fund transfer instruction sequence comprises: sequentially identifying instruction contents of each escrow fund transfer instruction in the escrow fund transfer instruction sequence to obtain M instruction contents; sequentially obtaining classification variables from each instruction content to obtain M classification variable sets; and arranging the M classification variable sets according to the order of the instruction contents to obtain the classification variable sequence.

[0012] Optionally, the generation of the classification variable sequence of the escrow fund transfer instruction sequence comprises: for each instruction content, in the case that the classification variable is obtained from each instruction content, judging whether the type and the number of the classification variable meet a preset requirement; in the case that the preset requirement is met, determining that the classification variable is correct; in the case that the preset requirement is not met, determining that the classification variable is incorrect, and determining missing data in the classification variable according to the preset requirement, and sending a prompt information to a user, wherein the prompt information is used to prompt the user to add the missing data.

[0013] Optionally, the classification variables are sequentially obtained from the instruction contents, and obtaining the M classification variable sets comprises: obtaining keywords indicated in the preset requirements to obtain a plurality of keywords; identifying instruction information corresponding to each keyword from the instruction contents to obtain a plurality of instruction information; sequentially obtaining instruction data corresponding to each instruction information, and determining a set of instruction information and instruction data as a classification variable to obtain M classification variables.

[0014] In order to achieve the above object, according to another aspect of the present application, a business state identification device is provided. The device comprises: a generation unit configured to receive a sequence of escrow fund transfer instructions sent by a user and generate a sequence of classification variables of the sequence of escrow fund transfer instructions, wherein the sequence of escrow fund transfer instructions comprises M escrow fund transfer instructions, and M is a positive integer; a first acquisition unit configured to acquire target parameters in a probability prediction model, wherein the target parameters comprise an initial probability vector representing an initial state probability, a state transition probability matrix representing a state transition probability, and an observation probability matrix representing an observation probability; and a calculation unit configured to calculate the target parameters and the sequence of classification variables by using a Viterbi algorithm to obtain a sequence of business states corresponding to the sequence of escrow fund transfer instructions.

[0015] In order to achieve the above object, according to another aspect of the present application, an electronic device is provided. The electronic device comprises a memory storing an executable program; and a processor configured to run the program, wherein the program, when running, performs the business state identification method described above.

[0016] In order to achieve the above object, according to another aspect of the present application, a computer program product is provided. The computer program product comprises computer instructions, and the computer instructions, when executed by a processor, implement the steps of the business state identification method described above.

[0017] In this embodiment, a sequence of escrow fund transfer instructions sent by a user is received, and a categorical variable sequence of the escrow fund transfer instruction sequence is generated. The escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer. Target parameters in the probability prediction model are obtained, including an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability. The target parameters and the categorical variable sequence are calculated using the Viterbi algorithm to obtain a business state sequence corresponding to the escrow fund transfer instruction sequence. This business state sequence indicates the business state of each escrow fund transfer instruction. By using the target parameters configured in the probability prediction model and the order of the escrow fund transfer instructions in the sequence, the business state of each escrow fund transfer instruction is calculated. This achieves the goal of eliminating the need for manual determination of the business state of the escrow fund transfer instructions, thus realizing the technical effect of accurately determining the business state of each escrow fund transfer instruction. This solves the problem of low accuracy and efficiency in related technologies where the operational state of the business indicated by the instructions is determined manually. Attached Figure Description

[0018] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0019] Figure 1 A hardware structure block diagram of a computer terminal for implementing a method for identifying business status is shown.

[0020] Figure 2 This is a flowchart of a service status identification method according to Embodiment 1 of this application;

[0021] Figure 3 This is a schematic diagram of a service status identification device according to Embodiment 2 of this application;

[0022] Figure 4 This is a structural block diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0023] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0026] It should be noted that the business status identification method, device, and electronic equipment defined in this disclosure can be used in the big data field, or in any field other than the big data field. The application fields of the business status identification method, device, and electronic equipment defined in this disclosure are not limited.

[0027] It should be noted that all information, 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, and displayed data) used in this application are information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data all comply with the relevant laws, regulations, and standards of the relevant regions, have taken necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entry points for users to choose to authorize or refuse use. If the user chooses to refuse, the process proceeds to the expert decision-making process. For example, this system has interfaces with relevant users or institutions. Before obtaining relevant information, a request to obtain the information needs to be sent to the aforementioned user or institution through the interface, and the relevant information is obtained only after receiving consent from the aforementioned user or institution.

[0028] The embodiments or examples disclosed herein are not exhaustive, but merely illustrative of some embodiments or examples, and are not intended to limit the scope of protection of this disclosure. Unless otherwise specified, each step in a particular embodiment or example can be implemented as an independent embodiment, and the steps can be arbitrarily combined. For example, a solution after removing some steps in a particular embodiment or example can also be implemented as an independent embodiment, and the order of the steps in a particular embodiment or example can be arbitrarily interchanged. Furthermore, optional methods or examples in a particular embodiment or example can be arbitrarily combined; moreover, embodiments or examples can be arbitrarily combined. For example, some or all steps of different embodiments or examples can be arbitrarily combined, and a particular embodiment or example can be arbitrarily combined with optional methods or examples of other embodiments or examples.

[0029] For ease of description, the following explains some of the nouns or terms used in the embodiments of this application:

[0030] Hidden Markov Models (HMMs) are statistical models that introduce the concept of "hidden states" into the traditional Markov model. In a traditional Markov model, the states are visible, while in an HMM, the true states of the system are not visible; only the outputs or observations related to these states can be observed.

[0031] Viterbi algorithm: a dynamic programming algorithm used to find the most likely sequence of hidden states given an observation sequence.

[0032] Custodian fund transfer instructions: In fund custody business, these are orders issued by the fund manager or authorized party to the custodian institution, instructing how funds should be transferred from one account to another for the purpose of managing fund assets. These instructions typically include detailed information about the fund transfer, such as the amount transferred, the two accounts involved (source and target accounts), the date and time of the transfer, and the purpose or reason for the transfer. The custodian institution executes the corresponding fund transfer operations according to these instructions to ensure that the fund's cash flow complies with the fund's operating strategy and legal requirements.

