Entry classification method and device, electronic equipment and storage medium
Through the entry classification model trained by the deep learning model, the entry classification in the enterprise information list business is automatically processed, solving the problems of time-consuming and low accuracy in the existing technology, and achieving efficient and accurate entry classification.
Patent Information
- Application Number
- CN202510643150.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-19
- Publication Date
- 2025-08-26
AI Technical Summary
In the enterprise information sheet business, entry classification takes time and is prone to errors. The existing classification methods based on data dictionary cannot adapt to entry changes, resulting in low classification accuracy and high maintenance costs.
The entry classification model trained by the deep learning model is used to obtain the entries in the table to be classified, perform feature mining and fusion, and automatically determine the entry type and add it to the table to reduce manual participation.
It improves the accuracy of item classification, shortens classification time, improves user experience, reduces labor costs, and ensures the accuracy and intuitiveness of classification results.
Smart Images
Figure CN120541572A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and in particular to an entry classification method, device, electronic device and storage medium. Background Art
[0002] The enterprise information sheet service provides electronic information sheet distribution services to employees. This service not only simplifies the issuance process but also improves internal information management efficiency, providing employees with a more convenient and efficient way to access information. The electronic information sheet service primarily features two functions: information sheet format definition and timesheet upload. Information sheet format definition determines the items in each information sheet uploaded by the enterprise. Common information sheet items include payable data, deduction data, and basic data. Timesheet upload allows users to upload specific timesheet files based on the information sheet format definition. To meet regulatory requirements, clients must specify the categories corresponding to the information sheet items during the information sheet format definition process. For large enterprises with a large number of information sheet items, categorizing the items during the information sheet format definition process can be time-consuming for operators. For example, a large corporate client may have over 100 information sheet items on a daily basis, and categorizing them takes an average of four minutes. This extensive categorization process can be tedious for clients, leading to haphazard categorization and errors.
[0003] Existing solutions maintain common entries in a data dictionary and maintain corresponding categories for the data dictionary. When classifying an entry, the data dictionary is first searched to see if the content is included. If it is included, it is classified according to the categories maintained in the data dictionary. If it is not in the data dictionary, it cannot be classified. Solutions that classify entries based on data dictionaries are only suitable for entries that remain fixed for a long time. In reality, the entries in a company's information sheets are constantly changing. If entry classification is based on a data dictionary, new entries cannot be matched to the data dictionary, and the classification accuracy is low. The core of the solution for classifying entries based on a data dictionary is an accurate data dictionary. Dedicated personnel need to be assigned to regularly maintain the content and corresponding categories in the data dictionary. Due to the large number of entries, the maintenance cost of the data dictionary is also high. Summary of the Invention
[0004] The present invention provides an entry classification method, device, electronic device and storage medium to automatically classify entries, improve classification accuracy and enhance user experience.
[0005] According to one aspect of the present invention, there is provided an entry classification method, comprising:
[0006] Acquire a table to be classified and at least one entry to be classified in the table to be classified, wherein the entry to be classified indicates an entry in the table to be classified that is related to the enterprise information sheet business;
[0007] Inputting the table to be classified into an item classification model to obtain an item type of each item to be classified;
[0008] The entry type corresponding to each of the to-be-classified entries is added to the to-be-classified table.
[0009] According to another aspect of the present invention, there is provided an entry classification device, comprising:
[0010] An acquisition module, configured to acquire a table to be classified and at least one item to be classified in the table to be classified, wherein the item to be classified indicates an item in the table to be classified that is related to the enterprise information sheet business;
[0011] A classification module, configured to input the table to be classified into an item classification model to obtain an item type of each item to be classified;
[0012] The adding module is used to add the entry type corresponding to each of the to-be-classified entries to the to-be-classified table.
[0013] According to another aspect of the present invention, an electronic device is provided, comprising:
[0014] at least one processor; and
[0015] a memory communicatively connected to the at least one processor; wherein,
[0016] The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor so that the at least one processor can perform the item classification method according to any embodiment of the present invention.
[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the item classification method according to any embodiment of the present invention when executed.
