Training method and device of transaction data processing model, medium and electronic equipment

Abstract

The invention relates to the technical field of artificial intelligence, and provides a training method of a transaction data processing model, a training device of the transaction data processing model, a computer readable storage medium and an electronic device.The training method of the transaction data processing model comprises the steps that article data sets of multiple articles in a historical period are acquired, dividing the article data set into a training data set and a training label set according to the time division points; calling an original transaction data processing model; the original transaction data processing model is composed of a plurality of networks; performing iterative training on the original transaction data processing model by using the training data set and the training label set to update network parameters of partial specified networks in the plurality of networks, and obtaining a trained transaction data processing model; wherein the partial specified network comprises a time series data coding network. The method can guarantee the training efficiency of the model and improve the processing performance of the model for the time series data.

Description

Background Art

[0002] With the rapid development and progress of big data and artificial intelligence technology, the relevant data processing field is also constantly innovating and progressing. As an important part of this field, transaction data prediction is of great significance to the company's inventory management, supply chain optimization, marketing strategy formulation, etc.

[0003] In the related technology, the transaction data of related items can be predicted through machine learning models. However, the related machine learning models have poor performance when processing time series data, which affects the prediction accuracy of the output data.

[0004] In view of this, there is an urgent need in this field to develop a new training method and device for transaction data processing models.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure. Summary of the invention

[0006] The purpose of the present disclosure is to provide a training method for a transaction data processing model, a training device for a transaction data processing model, a computer-readable storage medium and an electronic device, thereby at least to a certain extent overcoming the technical problem of insufficient data prediction accuracy caused by the limitations of related technologies.

[0007] Other features and advantages of the present disclosure will become apparent from the following detailed description, or may be learned in part by the practice of the present disclosure.

[0008] According to a first aspect of the present disclosure, a training method for a transaction data processing model is provided, comprising:

[0009] Acquire an item data set of multiple items in a historical period, and divide the item data set into a training data set and a training label set according to a time segmentation point; the item data set includes an item description text of each item corresponding to the historical period and historical transaction time series data of each item in the historical period;

[0010] Calling an original transaction data processing model; the original transaction data processing model is composed of multiple networks;

[0011] The original transaction data processing model is iteratively trained using the training data set and the training label set to update network parameters of some specified networks among the multiple networks to obtain a trained transaction data processing model; wherein the some specified networks include a time series data encoding network.

[0012] In an exemplary embodiment of the present disclosure, the item description text includes at least one of the following:

[0013] The basic information of each item, the transaction market information of each item in the historical period, and the characteristic description information corresponding to the historical transaction time series data.

[0014] In an exemplary embodiment of the present disclosure, after acquiring an item dataset of a plurality of items within a historical period, the method further includes:

[0015] Preprocessing the item data set of the plurality of items in the historical period to obtain the preprocessed item data set;

[0016] The preprocessing includes at least one of the following: outlier removal processing, missing value filling processing and data standardization processing.

[0017] In an exemplary embodiment of the present disclosure, dividing the object dataset into a training dataset and a training label set according to time segmentation points includes:

[0018] Dividing the historical period into a first period and a second period according to the time segmentation point to obtain a training data set corresponding to the first period and a training label set corresponding to the second period;

[0019] The first time period starts at the start time of the historical time period and ends at the time division point; the training data set includes the first item description text of each item corresponding to the first time period and the first transaction time series data of each item in the first time period;

[0020] The second time period starts at the time division point and ends at the end time of the historical time period; the training label set includes the second item description text of each item corresponding to the second time period and the second transaction time series data of each item in the second time period.

[0021] In an exemplary embodiment of the present disclosure, the iterative training of the original transaction data processing model using the training data set and the training label set to update network parameters of a portion of designated networks in the multiple networks includes:

[0022] Inputting the training sample set into the original transaction data processing model, and outputting the predicted transaction time series data of each item corresponding to the second time period through the original transaction data processing model;

[0023] Determining a first loss value of the original transaction data processing model according to a difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period;

[0024] The network parameters of a portion of the specified networks among the multiple networks are updated according to the first loss value.

[0025] In an exemplary embodiment of the present disclosure, the plurality of networks include a text information encoding network, a time series data encoding network, and a decoding network;

[0026] The step of inputting the training sample set into the original transaction data processing model and outputting the predicted transaction time series data of each item corresponding to the second time period through the original transaction data processing model includes:

[0027] Encoding the first item description text corresponding to the first time period for each item through the text information encoding network to obtain a word embedding vector corresponding to each item;

[0028] Encode the first transaction time series data of each item in the first time period through the time series data encoding network to obtain a time series embedding vector corresponding to each item;

[0029] Concatenate the word embedding vector and the time series embedding vector corresponding to each item to obtain a concatenated vector;

[0030] The concatenated vector corresponding to each item is decoded through the decoding network to obtain predicted transaction time series data of each item corresponding to the second time period.

[0031] In an exemplary embodiment of the present disclosure, encoding the first item description text corresponding to the first time period of each item by the text information encoding network to obtain the word embedding vector corresponding to each item includes:

[0032] Performing word segmentation processing on the first item description text corresponding to the first time period of each item through the text information encoding network to obtain multiple target word segments;

[0033] The standard word vector corresponding to each target word segment is matched from the standard word list to obtain the word embedding vector corresponding to each item.

[0034] In an exemplary embodiment of the present disclosure, encoding the first transaction time series data of each item in the first time period through the time series data encoding network to obtain the time series embedding vector corresponding to each item includes:

[0035] Splitting the first transaction time series data of each item in the first time period by a splitting function to obtain a plurality of subsequences;

[0036] Perform sequence encoding on each of the subsequences by using an embedding function to obtain each subvector;

[0037] According to the multiple sub-vectors corresponding to the multiple sub-sequences, a temporal embedding vector corresponding to each item is determined.