[0033] Example 1

[0034] According to an embodiment of this application, an embodiment of a method for identifying business status is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0035] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device.Figure 1 A hardware block diagram of a computer terminal for implementing a method for identifying business status is shown. Figure 1 As shown, the computer terminal 10 (or mobile device) may include one or more processors 102 (shown as 102a, 102b, ..., 102n in the figure) 102 (processor 102 may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0036] It should be noted that the aforementioned one or more processors 102 and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10 (or mobile device). As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0037] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the business status identification method in this embodiment. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned business status identification method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0038] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the communication provider of the computer terminal 10. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module, used for wireless communication with the Internet.

[0039] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10 (or mobile device).

[0040] Under the aforementioned operating environment, this application provides the following: Figure 2 The method for identifying the business status is shown. Figure 2 This is a flowchart of the service status identification method provided in Embodiment 1 of this application, as follows: Figure 2 As shown, the method includes:

[0041] Step S101: Receive the sequence of escrow fund transfer instructions sent by the user, and generate a categorical variable sequence of the escrow fund transfer instruction sequence, wherein the escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer.

[0042] It should be noted that the sequence of custodial fund transfer instructions sent by the user refers to the set of instructions sent by the fund manager or other authorized users through the custodial system interface to instruct the custodian institution to perform fund transfer operations. This set of instructions may include multiple instructions, each corresponding to a custodial fund transfer instruction.

[0043] Specifically, after receiving the sequence of escrow fund transfer instructions sent by the user, each escrow fund transfer instruction can be determined sequentially within the sequence, and the categorical variables of each instruction can be obtained, thus generating a categorical variable sequence. These categorical variables can be key observation variables describing the operational status of the business indicated by the escrow fund transfer instructions. For example, key observation variables for each instruction might include: receiving account number (long, short), amount (large, small), and summary (present, absent). For example, the key observation variables for escrow fund transfer instruction A might be (long, large, present), and for instruction B, (long, large, absent). This process yields the key observation variables for each instruction. Based on the categorical variables obtained from the combination of these key observation variables, a categorical variable sequence for the escrow fund transfer instruction sequence is generated. Finally, the operational status of each escrow fund transfer instruction is determined in subsequent operations based on this categorical variable sequence.

[0044] Step S102: Obtain the target parameters in the probability prediction model, wherein the target parameters include: an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability.

[0045] It should be noted that the probabilistic prediction model can be a Hidden Markov Model (HMM) used to predict states in sequential data. The target parameters include the initial probability vector, the state transition probability matrix, and the observation probability matrix; these parameters define the model's behavioral characteristics. The initial probability vector represents the prior probabilities of each hidden state at the start of the model. The state transition probability matrix describes the probability of transitioning from one hidden state to another. The observation probability matrix defines the probability of observing a specific type of instruction given a hidden state. Here, the hidden state is the operational state of the business.

[0046] Specifically, when predicting the status of business operations, it is first necessary to use a probabilistic prediction model that has been trained and obtain the target parameters in the probabilistic prediction model, and then use the target parameters to execute the subsequent operation status prediction process.

[0047] Step S103: The target parameters and categorical variable sequences are calculated using the Viterbi algorithm to obtain the business status sequence corresponding to the escrow fund transfer instruction sequence. The business status sequence is used to indicate the business status of each escrow fund transfer instruction.

[0048] It should be noted that the business status sequence is calculated using the Viterbi algorithm and indicates the specific business status corresponding to each instruction in the escrow fund transfer instruction sequence.

[0049] Specifically, given the target parameters and the sequence of categorical variables, the target parameters and the sequence of categorical variables can be substituted into the Viterbi algorithm to determine the globally optimal path. Then, based on the probability vectors in the target parameters, the business operation state most likely to correspond to each categorical variable in the sequence of categorical variables can be determined, thereby obtaining the business operation state of each categorical variable. Finally, the business operation states are combined into a business state sequence according to the order of instructions.

[0050] For example, given the target probability λ = (π, A, B), the optimal path for matching variables is calculated using the Viterbi algorithm. Assuming the categorical variable sequence is {(long, large, yes), (long, large, no), (long, small, yes), (long, small, no), (short, large, yes), (short, large, no), (short, small, yes), (short, small, no), (long, large, yes), (long, large, no)}, the most likely business state sequence obtained using the target probability λ = (π, A, B) can be: {interest injection into collection accounts, payment of performance fees, payment of audit fees, interest injection into entrusted accounts, interest withdrawal from entrusted accounts, payment of performance fees, payment of performance fees, payment of audit fees, interest injection into entrusted accounts, payment of performance fees}.

[0051] The specific method is as follows:

[0052] Algorithm input: Target parameter λ = (π, A, B), categorical variable sequence X = (x1, x2, ..., x...). T ), where x1 to xT each correspond to a categorical variable. Algorithm output: Best match (Z=(Z1,Z2,...,ZT)). 10 )).

[0053] Step 1: Initialization. At t=1, for the ten states i, calculate the probabilities of (long, big, have) for the five outcomes (interest injection into the collection account, payment of performance fee, payment of audit fee, interest injection into the trustee account, and interest withdrawal from the trustee account) using the initial probability λ=(π,A,B).

[0054] Step 2: Recursively calculate the most likely outcome for t = 2, 3, ... 10 from (long, large, yes) to (long, large, no). That is, calculate the probability that (long, large, yes) is "interest injection for the collection account" to (long, large, no) is "interest injection for the collection account". Then calculate the probability that (long, large, yes) is "interest injection for the collection account" to (long, large, no) is "payment of performance fee". Continue this process to find the maximum probability and the path.

[0055] Step 3: Find the most likely outcome from (long, large, none) to (long, small, present), and continue to find the most likely outcome when the last observation result is (long, large, none), that is, find the probability of the most likely outcome sequence.

[0056] Step 4: Determine the path Ψ when the optimal probability of the global sequence is obtained, that is, {interest injection for the aggregate account, payment of performance fee, payment of audit fee, interest injection for the entrusted account, interest withdrawal for the entrusted account, payment of performance fee, payment of performance fee, payment of audit fee, interest injection for the entrusted account, payment of performance fee}.