[0018] The technical solution of an embodiment of the present invention obtains a table to be classified and at least one item to be classified in the table to be classified, inputs the table to be classified into an item classification model, obtains the item type of each item to be classified, and adds the item type corresponding to each item to be classified to the table to be classified. By obtaining the item type of each item to be classified through the item classification model, the operation of automatically classifying the items to be classified is realized. The classification method of the item classification model shortens the time for manual participation, thereby reducing the occurrence of classification errors and improving the user experience. Finally, the item type is added to the table to be classified, ensuring that the user can clearly and intuitively see the item type of each item to be classified, effectively meeting the user's needs.
[0019] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0021] Figure 1 This is a flowchart of an entry classification method provided according to the first embodiment of the present invention;
[0022] Figure 2 This is a flow chart of a classification processing method provided according to the first embodiment of the present invention;
[0023] Figure 3 This is a flowchart of a method for determining an entry type provided in accordance with a second embodiment of the present invention;
[0024] Figure 4 This is a schematic diagram of the structure of an entry classification model provided according to the second embodiment of the present invention;
[0025] Figure 5 This is a schematic diagram of the structure of an item classification device provided according to the third embodiment of the present invention;
[0026] Figure 6 This is a block diagram of an electronic device provided according to a fourth embodiment of the present invention. DETAILED DESCRIPTION
[0027] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.
[0028] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0029] Example 1
[0030] Figure 1 This is a flowchart of an item classification method provided according to the first embodiment of the present invention. This embodiment is applicable to the case of classifying items in an enterprise information sheet business. The method can be executed by an item classification device. The item classification device can be implemented in the form of hardware and / or software. The item classification device can be configured in an electronic device, which can be a mobile terminal, PC or server, etc. Figure 1 As shown, the method includes:
[0031] S110 : Acquire a table to be classified and at least one item to be classified in the table to be classified, where the item to be classified indicates an item in the table to be classified that is related to the enterprise information sheet business.
[0032] In this embodiment, the "to-be-classified table" can be understood as an electronic form provided to users by an enterprise through its online service. The "to-be-classified table" can store data related to enterprise information sheets. "to-be-classified items" can be understood as items of data stored in the "to-be-classified table." "to-be-classified items" can indicate the composition of an enterprise information sheet and can be stored in a row or column of the "to-be-classified table." The enterprise information sheet service can be understood as a service provided by an enterprise through its online service, whereby an enterprise distributes information sheets to users.
[0033] Specifically, when a user initiates an item classification request, the system first retrieves the information sheet to be classified. It then determines whether the user has previously uploaded a template with the same elements. If so, the existing template is directly used to classify the items in the information sheet. If not, the user's input information sheet is used as the table to be classified, and at least one item to be classified stored in the table is retrieved.
[0034] For example, a table to be classified may include more than 100 items to be classified, and the items to be classified may include basic values, traffic values, subsidy values, grant values, payable values, deductible values and actual paid values in the enterprise information sheet business.
[0035] S120: Input the table to be classified into an item classification model to obtain an item type of each item to be classified.
[0036] In this embodiment, the item classification model can be understood as a model used to output the item type to which the item to be classified belongs. The item classification model can be trained using historical data based on a deep learning model. The historical data can be data related to the enterprise information sheet business and corresponding to items of known types. The item type can be understood as the type to which the item to be classified belongs. The item type can be a standard type established for the enterprise information sheet business.
[0037] Specifically, before obtaining the table to be classified, an item classification model is trained using historical data based on a deep learning model. The table to be classified is then fed into the trained item classification model, which then performs operations such as feature mining on each item to be classified, ultimately outputting the item type corresponding to each item to be classified. Item types can include basic information items, payable items, deductible items, and other information items.
[0038] For example, in an enterprise information sheet service, basic information items may include an item type indicating the user's identity and the time the corresponding information was issued, within the items to be classified. Items to be issued may include an item type indicating the amount the user should receive before any deductions are made. Items to be deducted may include an item type indicating the various amounts to be deducted from the amount the user should receive. Other information items may include supplementary items in addition to the three aforementioned item types.
[0039] S130: Add the entry type corresponding to each of the to-be-classified entries to the to-be-classified table.