[0038] In an exemplary embodiment of the present disclosure, after obtaining the predicted transaction time series data corresponding to the second time period for each item, the method further includes:

[0039] Determining a first loss value of the original transaction data processing model according to a difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period;

[0040] Determining a second loss value of the original transaction data processing model according to a degree of similarity between each sub-vector associated with each item and each standard word vector in a standard word list;

[0041] Determining a comprehensive loss value of the original transaction data processing model according to the first loss value and the second loss value;

[0042] The network parameters of a portion of the plurality of networks are updated according to the comprehensive loss value.

[0043] In an exemplary embodiment of the present disclosure, the partially specified network further includes a decoding network.

[0044] In an exemplary embodiment of the present disclosure, after obtaining the trained transaction data processing model, the method further includes:

[0045] Inputting the item data set of the plurality of items in the historical period into the trained transaction data processing model;

[0046] The trained transaction data processing model is used to predict future transaction time series data of each item in a future time period based on the item data set of the multiple items in a historical time period.

[0047] According to a second aspect of the present disclosure, there is provided a training device for a transaction data processing model, comprising:

[0048] A data acquisition module is used to acquire an item data set of multiple items in a historical period, and divide the item data set into a training data set and a training label set according to a time segmentation point; the item data set includes an item description text of each item corresponding to the historical period and a historical transaction time series data of each item in the historical period;

[0049] A model calling module, used to call an original transaction data processing model; the original transaction data processing model is composed of multiple networks;

[0050] A model training module is used to iteratively train the original transaction data processing model using the training data set and the training label set to update the network parameters of some specified networks among the multiple networks to obtain a trained transaction data processing model; wherein the some specified networks include a time series data encoding network.

[0051] According to a third aspect of the present disclosure, there is provided a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the training method of the transaction data processing model described in the first aspect is implemented.

[0052] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the training method of the transaction data processing model described in the first aspect above by executing the executable instructions.

[0053] It can be seen from the above technical solutions that the training method of the transaction data processing model, the training device of the transaction data processing model, the computer-readable storage medium and the electronic device in the exemplary embodiments of the present disclosure have at least the following advantages and positive effects:

[0054] In the technical solutions provided by some embodiments of the present disclosure, an item data set of multiple items in a historical period is obtained, and the item data set is divided into a training data set and a training label set according to a time segmentation point; the item data set includes an item-related text information item description text corresponding to each item in the historical period and the historical transaction time series data of each item in the historical period; the original transaction data processing model is called; the original transaction data processing model is composed of multiple networks; the original transaction data processing model is iteratively trained using the training data set and the training label set to update the network parameters of some specified networks (including the time series data encoding network) in the multiple networks, and the trained The transaction data processing model, on the one hand, enables the model to use the time series transaction data of items in combination with unstructured text information to learn the characteristics of items from multiple dimensions, capture the characteristics of items more comprehensively, thereby improving data utilization efficiency and helping to improve the prediction accuracy of the model; further, it can maximize the use of parameter configurations of existing large models, avoiding training a large time series inference model with matching parameter magnitude from scratch, thereby reducing investment in data, computing power and human resources, and reducing operating costs; on the other hand, it can focus on iterative training of the time series data encoding network, thereby enhancing the performance of the large model in processing time series data and further ensuring the prediction accuracy of the model.

[0055] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.

[0057] Figure 1 A schematic diagram showing a flow chart of a training method for a transaction data processing model in an embodiment of the present disclosure;

[0058] Figure 2 A schematic diagram showing a flow chart of how to update network parameters of a portion of a specified network in an embodiment of the present disclosure;

[0059] Figure 3 A flow chart showing how to output predicted transaction time series data corresponding to the second time period for each item through the original transaction data processing model in an embodiment of the present disclosure;

[0060] Figure 4 A flow chart showing how to encode the first transaction time series data of each item in the first time period through a time series data encoding network in an embodiment of the present disclosure to obtain a time series embedding vector corresponding to each item;

[0061] Figure 5 A schematic diagram showing another flow chart of how to update network parameters of a portion of a specified network in an embodiment of the present disclosure;

[0062] Figure 6 The overall network architecture diagram of the transaction data processing model in the embodiment of the present disclosure is shown;

[0063] Figure 7 A schematic diagram showing the overall training process of the transaction data processing model in an embodiment of the present disclosure is shown;

[0064] Figure 8 A schematic diagram showing the structure of a training device for a transaction data processing model in an exemplary embodiment of the present disclosure;

[0065] Fig. 9 A schematic structural diagram of an electronic device in an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0066] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0067] The terms "a", "an", "the" and "said" are used in this specification to indicate the presence of one or more elements / components / etc.; the terms "including" and "having" are used to express an open-ended inclusion and mean that additional elements / components / etc. may exist in addition to the listed elements / components / etc.; the terms "first" and "second" etc. are used only as labels and are not intended to limit the quantity of their objects.

[0068] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings represent the same or similar parts, and their repeated descriptions will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities.

[0069] In the process of realizing automation and intelligence in many commodity trading scenarios, the time series prediction of commodity trading volume plays a very important role. For example, on some online shopping websites, the transaction volume of each category of commodities in the future is a variable that needs to be considered in a series of intelligent decisions (such as stocking, promotion, etc.). Therefore, the ability of prediction technology will ultimately have an important impact on sales revenue, inventory costs, etc. At the same time, the number of commodities sold at the same time on large online shopping websites can reach up to millions, and large-scale time series have brought new challenges to modern time series prediction technology.

[0070] Existing transaction volume forecasting methods mainly use traditional statistical methods, machine learning models, etc. These methods predict future sales by analyzing historical sales data and combining market trends, seasonal factors, promotional activities and other information. With the improvement of big data and computing power, these technologies have significantly improved in forecasting accuracy and efficiency.