[0057] The business status identification method provided in this application embodiment receives a sequence of escrow fund transfer instructions sent by a user and generates a categorical variable sequence of the escrow fund transfer instruction sequence, wherein the escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer; obtains target parameters in a probability prediction model, wherein the target parameters include: an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability; calculates the business status sequence corresponding to the escrow fund transfer instruction sequence using the Viterbi algorithm, wherein the business status sequence is used to indicate the business status of each escrow fund transfer instruction. By using the target parameters configured in the probability prediction model and the order of each escrow fund transfer instruction in the escrow fund transfer instruction sequence, the business status to which each escrow fund transfer instruction belongs is calculated, achieving the goal of eliminating the need for manual determination of the business status of the escrow fund transfer instructions, thereby achieving the technical effect of accurately determining the business status of each escrow fund transfer instruction, and solving the problem of low accuracy and efficiency in related technologies where the operational status of the business indicated by the instructions is determined manually.

[0058] Optionally, in the business status identification method provided in this application embodiment, before obtaining the target parameters in the probability prediction model, the method further includes: obtaining multiple historical escrow fund transfer instruction sequences and obtaining the historical business status sequence corresponding to each historical escrow fund transfer instruction sequence; determining each historical escrow fund transfer instruction sequence and the corresponding historical business status sequence as a set of sample data to obtain multiple sets of sample data; and training the probability prediction model based on the multiple sets of sample data using the expectation-maximization algorithm to obtain the target parameters in the probability prediction model.

[0059] It should be noted that the historical escrow fund transfer instruction sequence refers to a sequence of fund transfer instructions that occurred in the past and are recorded in the escrow system. These sequences contain instances of historical fund transfer instructions. The historical business status sequence corresponds to the historical escrow fund transfer instruction sequence, recording the business status at the time of actual execution of each transfer instruction in the past. The Expectation Maximization Algorithm (EM algorithm) is an iterative algorithm used to find parameter estimates for a given dataset in a probabilistic model. In this embodiment, the EM algorithm is used to train and adjust the parameters of the probabilistic prediction model based on historical sample data to improve the model's prediction accuracy.

[0060] Specifically, the system first reads multiple historical custodian fund transfer instruction sequences from the database. Each sequence represents a sequence of fund transfer instructions issued by the fund manager to the custodian institution within a past period. Simultaneously, the system also acquires historical business status sequences corresponding to these instruction sequences, recording the specific business operation status of each instruction during its actual execution. This process collects sufficient historical data to provide a foundation for model training, ensuring that the model can learn effective parameter estimates from real-world custodian business scenarios.

[0061] It should be noted that the system assembles each pair of historical escrow fund transfer instruction sequences and corresponding historical business operation status sequences into a sample dataset. Then, all these datasets are combined to form a large dataset containing multiple sets of sample data. Each sample dataset contains a sequence of fund transfer instructions and a sequence of business statuses during the actual execution of those instructions, forming an input-output pair for model training.

[0062] It should be noted that, given sample data, the system can use the EM algorithm to train the probabilistic prediction model. The EM algorithm is an iterative algorithm that progressively optimizes model parameters by alternating between "expectation" (E-step) and "maximization" (M-step) steps to improve the model's accuracy in predicting unknown data. Specifically, the algorithm continuously adjusts the model's initial probability vector, state transition probability matrix, and observation probability matrix based on fund transfer instruction sequences and historical business state sequences from multiple sets of sample datasets until it finds a set of parameters that maximizes the log-likelihood function of the sample dataset. Through algorithm training, the target parameters of the probabilistic prediction model are optimized, enabling the model to accurately predict the business state corresponding to each instruction in the future fund transfer instruction sequence, thereby improving the automation level and accuracy of fund transfer instruction processing in the custody business.

[0063] This embodiment trains a high-precision probabilistic prediction model using historical data. This model can automatically identify and classify the business scenarios corresponding to each instruction based on the characteristics of the escrow fund transfer instruction sequence, greatly improving the automation level, efficiency, and accuracy of escrow business fund transfer instruction processing.

[0064] Optionally, in the business state identification method provided in this application embodiment, the target parameters in the probability prediction model are obtained by training the probability prediction model based on multiple sets of sample data using the expectation-maximization algorithm, including: randomly generating target parameters to obtain random probability parameters; estimating the probability distribution of the hidden state sequence using the random probability parameters for each set of sample data; adjusting the random probability parameters according to the probability distribution until the log-likelihood of the probability prediction model under the random probability parameters is maximized, and determining the random probability parameters at the current time as the target parameters.

[0065] It should be noted that the random probability parameter can be a random initial value assigned to the model parameters at the beginning of algorithm training, serving as the starting point for iterative optimization of the EM algorithm. The hidden state sequence can be the sequence of actual business states that the model attempts to estimate behind each sequence of escrow fund transfer instructions, i.e., the "hidden" true state.

[0066] Specifically, before training begins, the system needs to set initial values ​​for the initial probability vector, state transition probability matrix, and observation probability matrix of the probabilistic prediction model. These values ​​are typically generated randomly to provide a starting point for the EM algorithm. For example, for a model with three business states, the initial probability vector can be randomly initialized to [0.3, 0.3, 0.4], and the elements of the state transition probability matrix and observation probability matrix can also be randomly initialized to a reasonable range of probability values.

[0067] Optionally, the system needs to estimate the probability distribution of a hidden state sequence for each set of historical sample data based on the current random probability parameters. This involves calculating the forward and backward probabilities of the model to determine the probability of each hidden state sequence under a given observation sequence (i.e., the historical escrow fund transfer instruction sequence).

[0068] In the "Expectation Step" (E-step) of the EM algorithm, the system calculates the log-likelihood of the current model parameters based on the obtained probability distribution. Subsequently, in the "Maximization Step" (M-step), the system adjusts the model parameters, including the initial probability vector, state transition probability matrix, and observation probability matrix, according to the principle of maximizing the log-likelihood. This process is repeated until the change in log-likelihood before and after parameter adjustment is less than a predetermined threshold, or the maximum number of iterations is reached. The model parameters at this point are the target parameters.