[0040] Specifically, the four entry types are encoded to obtain a serial number corresponding to each entry type. After determining the entry type of each entry to be classified in the table to be classified, the serial number of the entry type corresponding to the entry to be classified is added to the table to be classified.
[0041] For example, the serial number of an item whose type is a basic information item is coded as A, the serial number of an item whose type is a payable item is coded as B, the serial number of an item whose type is a deductible item is coded as C, and the serial number of an item whose type is other information item is coded as D. Assuming that the item type of an item to be classified is a basic information item, the serial number A corresponding to the basic information item is added to the position corresponding to the item to be classified in the table to be classified.
[0042] The technical solution of an embodiment of the present invention obtains a table to be classified and at least one item to be classified in the table to be classified, inputs the table to be classified into an item classification model, obtains the item type of each item to be classified, and adds the item type corresponding to each item to be classified to the table to be classified. By obtaining the item type of each item to be classified through the item classification model, the operation of automatically classifying the items to be classified is realized. The classification method of the item classification model shortens the time for manual participation, thereby reducing the occurrence of classification errors and improving the user experience. Finally, the item type is added to the table to be classified, ensuring that the user can clearly and intuitively see the item type of each item to be classified, effectively meeting the user's needs.
[0043] Based on the above embodiment, a modified embodiment of the above embodiment is proposed. It should be noted that, in order to simplify the description, only the differences from the above embodiment are described in the modified embodiment.
[0044] In one embodiment, the training operation of the item classification model includes:
[0045] Obtaining a history table and the historical data in the history table;
[0046] Cleaning the historical data and performing statistics on the cleaned historical data to obtain training data, wherein the training data includes historical entry data corresponding to historical entries and types of the historical entries, wherein the historical entries indicate entries in the historical table that are related to the enterprise information sheet business;
[0047] An initial item classification model is trained based on the training data to obtain a trained item classification model.
[0048] In this embodiment, historical data related to enterprise information sheets can be stored in the history table. The history table can be understood as a stored record of past information sheets, and the type of each historical entry in the history table is known. Historical entries can be understood as entries used in the item classification model training process, and historical entries indicate items of data stored in the history table. Historical data can be understood as data stored in the history table, and historical data can include data indicating items of historical entries in the history table, as well as data indicating the type of historical entries. Training data can be understood as data used for item classification model training, and training data can be data obtained after cleaning, statistics, and other operations on historical data. The initial item classification model can be understood as a deep learning model based on the attention mechanism (Transformer architecture), and the Transformer architecture is mainly composed of an encoder (Encoder) and a decoder (Decoder).
[0049] Specifically, obtain the historical table where the historical entries of known types are located, and take the data in the historical table indicating the data related to the historical entries and the data indicating the type of the historical entries as the historical data. Clean the historical data, including deleting abnormal values in the historical data, removing abnormal characters such as spaces, etc., to obtain the cleaned historical data. Then, count the cleaned historical data, including counting the number of historical entries, the number of types of historical entries, etc., to obtain training data. Therefore, the training data includes the historical entry data corresponding to the historical entries, and the data indicating the type of the historical entries. Finally, train the initial entry classification model based on the training data to obtain the entry classification model. The historical entry data in the training data can be input into the initial entry classification model to obtain the type output by the initial entry classification model. The type of the known historical entry is used as the target, and the parameters of the initial entry classification model are adjusted using the cross entropy loss function to obtain the entry classification model.
[0050] Exemplarily, after the item classification model is trained, the trained item classification model can be deployed to run in multiple environments, such as: a multi-tenant environment, multiple servers, and / or multiple running terminals. Multiple environments can realize the isolated operation of the item classification model and achieve high concurrent calls. At the same time, recent historical tables will be synchronized regularly in multiple environments to regularly adjust the parameters of the item classification model and realize automatic model updates. During the operation of the item classification model, the operation of the item classification model can be dynamically monitored. The monitoring method can be: when the user uses the item classification model for classification, the item type obtained is buried to record the user's evaluation index of the item type, and at the same time, the data corresponding to the item type that the user believes the item classification model has misidentified is recorded, so as to upgrade and optimize the item classification model.