[0071] With the rise and application of large language models (LLMs), these models have demonstrated excellent understanding and processing capabilities for massive text data with the help of deep learning algorithms and advanced natural language processing technologies. This not only provides a powerful tool for analyzing and interpreting complex language patterns, but also opens up new perspectives and space for completing time series-related reasoning tasks.

[0072] Although the existing transaction volume forecasting technology has improved the efficiency and accuracy of forecasting to a certain extent, the following problems still exist:

[0073] First, traditional statistical models and machine learning models often rely on expert experience and manual intervention. For example, for brand, category, attributes, etc. of an item, the common method is to process them into qualitative or quantitative features, which makes the data processing process complicated and time-consuming. These models require a lot of data preprocessing and feature engineering, which increases the complexity of the system and increases maintenance costs. Although these methods can improve the accuracy of predictions, they still require professionals to perform tedious feature selection and parameter adjustment;

[0074] Second, although open source large language models perform well in text generation and reasoning, they are mainly designed to process text data and often perform poorly when directly applied to the prediction of time series data.

[0075] In view of the above problems, the existing inventory management methods often appear to be insufficient in prediction accuracy, lack of adaptability, cumbersome data processing and poor real-time performance when dealing with the rapid changes in the market environment. Therefore, it is urgent to develop a smarter, more efficient and more flexible transaction volume prediction method and system to overcome the limitations of existing technologies and meet the needs of relevant online shopping platforms and users.

[0076] In the embodiments of the present disclosure, a training method for a transaction data processing model is first provided, which at least to some extent overcomes the defect of insufficient prediction accuracy in the related art.

[0077] Figure 1 A flow chart of a method for training a transaction data processing model in an embodiment of the present disclosure is shown. The execution subject of the method for training a transaction data processing model may be a server that trains the transaction data processing model.

[0078] refer to Figure 1 According to an embodiment of the present disclosure, a training method for a transaction data processing model includes the following steps:

[0079] Step S110, obtaining an item data set of multiple items in a historical period, and dividing the item data set into a training data set and a training label set according to a time segmentation point; the item data set includes an item description text corresponding to each item in the historical period and historical transaction time series data of each item in the historical period;

[0080] Step S120, calling the original transaction data processing model; the original transaction data processing model is composed of multiple networks;

[0081] Step S130, iteratively training the original transaction data processing model using the training data set and the training label set to update network parameters of some specified networks among the multiple networks to obtain a trained transaction data processing model; wherein the some specified networks include a time series data encoding network.

[0082] exist Figure 1 In the technical solution provided by the illustrated embodiment, an item data set of multiple items in a historical period is obtained, and the item data set is divided into a training data set and a training label set according to a time segmentation point; the item data set includes an item-related text information item description text of each item corresponding to the historical period and the historical transaction time series data of each item in the historical period; an original transaction data processing model is called; the original transaction data processing model is composed of multiple networks; the original transaction data processing model is iteratively trained using the training data set and the training label set to update the network parameters of some specified networks (including the time series data encoding network) among the multiple networks. To obtain a trained transaction data processing model, on the one hand, the model can use the time series transaction data of items in combination with unstructured text information to learn the characteristics of items from multiple dimensions, capture the characteristics of items more comprehensively, thereby improving data utilization efficiency and helping to improve the prediction accuracy of the model; further, it can maximize the use of the parameter configuration of the existing large model and avoid training a time series reasoning large model with a matching parameter magnitude from scratch, thereby reducing the investment in data, computing power and human resources and reducing operating costs; on the other hand, it can focus on iterative training of the time series data encoding network, thereby enhancing the performance of the large model in processing time series data and further ensuring the prediction accuracy of the model.

[0083] The following Figure 1 The specific implementation process of each step is described in detail:

[0084] In step S110, an item dataset of multiple items in a historical period is obtained, and the item dataset is divided into a training dataset and a training label set according to time division points.

[0085] In this step, an item data set of multiple items in a historical period may be obtained, and the item data set may include an item description text of each item corresponding to the historical period and historical transaction time series data of each item in the historical period.

[0086] Among them, the above historical period can be the past two years, which can be set according to actual conditions, and the present disclosure does not make any special limitations on this.

[0087] The above-mentioned item description text includes at least one of the following: basic item information of each item, transaction market information of each item in a historical period, and characteristic description information corresponding to historical transaction time series data.

[0088] The basic information of an item refers to the most basic information related to the item. These data are the most basic elements for understanding and analyzing the characteristics of the item, and are also the basis for various data analysis, modeling and decision-making. For example, the basic information of an item may include the following:

[0089] ① Item basis: the name of the item displayed on the e-commerce platform, such as "Smartphone A1", "Smartphone A2", etc.;

[0090] ② Category: the category to which the item belongs, such as “electronic products - mobile phones”;

[0091] ③ Brand: the brand of the item, such as "Brand X";

[0092] ④ Attributes: Other specific attributes of the item, such as: color: "black", size: "6.5 inches", storage capacity: "128GB", processor: "octa-core processor", battery capacity: "4000mAh".