[0069] The following is an optional procedure for determining the target parameters:

[0070] Step 1: Initialize model parameters π, A, and B.

[0071] Implementation process: Select a random number generator to generate random probability values ​​for the initial state probability vector π, the state transition probability matrix A, and the observation probability matrix B. Ensure that the sum of all elements in π is 1, the sum of all elements in each row or column of A and B is also 1, and all probability values ​​are non-negative.

[0072] For example: For states with N=5 (interest injection into collection accounts, payment of performance fees, payment of audit fees, interest injection into trustees, interest withdrawal from trustees) and observations with M=8 (long, large, present), (long, large, absent), (long, small, present), (long, small, absent), (short, large, present), (short, large, absent), (short, small, present), (short, small, absent), initialize π, A, and B. For example, π can be initialized to have an equal probability of occurrence for each state, i.e., π is [0.2, 0.2, 0.2, 0.2, 0.2]; each element in A and B can be initialized to a random number that meets the probability distribution requirements, such as a row in A being initialized to [0.2, 0.1, 0.3, 0.2, 0.2], ensuring that the row sum is 1.

[0073] Step 2: Perform step E (expected step).

[0074] Implementation process: Using the current π, A, and B parameters, calculate the posterior probability for each set of historical data (historical escrow fund transfer instruction sequence and corresponding historical business type sequence), that is, the probability of being in each state at each moment under the parameters.

[0075] For example, for a specific instruction sequence, such as {(long, large, yes), (long, large, no)}, using the current π, A, and B parameters, calculate the probability of being in each business type state at each time step. This step may require using a forward or backward algorithm to calculate the posterior probability of each state at each time step.

[0076] Step 3: Perform the M-step (maximum step).

[0077] Implementation process: Based on the posterior probabilities calculated in step 2, re-estimate π, A, and B to maximize the log-likelihood of the model based on the new parameters.

[0078] For example, based on the posterior probability calculated in the previous step, the elements in π, A, and B are adjusted to maximize the log-likelihood of the observed sequence under the new parameters. For example, if it is found in step 2 that the probability of "interest injection for the aggregator" at the first moment is relatively high, then in step M, the value of "interest injection for the aggregator" in π will be adjusted so that its probability is higher in the updated π.

[0079] Step 4: Evaluate whether the log-likelihood function of the model meets the convergence criteria.

[0080] Implementation process: Calculate the log-likelihood function value under the current parameters, compare it with the log-likelihood function value of the previous iteration, and determine whether the convergence criterion has been met, that is, the increase in the log-likelihood function is less than the set threshold.

[0081] For example: Suppose the log-likelihood function value under the current parameters is L1, and the log-likelihood function value of the previous iteration is L0. If L1-L0<ε (ε is the convergence threshold, which can be set to 0.001 or smaller), then the model parameters are considered to have converged, and the iteration stops.

[0082] Step 5: Once the model parameters are adjusted to convergence, output the final π, A, and B as preset probability parameters.

[0083] Implementation process: When the evaluation results show that the model parameters no longer change or change only slightly, stop the EM algorithm iteration and output the current values ​​of π, A and B parameters.

[0084] For example, suppose that after multiple iterations, the values ​​of parameters π, A, and B are π_final, A_final, and B_final, respectively. These values ​​are the parameters of the probability prediction model after training, which are also the target parameters.

[0085] This embodiment uses the EM algorithm training process to automatically optimize the parameters of the probabilistic prediction model, enabling the model to more accurately identify and classify the business operation status corresponding to the escrow fund transfer instructions, thereby improving the model's adaptability and prediction accuracy.

[0086] Optionally, in the business status identification method provided in this application embodiment, calculating the target parameters and the categorical variable sequence using the Viterbi algorithm to obtain the business status sequence corresponding to the escrow fund transfer instruction sequence includes: calculating the initial probability vector of each business status corresponding to the first feature element of the categorical variable sequence using the initial probability vector in the target parameters; for each subsequent feature element in the categorical variable sequence, calculating the probability of each business status when reaching the next feature element based on the current target path, transition probability, and observation probability, and adding the business status with the highest probability to the target path to obtain the updated target path, wherein the target path is composed of the business status with the highest probability; after processing the last feature element in the categorical variable sequence, determining the business status contained in the target path as the business status sequence.

[0087] It should be noted that the initial probability vector is the probability distribution of each business state in the model as the starting state, reflecting the model's prior probability estimate of the business states when there is no observation information. The transition probability is the probability of transitioning from one business state to another in the probabilistic prediction model, used to describe the dynamic changes between states. The observation probability is the probability of observing a specific type of instruction given a business state, used to measure the degree of matching between the model output and the actual observation. The target path, in the Viterbi algorithm, is the series of business states with the highest probability given a sequence of observations; this series of states is the "target path".

[0088] Specifically, when determining the business state sequence, the algorithm can calculate the initial probability of each business state corresponding to the first instruction (the first feature element of the categorical variable sequence) in the escrow fund transfer instruction sequence based on the initial probability vector already trained in the probabilistic prediction model. Then, the algorithm begins to process subsequent instructions in the categorical variable sequence. For each instruction, based on the currently calculated "target path" (i.e., the business state sequence with the highest probability so far), the model's transition probability matrix, and the observation probability matrix, the algorithm calculates the probability of each business state when the instruction is received, and retains the path with the highest probability. That is, the newly calculated most probable business state is added to the target path to form an updated path, and thus gradually constructs the most probable business state sequence. By considering the transition probabilities between states and the probability of observing the instruction, the algorithm can dynamically adjust the path to match the observed data to the greatest extent.

[0089] When the last instruction in the categorical variable sequence (i.e., the sequence of instructions for escrow fund transfer) is processed, the algorithm stops iterating and determines all business states contained in the target path as the final business state sequence, thus obtaining the business state sequence.

[0090] This embodiment uses the Viterbi algorithm to enable the system to identify the business state corresponding to each instruction in the escrow fund transfer instruction sequence based on the probability prediction model, and construct the most likely business state sequence. This improves the automation level of fund transfer instruction processing in escrow business, reduces the need for manual intervention, and increases the objectivity and accuracy of decision-making by considering the transition probability between states and the observation probability of instructions, thereby improving the efficiency of escrow services and customer satisfaction.