[0051] In one embodiment, after adding the entry type corresponding to each of the to-be-classified entries to the to-be-classified table, the method further includes:
[0052] For the entry type corresponding to each entry to be classified in the table to be classified, in response to the verification operation of the entry type on the operation interface, the verification result corresponding to the entry type is obtained. When the verification result indicates that the entry type is incorrect, in response to the modification operation of the entry type on the operation interface, the modified entry type corresponding to the entry to be classified is obtained, and the entry type corresponding to the entry to be classified is updated to the modified entry type.
[0053] In this embodiment, the operation interface can be understood as a visual interface for user operations and a medium for human-computer interaction. Users can verify and modify item types on the operation interface. For an item to be classified, the verification result can be understood as the user's judgment on whether the item type output by the item classification model is correct. If the user believes that the item type output by the item classification model is not the type of the item to be classified, the user can modify the type of the item to be classified on the operation page. The modified item type can be understood as the item type obtained after performing the modification operation on the item to be classified.
[0054] Specifically, after adding the corresponding entry type for the entry to be classified in the table to be classified, the content in the table to be classified is displayed on the operation interface. For the entry type corresponding to each entry to be classified, the user's verification operation on the entry type in the operation interface is received, and the verification result of the entry type corresponding to each entry to be classified is obtained. For one of the entries to be classified: when the verification result indicates that the entry type is correct, it means that the user believes that the entry type output by the entry classification model is correct, so the entry type corresponding to the entry to be classified can be directly output; when the verification result indicates that the entry type is incorrect, the user's modification operation on the entry type in the operation interface is received, and the modified entry type is obtained. The modified entry type is the entry type of the entry to be classified that the user believes, and the entry type corresponding to the entry to be classified is updated to the modified entry type.
[0055] For example, Figure 2 This is a flow chart of a classification processing method provided according to the first embodiment of the present invention. Figure 2 As shown, the entry type is displayed to the user on the operation interface, and the user's verification result is received. If the verification result is correct, the process is directly ended; otherwise, the user's modification operation is received, and the entry type is modified to the modified entry type.
[0056] In one embodiment, obtaining the table to be classified includes:
[0057] In response to a request operation on a request interface, a request message is obtained, wherein the request message includes a standard table to be classified;
[0058] The standard table to be classified in the request message is parsed to obtain a table to be classified.
[0059] In this embodiment, the request interface can be understood as an interface that can receive requests. The request interface defines communication and call specifications. The specifications specified by the request interface can be used to assemble a request message. The request message can be understood as a collection of specific request content, including a standard classification table for items to be classified. The format of the standard classification table matches the format required by the request interface.
[0060] Specifically, users can organize the items to be classified into a standard interface format, that is, a standard table to be classified, and then send a request to the system. After the request is received, the system will parse the request message and obtain the table to be classified.
[0061] For example, Figure 2 As shown, the local file is the item that needs to be classified, and the corresponding information sheet template is generated for it. Then, it is determined whether the user has previously uploaded a template with the same elements. The elements can include the items to be classified in the table to be classified, that is, it is determined whether the user has previously performed classification operations on the same table to be classified. If so, the uploaded template is directly called to directly classify the items to be classified in the table to be classified. If not, the decision engine interface is called for classification, that is, a request is sent to the system. After the request is received, the system parses the request message to obtain the table to be classified, and then classifies the table to be classified through the item classification model.
[0062] Example 2
[0063] Figure 3 This is a flow chart of a method for determining an entry type according to a second embodiment of the present invention. This embodiment is based on the method for determining the entry type of an entry to be classified according to the above embodiment. Figure 3 As shown, the method includes:
[0064] S210: Acquire a table to be classified and at least one entry to be classified in the table to be classified.
[0065] By using the item classification model, the following steps S220-S250 are performed for each item to be classified in the table to be classified:
[0066] S220: Determine the text-side classification data and the object-side classification data corresponding to the entry to be classified.
[0067] The text-side classification data indicates text information of the item to be classified, and the object-side classification data indicates object information corresponding to the item to be classified.
[0068] In this embodiment, the text-side classification data can be understood as data indicating the text information of the item to be classified, such as the name of the item to be classified. The object-side classification data can be understood as data indicating the object information of the item to be classified, such as data related to the enterprise information corresponding to the item to be classified.