[0093] The historical transaction time series data of each item in the historical period can be arranged by the daily transaction volume of each item in the past two years. The daily transaction volume of each item in the past two years can refer to Table 1:

[0094] Table 1

[0095]

[0096] The characteristic description information corresponding to the above historical transaction time series data can be used to describe the data characteristics corresponding to the historical transaction time series data. For example, reference can be made to Table 2, which shows an exemplary table of characteristic description information corresponding to the historical transaction time series data:

[0097] Table 2

[0098]

[0099]

[0100] The transaction market information of each item in the historical period refers to the summary of the transaction activities and related information of each item in the historical period, which can help analyze the market performance, consumer behavior and market trends of the items. For example, the market information can be the promotional discount price of each mobile phone every day in the past two years. For example, Table 3 is shown as an exemplary table of the above market information:

[0101] Table 3

[0102] date Phone model Original price (unit: yuan) Promotional discount price (unit: yuan) 2022-01-01 Smartphone A1 5000 4500 2022-01-01 Smartphone A2 4000 3600 2022-01-02 Smartphone A1 3500 3200

[0103] After obtaining the item datasets of the above-mentioned multiple items in the historical period, the present disclosure may also preprocess the item datasets of the multiple items in the historical period to obtain the preprocessed item datasets.

[0104] The above preprocessing includes at least one of the following: outlier removal, missing value filling and data standardization.

[0105] Among them, outliers refer to data points in the data set that deviate significantly from other observations. These data points may be caused by measurement errors, input errors, or other reasons. Therefore, outliers can be identified based on the experience or expertise of domain experts and removed from the item data set to ensure data accuracy and reduce the impact of outliers on analysis results.

[0106] Missing values ​​refer to data points where some observations in a data set are missing. These missing values ​​may affect subsequent analysis and modeling. Therefore, missing values ​​can be filled by interpolation, averaging or median, data prediction, etc. to ensure the integrity and availability of the data set.

[0107] Data normalization refers to the process of converting data to the same scale, usually to eliminate the dimensional differences between different features and make the data more suitable for modeling.

[0108] After preprocessing the item data set, the item data set can be divided into a training data set and a training label set according to the time division point. Specifically, the historical period can be divided into a first period and a second period according to the time division point t0 to obtain a training data set corresponding to the first period and a training label set corresponding to the second period. Exemplarily, taking the historical period as (1, ..., t1) as an example, the first period can be expressed as (1, ..., t0), and the second period can be expressed as (t0, ..., t1), that is, the first period starts at the start time 1 of the historical period and ends at the time division point t0, and the second period starts at the time division point t0 and ends at the end time of the above historical period.

[0109] Thus, the training data set may include the first item description text corresponding to the above-mentioned first time period for each item and the first transaction time series data of each item in the first time period; and the training label set may include the second item description text corresponding to the second time period for each item and the second transaction time series data of each item in the second time period.

[0110] After obtaining the above training data set and the above training label set, the following step S120 may be performed.

[0111] In step S120, the original transaction data processing model is called.

[0112] In this step, the original transaction data processing model can be called, and the original data processing model can be a transaction data processing model that has a prototype in the existing framework. The original transaction data processing model can be composed of multiple networks, and the multiple networks can include a text information encoding network, a time series data encoding network, and a decoding network. The specific processing process of each network is described in step S130 below.

[0113] In step S130, the original transaction data processing model is iteratively trained using the training data set and the training label set to update network parameters of some designated networks in the multiple networks and obtain a trained transaction data processing model.

[0114] In this step, the original transaction data processing model can be iteratively trained using the training data set and the training label set to update the network parameters of some specified networks in the multiple networks to obtain a trained transaction data processing model. Specifically, the present disclosure can freeze the network parameters of the text information encoding network, so that only the parameters of the time series data encoding network and the decoding network are iterated, thereby maximizing the use of the existing model framework, improving the training efficiency of the model, and optimizing the time series data processing performance of the large model in a targeted manner, thereby improving the processing performance of the large model for time series data.

[0115] In a first optional implementation, reference Figure 2 , Figure 2 A schematic diagram showing a flow chart of how to update network parameters of some designated networks in multiple networks in an embodiment of the present disclosure includes steps S201 to S203:

[0116] In step S201, the training sample set is input into the original transaction data processing model, and the original transaction data processing model outputs the predicted transaction time series data of each item corresponding to the second time period.

[0117] In this step, the training sample set may be input into the original transaction data processing model, and the original transaction data processing model may be used to output the predicted transaction time series data of each item corresponding to the second time period.

[0118] Specifically, please refer to Figure 3 , Figure 3 A flow chart showing how to output predicted transaction time series data corresponding to the second time period for each item through the original transaction data processing model in an embodiment of the present disclosure includes steps S301 to S304:

[0119] In step S301, the first item description text corresponding to the first time period of each item is encoded through a text information encoding network to obtain a word embedding vector corresponding to each item.

[0120] In this step, the first item description text corresponding to the first time period of each item can be encoded by the text information encoding network to obtain the M-dimensional word embedding vector corresponding to each item. Word embedding is a technology that maps words in a text to a multidimensional vector space. Each word (or "token") has a corresponding vector representation in the word embedding space. These vectors can not only reflect the semantic information of the word itself, but also reflect the relationship between words.

[0121] Specifically, the first item description text corresponding to the first time period of each item can be segmented through the above-mentioned text information encoding network to obtain multiple target segmentation tokens. After that, the standard word vector corresponding to each target segmentation can be matched from the standard word list to obtain the word embedding vector corresponding to each item. Exemplarily, taking the above-mentioned original transaction data processing model as GPT-2-small as an example, its standard word list can contain 50257 tokens, and each token has a corresponding standard word vector with a dimension of 768. Thus, each of the above-mentioned target segmentations can correspond to a word vector with a dimension of 768.

[0122] In step S302, the first transaction time series data of each item in the first time period is encoded through a time series data encoding network to obtain a time series embedding vector corresponding to each item.

[0123] In this step, the first transaction time series data of each item in the first time period can be encoded by the above-mentioned time series data encoding network to obtain the time series embedding vector corresponding to each item.

[0124] For details, please refer to Figure 4 , Figure 4A flow chart showing how to encode the first transaction time series data of each item in the first time period through a time series data encoding network in an embodiment of the present disclosure to obtain a time series embedding vector corresponding to each item includes steps S401 to S403:

[0125] In step S401, each first transaction time series data is segmented to obtain multiple subsequences.