[0091] The following is an example of an overall process for calculating a business state sequence. The specific process of this example is as follows:

[0092] In pension fund custody operations, upon receiving an instruction sequence, observational data (categorical variables) for each instruction can be obtained. These data can include: the length of the receiving account number (long / short), the amount transferred (large / small), and the presence or absence of a transfer summary (yes / no). Hidden states include various business instruction types, such as interest injection into the collection account, payment of performance fees, payment of audit fees, interest injection into the trustee account, and interest withdrawal from the trustee account. By obtaining the categorical variables for each instruction, a sequence of categorical variables corresponding to the instruction sequence can be derived, thereby determining the operational state of each instruction based on these categorical variables.

[0093] Having trained the HMM model using the EM algorithm and obtained the model parameters π (initial state probability vector), A (state transition probability matrix), and B (observation probability matrix), and given the existing categorical variable sequence as: O = {(long, large, present), (long, large, present), (long, small, absent)}, it is necessary to identify the most likely corresponding fund transfer instruction state sequence.

[0094] Example model parameters: π = [0.3, 0.2, 0.1, 0.2, 0.2], which means that the initial probabilities of the model being in the states of interest injection into the collection account, payment of performance fees, payment of audit fees, interest injection into the trustee account, and interest withdrawal from the trustee account are 0.3, 0.2, 0.1, 0.2, 0.2, respectively. A = state transition probability matrix, which can be shown in Table 1, representing the probability of transitioning from one state to another:

[0095] Table 1

[0096]

[0097] B = Observation probability matrix, which can be shown in Table 2 as representing the probability of observing a specific observation value given a state:

[0098] Table 2

[0099]

[0100]

[0101] Furthermore, the Viterbi algorithm can be performed based on the aforementioned matrix and parameters.

[0102] The following is the specific process of the Viterbi algorithm:

[0103] Initialization: For the first observation in the observation sequence, “(long, large, present)”, calculate the “local probability” of being in each hidden state. Using the formulas: Local Probability = π × B; Local Probability_Collective Account Interest Injection = π_Collective Account Interest Injection × B_Collective Account Interest Injection (Long, Large, Available) = 0.3 × 0.8 = 0.24; Local Probability_Payment of Performance Fee = π_Payment of Performance Fee × B_Payment of Performance Fee (Long, Large, Available) = 0.2 × 0.1 = 0.02; Local Probability_Payment of Audit Fee = π_Payment of Audit Fee × B_Payment of Audit Fee (Long, Large, Available) = 0.1 × 0.0 = 0.00; Local Probability_Entrusted Account Interest Injection = π_Entrusted Account Interest Injection × B_Entrusted Account Interest Injection (Long, Large, Available) = 0.2 × 0.0 = 0.00; Local Probability_Entrusted Account Interest Withdrawal = π_Entrusted Account Interest Withdrawal × B_Entrusted Account Interest Withdrawal (Long, Large, Available) = 0.2 × 0.0 = 0.00.

[0104] Recursion: For each subsequent observation in the observation sequence, calculate the "path probability" of that observation point being in each hidden state, and record which previous state has the highest probability. Path probability = max(probability AB of the previous state). The second observation in the observation sequence, "(long, large, present)":

[0105] Calculate the path probability of interest injection for the aggregator, and use the same calculation method to calculate the path probabilities of the other four states, and record the maximum path probability and the corresponding previous state.

[0106] Continue the recursion: For the third observation "(long, small, none)", continue the recursion process, but this time the "local probability" will be the maximum path probability calculated in the previous step multiplied by the observation probability of the corresponding state in the B matrix.

[0107] Backtracking: After all observations have been processed, find the state corresponding to the maximum path probability at the last observation, and backtrack the recorded previous state information until the sequence begins, thereby determining the most likely sequence of fund transfer instruction states.

[0108] In this embodiment, the Viterbi algorithm may ultimately predict the hidden state sequence corresponding to the observed sequence as "interest injection for the aggregator, interest injection for the aggregator, interest withdrawal for the trustee". This means that under the observed sequence "(long, large, present), (long, large, present), (long, small, absent)", the most likely fund transfer instruction state sequence is "interest injection for the aggregator, interest injection for the aggregator, interest withdrawal for the trustee". Furthermore, through the Viterbi algorithm, the most likely fund transfer instruction state sequence behind a given observed sequence can be efficiently identified, thereby helping operators to process and respond quickly, improving the efficiency of business processing.

[0109] Optionally, in the business status identification method provided in this application embodiment, generating the classification variable sequence of the escrow fund transfer instruction sequence includes: sequentially identifying the instruction content of each escrow fund transfer instruction in the escrow fund transfer instruction sequence to obtain M instruction contents; sequentially obtaining classification variables from each instruction content to obtain M classification variable sets; and arranging the M classification variable sets according to the order of the instruction contents to obtain the classification variable sequence.

[0110] It should be noted that the instruction content can refer to the detailed information in each escrow fund transfer instruction, including but not limited to instruction type, amount, account information, operation time, etc.

[0111] Specifically, upon receiving an instruction, it is necessary to accurately extract the categorical variables from the instruction so that the operation status can be determined subsequently based on the accurate categorical variables.

[0112] When the system begins processing the sequence of escrow fund transfer instructions, it first reads each instruction in the sequence and parses its content to extract key information for each instruction, providing a data foundation for subsequent classification processing. Based on this instruction content, the system needs to extract classification variables from each instruction according to a preset content extraction method, thereby obtaining a set of classification variables for each instruction.

[0113] For example, if the instruction includes: the receiving account number, the amount transferred, and the transfer summary, and the preset content extraction method is: to determine the length of the receiving account number (long / short), the size of the amount transferred (large / small), and the presence or absence of the transfer summary (yes / no), then the set of categorical variables corresponding to this instruction can be: long, large, and yes.

[0114] Once the set of categorical variables for each instruction is obtained, the set of categorical variables can be sorted according to the order of the instructions corresponding to the categorical variables in the instruction sequence, thereby obtaining the categorical variable sequence.