[0069] For example, text-side classification data can be data corresponding to the name of the item to be classified, such as basic value, traffic value, subsidy value, grant value, payable value, deductible value, and actual value. Object-side classification data can be data corresponding to the company information to which the item to be classified belongs, such as company size, user type, registration time, and registration value.
[0070] S230: Extracting text-side features corresponding to the text-side classification data and object-side features corresponding to the object-side classification data.
[0071] In this embodiment, text-side features can be understood as features obtained by mining features of text-side classification data, and the mining method can be a feature mining method based on the attention mechanism. Object-side features can be understood as features obtained by mining features of object-side classification data, and the mining method can be a feature mining method based on matrix transformation.
[0072] Specifically, the text-side classification data is processed based on the attention mechanism to obtain text-side features. The attention mechanism can be implemented based on an encoder and a decoder. The encoder is responsible for converting the text-side classification data into an intermediate representation, while the decoder generates input text-side features based on this intermediate representation. The attention mechanism allows the model to focus on information at different locations when processing text-side classification data, while the multi-head attention mechanism divides the text-side classification data into multiple different parts, performs attention calculations on each part, and then merges the results to obtain classification-side features. The object-side classification data is processed based on matrix transformation to obtain object-side features. Matrix transformation mainly builds the sequence matrix corresponding to the object-side classification data and transforms the matrix to obtain object-side features.
[0073] Optionally, extracting text-side features corresponding to the text-side classification data includes:
[0074] Splitting the text-side classification data according to a splitting order to obtain at least one unit constituting the text-side classification data, wherein the splitting order indicates an order of the units in the text-side classification data;
[0075] Encoding each of the units to obtain a unit code corresponding to each of the units;
[0076] The unit codes are combined in the splitting order to obtain the text-side features corresponding to the text-side classification data.
[0077] In this embodiment, the splitting order can be understood as the order in which the text-side classification data is split into smaller units according to a rule. The splitting order can be the positional order of the units that make up the text-side classification data in the text-side classification data. The units that make up the text-side classification data can be individual characters. The unit code can be understood as the sequence obtained by binary encoding the units that make up the text-side classification data.
[0078] Specifically, the text-side classification data is split, and all units that make up the text-side classification data are obtained, along with the order in which the data is split. Each unit is encoded using a multi-head attention mechanism to obtain the unit code corresponding to each unit. All unit codes are then combined according to the split order to obtain the text-side features corresponding to the text-side classification data.
[0079] For example, Figure 4 Schematic diagram of the structure of an entry classification model provided according to the second embodiment of the present invention. Figure 4 As shown, the left tower is the process of extracting the text-side features corresponding to the text-side classification data. For the text-side classification data, it is first split word by word to obtain multiple units. For each unit, the unit code of each unit can be obtained by positional encoding (PositionalEncoder) and encoder (Transformer Encoder) using binary encoding. Then, the unit codes are combined in the order of splitting, and a fully connected hierarchy is built based on the combined sequence to obtain the text-side features. The parameters of the fully connected hierarchy can be set to *2:12*32.
[0080] Optionally, extracting object-side features corresponding to the object-side classification data includes:
[0081] Classifying the object-side classification data to obtain discrete data and continuous data;
[0082] Determining a discrete matrix corresponding to the discrete data according to the discrete data;
[0083] The discrete matrix and the continuous data are concatenated to obtain object-side features corresponding to the object-side classification data.
[0084] In this embodiment, discrete data can be understood as data indicating semantic information of an enterprise in the object-side categorical data, and continuous data can be understood as data indicating numerical information of an enterprise in the object-side categorical data.
[0085] For example, according to the different information related to the enterprise indicated in the object-side classification data, the object-side classification data is classified to obtain discrete data and continuous data. Figure 4 As shown, the right tower is the process of extracting object-side features corresponding to object-side classification data. Discrete data can include data related to enterprise size, user type, etc. Continuous data can include data related to the registration time, registration value, etc. of the enterprise. Build a text sequence matrix for discrete data, and transform the matrix to obtain a discrete matrix corresponding to the discrete data. Use the fully connected hierarchy (for example: linear rectified unit activation) to splice the discrete matrix and continuous data to obtain the object-side features corresponding to the object-side classification data. Among them, the splicing process can go through three layers of fully connected hierarchy, and its parameters can be set to 256, 64, and 16 respectively.