[0126] In this step, the time series data encoding network can segment each of the above first transaction time series data into K (K is an integer greater than 1) non-overlapping subsequences by using the segmentation function F(s):x→s. Where x represents each first transaction time series data corresponding to each item, and s represents each subsequence after the segmentation. F(s) is usually a sliding window, each segment is a list of time points (t i and t j Indicates the endpoint time corresponding to the subsequence after splitting, 1≤t i ≤t j ≤T, T represents the total number of time points).

[0127] In step S402, each subsequence is sequence-encoded by an embedding function to obtain each subvector.

[0128] In this step, we can embed the function F e :s k ∈R D×T →e k ∈R M Sequence encoding is performed on each subsequence to obtain each M-dimensional subvector e k ,

[0129] Among them, the above embedding function can be a fully connected neural network model with parameters to be trained, whose input dimension is D*T (D represents the total number of items) and whose output dimension is M.

[0130] In step S403, a temporal embedding vector corresponding to each item is determined according to the multiple sub-vectors corresponding to the multiple sub-sequences.

[0131] In this step, for example, multiple sub-vectors e corresponding to multiple sub-sequences can be k Concatenate to determine the temporal embedding vector corresponding to each item.

[0132] After obtaining each word embedding vector and each time series embedding vector, you can then refer to Figure 3,In step S303, the word embedding vector and the time series embedding vector corresponding to each item are concatenated to obtain a concatenated vector.

[0133] In this step, the word embedding vector and the time series embedding vector corresponding to each item can be concatenated to obtain a concatenated vector.

[0134] In step S304, the concatenated vector corresponding to each item is decoded through a decoding network to obtain predicted transaction time series data corresponding to the second time period for each item.

[0135] In this step, the decoding module can be a fully connected neural network model to be trained, whose input dimension is M and output dimension is D*H (H is the total length of the time points included in the second time period to be predicted). Therefore, the splicing vector corresponding to each item can be decoded by the decoding network to remap the splicing vector back to the D*H dimension to obtain the predicted transaction time series data of each item corresponding to the second time period.

[0136] Next reference Figure 2 In step S202, a first loss value of the original transaction data processing model is determined according to the degree of difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period.

[0137] In this step, the first loss value lossA of the original transaction data processing model can be determined according to the difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period.

[0138] Specifically, considering that the predicted transaction time series data is composed of a plurality of predicted transaction data, the second transaction time series data is composed of a plurality of second transaction data, and the number of the predicted transaction data is the same as the number of the second transaction data, the first loss value lossA can be calculated based on the following formula 1:

[0139]

[0140] Among them, θ represents the network parameters corresponding to the specified network that needs to be trained, D represents the total number of items, and T represents the total number of prediction time points contained in the second period that needs to be predicted. It represents the degree of difference between the predicted transaction data and the corresponding second transaction data at each predicted time point for each item.

[0141] In step S203, network parameters of a portion of designated networks among the multiple networks are updated according to the first loss value.

[0142] In this step, based on the above-mentioned first loss value, the gradient descent algorithm can be used to update the network parameters corresponding to the above-mentioned time series data encoding network and the above-mentioned decoding network. When the above-mentioned first loss value tends to converge (for example: less than or equal to the preset loss value), the final network parameters corresponding to the above-mentioned time series data encoding network and the above-mentioned decoding network can be determined to obtain a trained transaction data processing model.

[0143] In a second optional implementation, after obtaining the first loss value, you can refer to Figure 5 , Figure 5 A schematic diagram showing another process of updating network parameters of some designated networks in multiple networks in an embodiment of the present disclosure includes steps S501 to S503:

[0144] In step S501, a second loss value of the original transaction data processing model is determined according to the degree of similarity between each sub-vector associated with each item and each standard word vector in the standard vocabulary.

[0145] In this step, each sub-vector e associated with each item can be k Each standard word vector word_embedding contained in the standard word list of the above original transaction data processing model j The similarity between the two is used to determine the second loss value of the original transaction data processing model. Exemplarily, the second loss value lossB can be calculated based on the following formula 2:

[0146]

[0147] Among them, K represents the total number of subsequences obtained after splitting the historical transaction time series data of each item, and P represents the total number of standard word vectors contained in the above standard vocabulary.

[0148] sim(e k ,word_embedding j ) represents the sub-vector e corresponding to each sub-sequence k And each standard word vector word_embedding j The similarity between them can be calculated, for example, by cosine similarity.

[0149] In step S502, a comprehensive loss value of the original transaction data processing model is determined based on the first loss value and the second loss value.

[0150] In this step, a comprehensive loss value lossC may be determined according to the first loss value and the second loss value. Specifically, the comprehensive loss value may be calculated based on the following formula 3:

[0151]

[0152] Wherein, λ is a weight parameter, which is a rational number greater than 0. The present disclosure solves the problem by making e k With the standard word vector word_embedding j The similarity is the greatest, which can force the sub-vector corresponding to the subsequence to be aligned with the standard word vector, and can realize the similarity between the text describing the word vector such as big, small, rising, falling, stable, and fluctuating and the specific numerical vector.

[0153] In step S503, network parameters of some designated networks among the multiple networks are updated according to the comprehensive loss value.

[0154] In this step, the network parameters corresponding to the above-mentioned time series data encoding network and the above-mentioned decoding network can be updated using the gradient descent algorithm based on the above-mentioned comprehensive loss value. When the above-mentioned comprehensive loss value tends to converge (for example: less than or equal to the preset loss value), the final network parameters corresponding to the above-mentioned time series data encoding network and the above-mentioned decoding network can be determined to obtain a trained transaction data processing model.