[0115] This embodiment generates a categorical variable sequence with ordered instructions by parsing and transforming the instruction content in the escrow fund transfer instruction sequence. This categorical variable sequence is then used as input to a probability prediction model, providing key data for the automated processing and instruction classification of escrow business, thereby improving the efficiency and accuracy of instruction processing.

[0116] Optionally, in the business status identification method provided in this application embodiment, generating the classification variable sequence of the escrow fund transfer instruction sequence includes: for each instruction content, in the case of obtaining classification variables from each instruction content, determining whether the type and quantity of classification variables meet preset requirements; if they meet the preset requirements, determining that the classification variables are correct; if they do not meet the preset requirements, determining that the classification variables are incorrect, and determining the missing data in the classification variables according to the preset requirements, and sending a prompt message to the user, wherein the prompt message is used to prompt the user to add the missing data.

[0117] Specifically, upon receiving an instruction, the system first parses the content of each escrow fund transfer instruction, extracting relevant categorical variables such as instruction type, amount, account information, and transaction time. Then, the system verifies whether the types and quantities of these categorical variables match preset requirements. For example, preset requirements might stipulate that each instruction must include instruction type, amount, and transaction time; if any one of these is missing, it is deemed non-compliant.

[0118] If the types and number of categorical variables meet the preset requirements, the system will determine that the extraction and identification of categorical variables are correct, and combine these categorical variables into a categorical variable sequence for subsequent business status prediction. If the types and number of categorical variables do not meet the preset requirements, the system will determine that the categorical variables are incorrect and automatically identify missing data items. For example, if an instruction does not specify the transaction time, the system will identify "transaction time" as missing data. Subsequently, the system will generate and send a prompt message to the user, clearly indicating which data items are missing and requesting the user to supplement this information.

[0119] This embodiment effectively avoids business processing problems caused by missing or incorrect data by performing data verification and integrity checks on the escrow fund transfer instructions, thereby ensuring the accuracy of subsequently determining the business operation status corresponding to each instruction.

[0120] Optionally, in the business status identification method provided in this application embodiment, obtaining classification variables from each instruction content in sequence to obtain a set of M classification variables includes: obtaining keywords indicated in the preset requirements to obtain multiple keywords; identifying instruction information corresponding to each keyword from the instruction content to obtain multiple instruction information; sequentially obtaining instruction data corresponding to each instruction information, and determining a set of instruction information and instruction data as a classification variable to obtain M classification variables.

[0121] Specifically, when acquiring categorical variables, the system first reads a list of keywords from preset requirements. These keywords indicate the types of information that need to be considered and extracted from the escrow fund transfer instructions. For example, the preset requirements might include the keyword "amount," used to extract the amount involved in the instruction, thereby determining the types of information the system needs to find and extract when parsing the instruction content.

[0122] Optionally, it is necessary to further analyze each instruction in the sequence of escrow fund transfer instructions to identify the instruction information corresponding to the keywords. For example, for the instruction "to inject 10,000 yuan into account A", the keyword can be identified as "amount", and the instruction information corresponding to this keyword is "transfer amount", and the instruction data corresponding to this instruction information is "10,000 yuan". Then, the various categorical variables in the transfer instructions are identified to obtain the instruction information and corresponding instruction data, and the instruction information and instruction data are combined to obtain the categorical variables.

[0123] This embodiment ensures the efficiency and accuracy of instruction processing by automatically parsing the sequence of escrow fund transfer instructions and generating a set of categorical variables.

[0124] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0125] Example 2

[0126] This application also provides a service status identification device. It should be noted that the service status identification device of this application can be used to execute the service status identification method provided in the above embodiments. The service status identification device provided in this application is described below.

[0127] According to an embodiment of this application, an apparatus for implementing the above-described method for identifying business status is also provided. Figure 3 This is a schematic diagram of a service status identification device according to Embodiment 2 of this application, as shown below. Figure 3 As shown, the device includes:

[0128] The generation unit 31 is used to receive the sequence of escrow fund transfer instructions sent by the user and generate a categorical variable sequence of the escrow fund transfer instruction sequence, wherein the escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer.

[0129] The first acquisition unit 32 is used to acquire the target parameters in the probability prediction model, wherein the target parameters include: an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability.

[0130] The calculation unit 33 is used to calculate the target parameters and the sequence of categorical variables using the Viterbi algorithm to obtain the business status sequence corresponding to the sequence of escrow fund transfer instructions.

[0131] The business status identification device provided in this application embodiment receives a sequence of escrow fund transfer instructions sent by a user through a generation unit 31, and generates a classification variable sequence of the escrow fund transfer instruction sequence, wherein the escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer; a first acquisition unit 32 acquires target parameters in a probability prediction model, wherein the target parameters include: an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability; a calculation unit 33 calculates the target parameters and the classification variable sequence using the Viterbi algorithm to obtain a business status sequence corresponding to the escrow fund transfer instruction sequence. By using the target parameters configured in the probability prediction model and the order of each escrow fund transfer instruction in the escrow fund transfer instruction sequence, the business status to which each escrow fund transfer instruction belongs is calculated, achieving the goal of eliminating the need for manual determination of the business status of the escrow fund transfer instructions, thereby achieving the technical effect of accurately determining the business status of each escrow fund transfer instruction, and solving the problem of low accuracy and efficiency in related technologies where the operational status of the business indicated by the instructions is determined manually.

[0132] Optionally, in the business status identification device provided in this application embodiment, before obtaining the target parameters in the probability prediction model, the device further includes: a second acquisition unit, used to acquire multiple historical escrow fund transfer instruction sequences and acquire the historical business status sequence corresponding to each historical escrow fund transfer instruction sequence; a determination unit, used to determine each historical escrow fund transfer instruction sequence and the corresponding historical business status sequence as a set of sample data to obtain multiple sets of sample data; and a training unit, used to train the probability prediction model based on the multiple sets of sample data using an expectation-maximization algorithm to obtain the target parameters in the probability prediction model.