[0086] S240: Fusing the text-side features and the object-side features to obtain fused features.
[0087] In this embodiment, the fused feature can be understood as a feature obtained by fusing the text-side feature and the object-side feature according to the requirements of the item classification model. The fused feature can indicate the complete information of the item to be classified.
[0088] Specifically, when the text-side features and the object-side features are fused, in order to avoid overfitting caused by too deep mining of enterprise-related features as much as possible, the fusion method adopts a method in which the final output dimension of the right tower is much smaller than that of the left tower to obtain the fused features.
[0089] For example, Figure 4 As shown, the features corresponding to the left tower are text-side features, while the features corresponding to the right tower are object-side features. The fully connected layer parameters of the final output of the left tower are *2:12*32, while the fully connected layer parameters of the final output of the right tower are 16. It can be seen that the final output dimension of the right tower is much smaller than that of the left tower.
[0090] S250: Determine the entry type of the entry to be classified according to the fusion feature.
[0091] For example, the fused features are passed through three layers of fully connected hierarchies in the item classification model. The parameters of the three layers of fully connected hierarchies can be set to 256, 64, and 16, respectively. After processing through the three layers of fully connected hierarchies, the item type of the item to be classified is obtained. The parameters of the fully connected hierarchies can be set in advance according to the classification requirements.
[0092] S260: Add the entry type corresponding to each of the to-be-classified entries to the to-be-classified table.
[0093] The technical solution of an embodiment of the present invention utilizes an item classification model to perform the following operations for each item to be classified in a table to be classified: determining text-side classification data and object-side classification data corresponding to the item to be classified, wherein the text-side classification data indicates the text information of the item to be classified, and the object-side classification data indicates the object information corresponding to the item to be classified; extracting text-side features corresponding to the text-side classification data and object-side features corresponding to the object-side classification data; fusing the text-side features and the object-side features to obtain fused features; and determining the item type of the item to be classified based on the fused features. The method for determining the item type is refined by processing the items to be classified within the item classification model. Because the items to be classified include data indicating different types of corporate information, the items to be classified are divided into text-side classification data and object-side classification data to ensure accurate item classification when different users have different categories for the same item. By extracting features separately and determining the item type of the item to be classified based on the fused features, the personalized needs of users are met. The operation of the item classification model effectively reduces labor costs and improves classification accuracy.
[0094] Example 3
[0095] Figure 5 This is a structural diagram of an item classification device provided by the third embodiment of the present invention. Figure 5 As shown, the device includes:
[0096] The acquisition module 310 is configured to acquire a table to be classified and at least one item to be classified in the table to be classified, wherein the item to be classified indicates an item in the table to be classified that is related to the enterprise information sheet business.
[0097] The classification module 320 is configured to input the table to be classified into an item classification model to obtain an item type of each item to be classified.
[0098] The adding module 330 is configured to add the entry type corresponding to each of the to-be-classified entries to the to-be-classified table.
[0099] The item classification device provided by the embodiment of the present invention obtains a table to be classified and at least one item to be classified in the table to be classified through an acquisition module, inputs the table to be classified into an item classification model through a classification module, obtains the item type of each item to be classified, and adds the item type corresponding to each item to be classified to the table to be classified through an addition module. Through the mutual cooperation between the modules, the item type of each item to be classified is obtained through the item classification model, and the operation of automatically classifying the items to be classified is realized. The classification method of the item classification model shortens the time for manual participation, thereby reducing the situation of classification errors and improving the user experience. Finally, the item type is added to the table to be classified to ensure that the user can clearly and intuitively see the item type of each item to be classified, effectively meeting the needs of the user.
[0100] In one embodiment, the classification model 320 includes:
[0101] The following operations are performed for each entry to be classified in the table to be classified using the entry classification model:
[0102] a first determining unit, configured to determine text-side classification data and object-side classification data corresponding to the item to be classified, wherein the text-side classification data indicates text information of the item to be classified, and the object-side classification data indicates object information corresponding to the item to be classified;
[0103] an extraction unit, configured to extract text-side features corresponding to the text-side classification data and object-side features corresponding to the object-side classification data;
[0104] a fusion unit, configured to fuse the text-side features and the object-side features to obtain fused features;
[0105] The second determining unit is configured to determine the entry type of the entry to be classified according to the fusion feature.