[0155] After obtaining the trained transaction data processing model, the present invention can input the item data set of multiple items in the historical period into the trained transaction data processing model; and predict the future transaction time series data of multiple items in the future period based on the item data set of multiple items in the historical period through the trained transaction data processing model.

[0156] Specifically, the text information encoding network of the above-mentioned trained transaction data processing model can encode the item description text corresponding to each item in the historical period to obtain the word embedding vector corresponding to each item; the time series data encoding network of the above-mentioned trained transaction data processing model can encode the historical transaction time series data of each item in the historical period to obtain the time series embedding vector corresponding to each item; then, the word embedding vector and the time series embedding vector corresponding to each item are spliced ​​to obtain a spliced ​​vector; and then the splicing vector corresponding to each item is decoded through the decoding network to obtain the future transaction time series data of each item in the future period.

[0157] refer to Figure 6 , Figure 6 The overall network architecture diagram of the transaction data processing model in the embodiment of the present disclosure is shown as follows: Figure 6 As shown:

[0158] The transaction data processing model includes a text information encoding network (parameters are frozen and do not need to be updated), a time series data encoding network and a decoding network. The text information encoding network is used to encode the item description text to obtain a word embedding vector. The time series data encoding network is used to encode the historical transaction time series data to obtain a time series embedding vector. The decoding network is used to dimensionally splice the word embedding vector and the time series embedding vector, and decode the spliced ​​vector to obtain the predicted transaction time series data of each item corresponding to the future time period.

[0159] refer to Figure 7 , Figure 7 The overall training process diagram of the transaction data processing model in the embodiment of the present disclosure is shown, including steps S701 to S706:

[0160] In step S701, start;

[0161] In step S702, item data is collected;

[0162] In step S703, data preprocessing (i.e., outlier removal, missing value filling and data standardization);

[0163] In step S704, model training process;

[0164] In step S705, a trained transaction data processing model is obtained;

[0165] In step S706, end.

[0166] Based on the technical solutions in this disclosure, this disclosure can at least achieve the following technical effects:

[0167] First, improve time series prediction capabilities and reduce training costs: In view of the shortcomings of existing open source large language models in time series data analysis, this paper strengthens the performance of these models on time series data by dividing time series data into subsequences, avoiding the need to train a large time series inference model with matching parameter magnitude from scratch, thereby significantly reducing the investment in data, computing power, and human resources;

[0168] Second, it simplifies data processing and feature engineering: By converting time series data into input suitable for large language models, the parameter weights of existing large models can be reused to the maximum extent, thereby using the text reasoning capabilities of existing open source large models to generate prediction results, reducing manual participation in complex data processing and feature engineering work. This not only improves the operating efficiency of the system, but also reduces the complexity and maintenance cost of the system.

[0169] Third, improve prediction accuracy: the model can make more accurate data predictions by integrating basic information of items, historical transaction volume, promotional information, etc.

[0170] Therefore, the present disclosure can improve the efficiency and accuracy of transaction volume prediction based on large models, reduce operating costs, improve the overall stability and responsiveness of the supply chain, and bring significant competitive advantages to enterprises.

[0171] The present disclosure also provides a training device for a transaction data processing model. Figure 8 A schematic diagram showing the structure of a training device for a transaction data processing model in an exemplary embodiment of the present disclosure is shown; Figure 8 As shown, the training device 800 for the transaction data processing model may include a data acquisition module 810, a model calling module 820 and a model training module 830. Among them:

[0172] The data acquisition module 810 is used to acquire an item data set of multiple items in a historical period, and divide the item data set into a training data set and a training label set according to a time segmentation point; the item data set includes an item description text of each item corresponding to the historical period and historical transaction time series data of each item in the historical period;

[0173] The model calling module 820 is used to call the original transaction data processing model; the original transaction data processing model is composed of multiple networks;

[0174] The model training module 830 is used to iteratively train the original transaction data processing model using the training data set and the training label set to update the network parameters of some specified networks among the multiple networks to obtain a trained transaction data processing model; wherein the some specified networks include a time series data encoding network.

[0175] In an exemplary embodiment of the present disclosure, the item description text includes at least one of the following:

[0176] The basic information of each item, the transaction market information of each item in the historical period, and the characteristic description information corresponding to the historical transaction time series data.

[0177] In an exemplary embodiment of the present disclosure, after acquiring an object dataset of multiple objects in a historical period, the data acquisition module 810 is configured to:

[0178] Preprocessing the item data set of the plurality of items in the historical period to obtain the preprocessed item data set;

[0179] The preprocessing includes at least one of the following: outlier removal processing, missing value filling processing and data standardization processing.

[0180] In an exemplary embodiment of the present disclosure, the data acquisition module 810 divides the object data set into a training data set and a training label set according to time segmentation points, including:

[0181] Dividing the historical period into a first period and a second period according to the time segmentation point to obtain a training data set corresponding to the first period and a training label set corresponding to the second period;

[0182] The first time period starts at the start time of the historical time period and ends at the time division point; the training data set includes the first item description text of each item corresponding to the first time period and the first transaction time series data of each item in the first time period;

[0183] The second time period starts at the time division point and ends at the end time of the historical time period; the training label set includes the second item description text of each item corresponding to the second time period and the second transaction time series data of each item in the second time period.

[0184] In an exemplary embodiment of the present disclosure, the model training module 830 uses the training data set and the training label set to iteratively train the original transaction data processing model to update network parameters of a portion of specified networks in the multiple networks, including:

[0185] Inputting the training sample set into the original transaction data processing model, and outputting the predicted transaction time series data of each item corresponding to the second time period through the original transaction data processing model;

[0186] Determining a first loss value of the original transaction data processing model according to a difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period;

[0187] The network parameters of a portion of the specified networks among the multiple networks are updated according to the first loss value.