[0133] Optionally, in the business state identification device provided in the embodiments of this application, the training unit includes: a generation module, used to randomly generate target parameters to obtain random probability parameters; an estimation module, used to estimate the probability distribution of the hidden state sequence using the random probability parameters for each set of sample data; and an adjustment module, used to adjust the random probability parameters according to the probability distribution until the log-likelihood of the probability prediction model under the random probability parameters is maximized, and to determine the random probability parameters at the current time as the target parameters.

[0134] Optionally, in the business state identification device provided in this application embodiment, the calculation unit 33 includes: a first calculation module, used to calculate the initial probability vector of each business state corresponding to the first feature element of the classification variable sequence through the initial probability vector in the target parameters; a second calculation module, used to calculate the probability of each business state when reaching the next feature element for each subsequent feature element in the classification variable sequence based on the current target path, transition probability and observation probability, and add the business state with the highest probability to the target path to obtain the updated target path, wherein the target path is composed of the business state with the highest probability; and a first determination module, used to determine the business states contained in the target path as a business state sequence after processing the last feature element in the classification variable sequence.

[0135] Optionally, in the business status identification device provided in this application embodiment, the generation unit 31 includes: an identification module, used to sequentially identify the instruction content of each escrow fund transfer instruction in the escrow fund transfer instruction sequence to obtain M instruction contents; an acquisition module, used to sequentially acquire classification variables from each instruction content to obtain M classification variable sets; and an arrangement module, used to arrange the M classification variable sets according to the order of the instruction contents to obtain a classification variable sequence.

[0136] Optionally, in the business status identification device provided in this application embodiment, the generation unit 31 includes: a judgment module, used to determine whether the type and quantity of the classification variables meet preset requirements when obtaining classification variables from each instruction content; a second determination module, used to determine that the classification variables are correct when the preset requirements are met; and a third determination module, used to determine that the classification variables are incorrect when the preset requirements are not met, and to determine the missing data in the classification variables according to the preset requirements, and to send a prompt message to the user, wherein the prompt message is used to prompt the user to add the missing data.

[0137] Optionally, in the business status identification device provided in the embodiments of this application, the acquisition module is used to include: an acquisition submodule, used to acquire the keywords indicated in the preset requirements to obtain multiple keywords; an identification submodule, used to identify the instruction information corresponding to each keyword from the instruction content to obtain multiple instruction information; and a determination submodule, used to sequentially acquire the instruction data corresponding to each instruction information, and determine a set of instruction information and instruction data as a classification variable to obtain M classification variables.

[0138] It should be noted that the above-mentioned generation unit 31, first acquisition unit 32, and calculation unit 33 correspond to steps S201 to S203 in Embodiment 1. The above modules and corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in Embodiment 1. It should be noted that the above modules or units can be hardware or software components stored in memory (e.g., memory 104) and processed by one or more processors (e.g., processors 102a, 102b, ..., 102n). The above modules can also be part of a device and run in the computer terminal 10 provided in Embodiment 1.

[0139] Example 3

[0140] Embodiments of this application may provide an electronic device. Figure 4 This is a structural block diagram of an electronic device according to an embodiment of this application. Figure 4 As shown, the electronic device may include: one or more ( Figure 4 (Only one is shown) processor 1002, memory 1004, memory controller, and peripheral interface, wherein the peripheral interface is connected to the radio frequency module, audio module and display.

[0141] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the methods and apparatus in the embodiments of this application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby implementing the above-described methods. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0142] The processor can access information and applications stored in memory via a transmission device to execute the following steps: receiving a sequence of escrow fund transfer instructions sent by the user, and generating a categorical variable sequence of the escrow fund transfer instruction sequence, wherein the escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer; obtaining the target parameters in the probabilistic prediction model, wherein the target parameters include: an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability; calculating the target parameters and the categorical variable sequence using the Viterbi algorithm to obtain the business state sequence corresponding to the escrow fund transfer instruction sequence, wherein the business state sequence is used to indicate the business state of each escrow fund transfer instruction.

[0143] The processor can access the information and application programs stored in the memory via the transmission device to perform the following steps: acquire multiple historical escrow fund transfer instruction sequences and acquire the historical business status sequences corresponding to each historical escrow fund transfer instruction sequence; determine each historical escrow fund transfer instruction sequence and the corresponding historical business status sequence as a set of sample data to obtain multiple sets of sample data; train the probabilistic prediction model based on the multiple sets of sample data using the expectation-maximization algorithm to obtain the target parameters in the probabilistic prediction model.

[0144] The processor can access the information and application programs stored in the memory via the transmission device to perform the following steps: randomly generate target parameters to obtain random probability parameters; for each set of sample data, use the random probability parameters to estimate the probability distribution of the hidden state sequence; adjust the random probability parameters according to the probability distribution until the log-likelihood of the probability prediction model under the random probability parameters is maximized, and determine the random probability parameters at the current time as the target parameters.

[0145] The processor can access the information and application programs stored in the memory via the transmission device to perform the following steps: Calculate the initial probability vector of each business state corresponding to the first feature element of the categorical variable sequence using the initial probability vector in the target parameters; for each subsequent feature element in the categorical variable sequence, calculate the probability of each business state when reaching the next feature element based on the current target path, transition probability, and observation probability, and add the business state with the highest probability to the target path to obtain the updated target path, where the target path consists of the business states with the highest probability; after processing the last feature element in the categorical variable sequence, determine the business states contained in the target path as the business state sequence.

[0146] The processor can access the information and application programs stored in the memory via the transmission device to execute the following steps: sequentially identify the instruction content of each escrow fund transfer instruction in the sequence of escrow fund transfer instructions to obtain M instruction contents; sequentially obtain categorical variables from each instruction content to obtain M sets of categorical variables; arrange the M sets of categorical variables according to the order of the instruction contents to obtain a sequence of categorical variables.

[0147] The processor can access information and applications stored in memory via a transmission device to execute the following steps: For each instruction, if categorical variables are obtained from each instruction, determine whether the type and quantity of the categorical variables meet preset requirements; if they meet the preset requirements, determine that the categorical variables are correct; if they do not meet the preset requirements, determine that the categorical variables are incorrect, identify missing data in the categorical variables according to the preset requirements, and send a prompt message to the user, wherein the prompt message is used to prompt the user to add the missing data.