[0106] In one embodiment, the extraction unit is specifically configured to:
[0107] Splitting the text-side classification data according to a splitting order to obtain at least one unit constituting the text-side classification data, wherein the splitting order indicates an order of the units in the text-side classification data;
[0108] Encoding each of the units to obtain a unit code corresponding to each of the units;
[0109] The unit codes are combined in the splitting order to obtain the text-side features corresponding to the text-side classification data.
[0110] In one embodiment, the extraction unit is specifically configured to:
[0111] Classifying the object-side classification data to obtain discrete data and continuous data;
[0112] Determining a discrete matrix corresponding to the discrete data according to the discrete data;
[0113] The discrete matrix and the continuous data are concatenated to obtain object-side features corresponding to the object-side classification data.
[0114] In one embodiment, the item classification device further includes a training module, specifically configured to:
[0115] Obtaining a history table and the historical data in the history table;
[0116] Cleaning the historical data and performing statistics on the cleaned historical data to obtain training data, wherein the training data includes historical entry data corresponding to historical entries and types of the historical entries, wherein the historical entries indicate entries in the historical table that are related to the enterprise information sheet business;
[0117] An initial item classification model is trained based on the training data to obtain a trained item classification model.
[0118] In one embodiment, the entry classification device further includes an updating module, specifically configured to:
[0119] For the entry type corresponding to each entry to be classified in the table to be classified, in response to the verification operation of the entry type on the operation interface, the verification result corresponding to the entry type is obtained. When the verification result indicates that the entry type is incorrect, in response to the modification operation of the entry type on the operation interface, the modified entry type corresponding to the entry to be classified is obtained, and the entry type corresponding to the entry to be classified is updated to the modified entry type.
[0120] In one embodiment, the acquisition module is specifically configured to:
[0121] In response to a request operation on a request interface, a request message is obtained, wherein the request message includes a standard table to be classified;
[0122] The standard table to be classified in the request message is parsed to obtain a table to be classified.
[0123] The entry classification device provided in the embodiment of the present invention can execute the entry classification method provided in any embodiment of the present invention, completes the classification of entries through the mutual cooperation and collaboration between modules, and has the functional modules and beneficial effects corresponding to the execution method.
[0124] Example 4
[0125] According to an embodiment of the present invention, the present invention further provides an electronic device, a computer-readable storage medium, and a computer program product.
[0126] Figure 6 is a block diagram of an electronic device provided according to embodiment four of the present invention, which can implement the entry classification method described in the embodiment of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the invention described and / or required herein.
[0127] like Figure 6 As shown, the electronic device 410 includes at least one processor 411, and a memory connected to the at least one processor 411 in communication, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, etc., wherein the memory stores a computer program that can be executed by the at least one processor, and the processor 411 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 412 or the computer program loaded from the storage unit 418 to the random access memory (RAM) 413. Various programs and data required for the operation of the electronic device 410 can also be stored in the RAM 413. The processor 411, ROM 412 and RAM 413 are connected to each other via a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.
[0128] Multiple components in the electronic device are connected to the I / O interface 415, including an input unit 416, such as a keyboard, mouse, etc.; an output unit 417, such as various types of displays, speakers, etc.; a storage unit 418, such as a magnetic disk, optical disk, etc.; and a communication unit 419, such as a network card, modem, wireless communication transceiver, etc. The communication unit 419 allows the electronic device to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.
[0129] Processor 411 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any other suitable processor, controller, microcontroller, etc. Processor 411 executes the various methods and processes described above, such as the entry classification method.
[0130] In some embodiments, the item classification method may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 418. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 410 via ROM 412 and / or communication unit 419. When the computer program is loaded into RAM 413 and executed by processor 411, one or more steps of the item classification method described above may be performed. Alternatively, in other embodiments, processor 411 may be configured to perform the item classification method in any other suitable manner (e.g., via firmware).
[0131] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.
[0132] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.
[0133] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).