[0188] In an exemplary embodiment of the present disclosure, the plurality of networks include a text information encoding network, a time series data encoding network, and a decoding network;

[0189] The model training module 830 inputs the training sample set into the original transaction data processing model, and outputs the predicted transaction time series data corresponding to the second time period for each item through the original transaction data processing model, including:

[0190] Encoding the first item description text corresponding to the first time period for each item through the text information encoding network to obtain a word embedding vector corresponding to each item;

[0191] Encode the first transaction time series data of each item in the first time period through the time series data encoding network to obtain a time series embedding vector corresponding to each item;

[0192] Concatenate the word embedding vector and the time series embedding vector corresponding to each item to obtain a concatenated vector;

[0193] The concatenated vector corresponding to each item is decoded through the decoding network to obtain predicted transaction time series data of each item corresponding to the second time period.

[0194] In an exemplary embodiment of the present disclosure, the model training module 830 encodes the first item description text corresponding to the first time period of each item through the text information encoding network to obtain the word embedding vector corresponding to each item, including:

[0195] Performing word segmentation processing on the first item description text corresponding to the first time period of each item through the text information encoding network to obtain multiple target word segments;

[0196] The standard word vector corresponding to each target word segment is matched from the standard word list to obtain the word embedding vector corresponding to each item.

[0197] In an exemplary embodiment of the present disclosure, the model training module 830 encodes the first transaction time series data of each item in the first time period through the time series data encoding network to obtain the time series embedding vector corresponding to each item, including:

[0198] Splitting the first transaction time series data of each item in the first time period by a splitting function to obtain a plurality of subsequences;

[0199] Perform sequence encoding on each of the subsequences by using an embedding function to obtain each subvector;

[0200] According to the multiple sub-vectors corresponding to the multiple sub-sequences, a temporal embedding vector corresponding to each item is determined.

[0201] In an exemplary embodiment of the present disclosure, after obtaining the predicted transaction time series data of each item corresponding to the second time period, the model training module 830 is configured to:

[0202] Determining a first loss value of the original transaction data processing model according to a difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period;

[0203] Determining a second loss value of the original transaction data processing model according to a degree of similarity between each sub-vector associated with each item and each standard word vector in a standard word list;

[0204] Determining a comprehensive loss value of the original transaction data processing model according to the first loss value and the second loss value;

[0205] The network parameters of a portion of the plurality of networks are updated according to the comprehensive loss value.

[0206] In an exemplary embodiment of the present disclosure, the partially specified network further includes a decoding network.

[0207] In an exemplary embodiment of the present disclosure, after obtaining a trained transaction data processing model, a model application module (not shown in the figure) is configured to:

[0208] Inputting the item data set of the plurality of items in the historical period into the trained transaction data processing model;

[0209] The trained transaction data processing model is used to predict future transaction time series data of each item in a future time period based on the item data set of the multiple items in a historical time period.

[0210] The specific details of each module in the training device of the above-mentioned transaction data processing model have been described in detail in the corresponding training method of the transaction data processing model, so they will not be repeated here.

[0211] It should be noted that, although several modules or units of the device for action execution are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0212] In addition, although the steps of the method in the present disclosure are described in a specific order in the drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps, etc.

[0213] Through the description of the above implementation, it is easy for those skilled in the art to understand that the example implementation described here can be implemented by software, or by software combined with necessary hardware. Therefore, the technical solution according to the implementation of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to execute the method according to the implementation of the present disclosure.

[0214] The present disclosure also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or may exist independently without being assembled into the electronic device.

[0215] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0216] Computer-readable storage media can send, propagate or transmit programs for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0217] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by an electronic device, the electronic device implements the method described in the above embodiments.

[0218] In addition, an electronic device capable of implementing the above method is also provided in an embodiment of the present disclosure.

[0219] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0220] Refer to the following Fig. 9 The electronic device 900 according to this embodiment of the present disclosure is described. Fig. 9 The electronic device 900 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0221] like Fig. 9 As shown, the electronic device 900 is in the form of a general computing device. The components of the electronic device 900 may include, but are not limited to: at least one processor 910, at least one memory 920, a bus 930 connecting different system components (including the memory 920 and the processor 910), and a display 940.

[0222] The memory stores program codes, which can be executed by the processor 910, so that the processor 910 performs the steps according to various exemplary embodiments of the present disclosure described in the above “Exemplary Method” section of this specification. For example, the processor 910 can perform the following steps: Figure 1 As shown in: Step S110, obtaining an item data set of multiple items in a historical period, and dividing the item data set into a training data set and a training label set according to a time segmentation point; the item data set includes an item description text corresponding to each item in the historical period and historical transaction time series data of each item in the historical period; Step S120, calling an original transaction data processing model; the original transaction data processing model is composed of multiple networks; Step S130, iteratively training the original transaction data processing model using the training data set and the training label set to update network parameters of some specified networks among the multiple networks to obtain a trained transaction data processing model; wherein the some specified networks include a time series data encoding network.

[0223] The memory 920 may include a readable medium in the form of a volatile storage, such as a random access memory (RAM) 9201 and / or a cache memory 9202 , and may further include a read-only memory (ROM) 9203 .

[0224] The memory 920 may also include a program / utility 9204 having a set (at least one) of program modules 9205, such program modules 9205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0225] Bus 930 may be a bus representing one or more of several types of bus structures, including a memory bus or memory controller, a peripheral bus, an accelerated graphics port, a processor, or a local bus using any of a variety of bus architectures.

[0226] The electronic device 900 may also communicate with one or more external devices 1000 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 900, and / or communicate with any device that enables the electronic device 900 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 950. In addition, the electronic device 900 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 960. As shown, the network adapter 960 communicates with other modules of the electronic device 900 via a bus 930. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 900, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0227] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary technical means in the art that are not disclosed in the present disclosure. The description and examples are to be considered exemplary only, and the true scope and spirit of the present disclosure are indicated by the claims.