[0148] The processor can access information and applications stored in memory via a transmission device to perform the following steps: obtain keywords indicated in a preset requirement to obtain multiple keywords; identify instruction information corresponding to each keyword from the instruction content to obtain multiple instruction information; sequentially obtain instruction data corresponding to each instruction information, and determine a set of instruction information and instruction data as a categorical variable to obtain M categorical variables.

[0149] Those skilled in the art will understand that Figure 4 The structure shown is for illustrative purposes only. Electronic devices can also be smartphones (such as Android phones, iOS phones, etc.), tablets, PDAs, mobile internet devices (MIDs), PADs, and other terminal devices. Figure 4 This does not limit the structure of the aforementioned electronic device. For example, electronic devices may also include components that are more... Figure 4 The more or fewer components shown (such as network interfaces, display devices, etc.), or having the same Figure 4 The different configurations shown.

[0150] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0151] Example 4

[0152] Embodiments of this application also provide a storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the service status identification method provided in Embodiment 1.

[0153] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0154] Embodiments of this application also provide a computer program product, which, when executed on a data processing device, is adapted to perform the steps of a business status identification method.

[0155] Embodiments of this application also provide a computer-readable storage medium, which includes a stored executable program, wherein the executable program controls the device where the computer-readable storage medium is located to perform the above-described business state identification method during runtime.

[0156] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0157] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0158] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0159] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0160] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0161] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0162] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for identifying business status, characterized in that, include: Receive a sequence of escrow fund transfer instructions sent by the user, and generate a categorical variable sequence of the escrow fund transfer instruction sequence, wherein the escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer; Obtain the target parameters in the probabilistic prediction model, wherein the target parameters include: an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability; The target parameters and the categorical variable sequence are calculated using the Viterbi algorithm to obtain the business status sequence corresponding to the escrow fund transfer instruction sequence, wherein the business status sequence is used to indicate the business status of each escrow fund transfer instruction.

2. The method according to claim 1, characterized in that, Before obtaining the target parameters in the probabilistic prediction model, the method further includes: Obtain multiple historical escrow fund transfer instruction sequences, and obtain the historical business status sequence corresponding to each historical escrow fund transfer instruction sequence; Each historical sequence of escrow fund transfer instructions and its corresponding historical business status sequence are identified as a set of sample data, resulting in multiple sets of sample data. The probability prediction model is trained using the expectation-maximization algorithm based on the multiple sets of sample data to obtain the target parameters in the probability prediction model.

3. The method according to claim 2, characterized in that, The probability prediction model is trained using the expectation-maximization algorithm based on the multiple sets of sample data, and the target parameters in the probability prediction model include: The target parameters are randomly generated to obtain random probability parameters; For each set of sample data, the probability distribution of the hidden state sequence is estimated using the random probability parameters; Based on the probability distribution, the random probability parameters are adjusted until the log-likelihood of the probability prediction model under the random probability parameters is maximized, and the random probability parameters at the current time are determined as the target parameters.

4. The method according to claim 1, characterized in that, The target parameters and the categorical variable sequence are calculated using the Viterbi algorithm to obtain the business status sequence corresponding to the escrow fund transfer instruction sequence, including: Using the initial probability vector in the target parameters, calculate the initial probability vector corresponding to each business state for the first feature element of the categorical variable sequence; For each subsequent feature element in the categorical variable sequence, based on the current target path, the transition probability, and the observation probability, the probability of each business state when reaching the next feature element is calculated, and the business state with the highest probability is added to the target path to obtain the updated target path, wherein the target path is composed of the business state with the highest probability. After processing the last feature element in the categorical variable sequence, the business states contained in the target path are determined as the business state sequence.

5. The method according to claim 1, characterized in that, The categorical variable sequence for generating the escrow fund transfer instruction sequence includes: The instruction content of each escrow fund transfer instruction in the sequence of escrow fund transfer instructions is identified sequentially to obtain M instruction contents; The categorical variables are obtained sequentially from each instruction, resulting in a set of M categorical variables; The M sets of categorical variables are arranged in the order of the instructions to obtain the categorical variable sequence.

6. The method according to claim 5, characterized in that, The categorical variable sequence for generating the escrow fund transfer instruction sequence includes: For each instruction, after obtaining the classification variables from each instruction, it is determined whether the type and quantity of the classification variables meet the preset requirements; If the preset requirements are met, the classification variable is determined to be correct; If the preset requirements are not met, the classification variable is determined to be incorrect, and the missing data in the classification variable is determined according to the preset requirements. A prompt message is then sent to the user, wherein the prompt message is used to prompt the user to add the missing data.

7. The method according to claim 6, characterized in that, The categorical variables are retrieved sequentially from each instruction, resulting in a set of M categorical variables, including: Obtain the keywords indicated in the preset requirements to obtain multiple keywords; The instruction information corresponding to each keyword is identified from the instruction content to obtain multiple instruction information; The instruction data corresponding to each instruction information is obtained sequentially, and a set of instruction information and instruction data is determined as a classification variable to obtain the M classification variables.

8. A device for identifying business status, characterized in that, include: A generation unit is used to receive a sequence of escrow fund transfer instructions sent by a user and generate a categorical variable sequence of the escrow fund transfer instruction sequence, wherein the escrow fund transfer instruction sequence includes M escrow fund transfer instructions, where M is a positive integer; The first acquisition unit is used to acquire target parameters in the probability prediction model, wherein the target parameters include: an initial probability vector representing the initial state probability, a state transition probability matrix representing the state transition probability, and an observation probability matrix representing the observation probability. The calculation unit is used to calculate the target parameters and the categorical variable sequence using the Viterbi algorithm to obtain the business status sequence corresponding to the escrow fund transfer instruction sequence.

9. A computer program product comprising computer instructions, characterized in that, When the computer instructions are executed by the processor, they implement the steps of the business status identification method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, include: Memory, which stores executable programs; A processor for running the program, wherein the program executes the business status identification method according to any one of claims 1 to 7 when it runs.

Images