[0135] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.
[0136] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.
[0137] The technical solution of the embodiment of the present invention is through an item classification method, device, electronic device, storage medium and program product. By obtaining a table to be classified and at least one item to be classified in the table to be classified, the table to be classified is input into an item classification model to obtain the item type of each item to be classified, and the item type corresponding to each item to be classified is added to the table to be classified. The item type of each item to be classified is obtained through the item classification model, and the operation of automatically classifying the items to be classified is realized. The classification method of the item classification model shortens the time of manual participation, thereby reducing the situation of classification errors and improving the user experience. Finally, the item type is added to the table to be classified to ensure that the user can clearly and intuitively see the item type of each item to be classified, effectively meeting the needs of the user.
[0138] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.
[0139] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.
Claims
1. A method for item classification, characterized in that: include: Acquire a table to be classified and at least one entry to be classified in the table to be classified, wherein the entry to be classified indicates an entry in the table to be classified that is related to the enterprise information sheet business; Inputting the table to be classified into an item classification model to obtain an item type of each item to be classified; The entry type corresponding to each of the to-be-classified entries is added to the to-be-classified table.
2. The method according to claim 1, characterized in that The step of inputting the table to be classified into an item classification model to obtain the item type of each item to be classified includes: The following operations are performed for each entry to be classified in the table to be classified using the entry classification model: Determining text-side classification data and object-side classification data corresponding to the item to be classified, wherein the text-side classification data indicates text information of the item to be classified, and the object-side classification data indicates object information corresponding to the item to be classified; extracting text-side features corresponding to the text-side classification data and object-side features corresponding to the object-side classification data; Fusing the text-side features and the object-side features to obtain fused features; The entry type of the entry to be classified is determined according to the fusion feature.
3. The method according to claim 2, characterized in that The extracting of text-side features corresponding to the text-side classification data includes: Splitting the text-side classification data according to a splitting order to obtain at least one unit constituting the text-side classification data, wherein the splitting order indicates an order of the units in the text-side classification data; Encoding each of the units to obtain a unit code corresponding to each of the units; The unit codes are combined in the splitting order to obtain the text-side features corresponding to the text-side classification data.
4. The method according to claim 2, characterized in that The extracting the object-side features corresponding to the object-side classification data includes: Classifying the object-side classification data to obtain discrete data and continuous data; Determining a discrete matrix corresponding to the discrete data according to the discrete data; The discrete matrix and the continuous data are concatenated to obtain object-side features corresponding to the object-side classification data.
5. The method according to claim 1, characterized in that The training operation of the item classification model includes: Obtaining a history table and the historical data in the history table; Cleaning the historical data and performing statistics on the cleaned historical data to obtain training data, wherein the training data includes historical entry data corresponding to historical entries and types of the historical entries, wherein the historical entries indicate entries in the historical table that are related to the enterprise information sheet business; An initial item classification model is trained based on the training data to obtain a trained item classification model.
6. The method according to claim 1, characterized in that After adding the entry type corresponding to each of the to-be-classified entries to the to-be-classified table, the method further includes: For the entry type corresponding to each entry to be classified in the table to be classified, in response to the verification operation of the entry type on the operation interface, the verification result corresponding to the entry type is obtained. When the verification result indicates that the entry type is incorrect, in response to the modification operation of the entry type on the operation interface, the modified entry type corresponding to the entry to be classified is obtained, and the entry type corresponding to the entry to be classified is updated to the modified entry type.
7. The method according to claim 1, characterized in that The step of obtaining the table to be classified includes: In response to a request operation on a request interface, a request message is obtained, wherein the request message includes a standard table to be classified; The standard table to be classified in the request message is parsed to obtain a table to be classified.
8. An item classification device, characterized in that: include: An acquisition module, configured to acquire a table to be classified and at least one item to be classified in the table to be classified, wherein the item to be classified indicates an item in the table to be classified that is related to the enterprise information sheet business; A classification module, configured to input the table to be classified into an item classification model to obtain an item type of each item to be classified; The adding module is used to add the entry type corresponding to each of the to-be-classified entries to the to-be-classified table.
9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the item classification method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the item classification method according to any one of claims 1 to 7 when executed.