Claims

1. A training method for a transaction data processing model, characterized in that: include: Acquire an item data set of multiple items in a historical period, and divide the item data set into a training data set and a training label set according to a time segmentation point; the item data set includes an item description text of each item corresponding to the historical period and historical transaction time series data of each item in the historical period; Calling an original transaction data processing model; the original transaction data processing model is composed of multiple networks; The original transaction data processing model is iteratively trained using the training data set and the training label set to update network parameters of some specified networks among the multiple networks to obtain a trained transaction data processing model; wherein the some specified networks include a time series data encoding network.

2. The method according to claim 1, characterized in that: The item description text includes at least one of the following: The basic information of each item, the transaction market information of each item in the historical period, and the characteristic description information corresponding to the historical transaction time series data.

3. The method according to claim 2, characterized in that After obtaining the object data set of the plurality of objects in the historical period, the method further includes: Preprocessing the item data set of the plurality of items in the historical period to obtain the preprocessed item data set; The preprocessing includes at least one of the following: outlier removal processing, missing value filling processing and data standardization processing.

4. The method according to claim 1 or 3, characterized in that: The step of dividing the item data set into a training data set and a training label set according to the time segmentation points includes: Dividing the historical period into a first period and a second period according to the time segmentation point to obtain a training data set corresponding to the first period and a training label set corresponding to the second period; The first time period starts at the start time of the historical time period and ends at the time division point; the training data set includes the first item description text of each item corresponding to the first time period and the first transaction time series data of each item in the first time period; The second time period starts at the time division point and ends at the end time of the historical time period; the training label set includes the second item description text of each item corresponding to the second time period and the second transaction time series data of each item in the second time period.

5. The method according to claim 4, characterized in that The iterative training of the original transaction data processing model using the training data set and the training label set to update network parameters of a portion of designated networks in the multiple networks includes: Inputting the training sample set into the original transaction data processing model, and outputting the predicted transaction time series data of each item corresponding to the second time period through the original transaction data processing model; Determining a first loss value of the original transaction data processing model according to a difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period; The network parameters of a portion of the specified networks among the multiple networks are updated according to the first loss value.

6. The method according to claim 5, characterized in that The multiple networks include a text information encoding network, a time series data encoding network and a decoding network; The step of inputting the training sample set into the original transaction data processing model and outputting the predicted transaction time series data of each item corresponding to the second time period through the original transaction data processing model includes: Encoding the first item description text corresponding to the first time period for each item through the text information encoding network to obtain a word embedding vector corresponding to each item; Encode the first transaction time series data of each item in the first time period through the time series data encoding network to obtain a time series embedding vector corresponding to each item; Concatenate the word embedding vector and the time series embedding vector corresponding to each item to obtain a concatenated vector; The concatenated vector corresponding to each item is decoded through the decoding network to obtain predicted transaction time series data of each item corresponding to the second time period.

7. The method according to claim 6, characterized in that The step of encoding the first item description text corresponding to the first time period for each item through the text information encoding network to obtain a word embedding vector corresponding to each item includes: Performing word segmentation processing on the first item description text corresponding to the first time period of each item through the text information encoding network to obtain multiple target word segments; The standard word vector corresponding to each target word segment is matched from the standard word list to obtain the word embedding vector corresponding to each item.

8. The method according to claim 6, characterized in that The encoding of the first transaction time series data of each item in the first time period by the time series data encoding network to obtain a time series embedding vector corresponding to each item includes: Splitting the first transaction time series data of each item in the first time period by a splitting function to obtain a plurality of subsequences; Perform sequence encoding on each of the subsequences by using an embedding function to obtain each subvector; According to the multiple sub-vectors corresponding to the multiple sub-sequences, a temporal embedding vector corresponding to each item is determined.

9. The method according to claim 8, characterized in that After obtaining the predicted transaction time series data corresponding to the second time period for each item, the method further includes: Determining a first loss value of the original transaction data processing model according to a difference between the predicted transaction time series data of each item corresponding to the second time period and the second transaction time series data of each item in the second time period; Determining a second loss value of the original transaction data processing model according to a degree of similarity between each sub-vector associated with each item and each standard word vector in the standard word list; Determining a comprehensive loss value of the original transaction data processing model according to the first loss value and the second loss value; The network parameters of a portion of the designated networks among the plurality of networks are updated according to the comprehensive loss value.

10. The method according to claim 5 or 9, characterized in that: The partially specified network also includes a decoding network.

11. The method according to claim 1, characterized in that: After obtaining the trained transaction data processing model, the method further includes: Inputting the item data set of the plurality of items in the historical period into the trained transaction data processing model; The trained transaction data processing model is used to predict future transaction time series data of each item in a future time period based on the item data set of the multiple items in a historical time period.

12. A training device for a transaction data processing model, characterized in that: include: A data acquisition module is used to acquire an item data set of multiple items in a historical period, and divide the item data set into a training data set and a training label set according to a time segmentation point; the item data set includes an item description text of each item corresponding to the historical period and a historical transaction time series data of each item in the historical period; A model calling module, used to call an original transaction data processing model; the original transaction data processing model is composed of multiple networks; A model training module is used to iteratively train the original transaction data processing model using the training data set and the training label set to update the network parameters of some specified networks among the multiple networks to obtain a trained transaction data processing model; wherein the some specified networks include a time series data encoding network.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the training method of the transaction data processing model described in any one of claims 1 to 11 is implemented.

14. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; Wherein, the processor is configured to execute the training method of the transaction data processing model described in any one of claims 1 to 11 by executing the executable instructions.

Images