Object classification method and device, electronic equipment, storage medium and program product

Abstract

The embodiment of the invention provides an object classification method and device, electronic equipment, a storage medium and a program product. The method comprises the following steps: receiving a classification request, wherein the classification request comprises multi-modal data of a target object; extracting a plurality of target object features of the target object from the multi-modal data according to the classification request; a target database is determined, multiple mapping rules are determined from the target database, and the multiple mapping rules are obtained through model extraction; and matching the target object features through the plurality of mapping rules to obtain a target classification identifier of the target object. According to the scheme, the pre-extracted multiple mapping rules can cover the mapping relation between multiple object features and classification, classification can be rapidly carried out through the multiple mapping rules, the problem that the complexity of manual classification is high is avoided, and therefore the classification efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to an object classification method, device, electronic device, storage medium and program product. Background Art

[0002] Before using an object, in order to ensure that the object is used correctly and / or used in the correct scenario, the object is checked and a check result is obtained. The check result includes the method of using the object and / or the usage scenario, and the check result can effectively guide the use of the object.

[0003] In the related art, objects are classified and processed by manual matching to obtain the inspection results.

[0004] However, the types of objects are complex and the manual matching method is inefficient. Summary of the invention

[0005] The embodiments of the present application provide an object classification method, an apparatus, an electronic device, a storage medium, and a program product to improve classification efficiency.

[0006] In a first aspect, an embodiment of the present application provides an object classification method, comprising: receiving a classification request, the classification request comprising multimodal data of a target object; extracting multiple target object features of the target object from the multimodal data according to the classification request; determining a target database, determining multiple mapping rules from the target database, the multiple mapping rules being obtained through model extraction; matching the target object features through the multiple mapping rules to obtain a target classification identifier of the target object.

[0007] In a possible implementation, extracting multiple target object features of the target object from the multimodal data includes: determining a data type of the multimodal data, where the data type is text or image; if the data type is text, extracting the multiple target object features from the multimodal data through natural language processing technology; if the data type is an image, extracting the multiple target object features from the multimodal data through computer vision technology and the natural language processing technology.

[0008] In a possible implementation, the multiple target object features are matched through the multiple mapping rules to obtain a target classification identifier of the target object, including: determining at least one target mapping rule from the multiple mapping rules, wherein the condition field of any target mapping rule includes the multiple target object features; and matching the multiple target object features according to the target mapping rules to obtain the target classification identifier.

[0009] In one possible implementation, there are multiple targets for at least one target mapping rule; matching processing is performed on the multiple target object features according to the target mapping rules to obtain the target classification identifier, including: determining field information under the conclusion field of multiple target mapping rules to obtain multiple classification identifiers to be selected; determining multiple priorities corresponding to the multiple target mapping rules, and determining the target classification identifier from the multiple classification identifiers to be selected according to the multiple priorities; or, determining historical matching records from the target database, and determining the target classification identifier from the multiple classification identifiers to be selected according to the historical matching records.

[0010] In a possible implementation, the method further includes: determining multiple sample data and multiple label classification identifiers corresponding to multiple sample objects; extracting multiple target object features of the target object from the multiple sample data; inputting the multiple target object features and the label classification identifiers into a relational extraction model to obtain the multiple mapping rules; and storing the multiple mapping rules in a target database.

[0011] In a possible implementation, the method further includes: determining multiple information gain values ​​corresponding to the multiple mapping rules; determining the mapping rule with the largest information gain value as the root node; starting with the root node, constructing a decision tree of the multiple mapping rules based on the multiple information gain values; and storing the decision tree in the database.

[0012] In a possible implementation, the method further includes: generating a plurality of modification controls according to a plurality of mapping rules, wherein the plurality of modification controls are used to modify the plurality of mapping rules; and generating a visualization page according to the decision tree and the plurality of modification controls.

[0013] In a second aspect, an embodiment of the present application provides an object classification device, comprising: a receiving module, used to receive a classification request, the classification request comprising multimodal data of a target object; an extraction module, used to extract multiple target object features of the target object from the multimodal data according to the classification request; a determination module, used to determine a target database, and determine multiple mapping rules from the target database, wherein the multiple mapping rules are obtained through model extraction; a matching module, used to match the target object features through the multiple mapping rules to obtain a target classification identifier of the target object.

[0014] In a possible implementation, the extraction module is specifically used to determine a data type of the multimodal data, where the data type is text or image; the extraction module is specifically used to extract the multiple target object features from the multimodal data through natural language processing technology if the data type is text; the extraction module is specifically used to extract the multiple target object features from the multimodal data through computer vision technology and the natural language processing technology if the data type is image.

[0015] In a possible embodiment, the device also includes: an execution module, used to determine at least one target mapping rule from the multiple mapping rules, wherein the condition field of any target mapping rule includes the multiple target object characteristics; the execution module is also used to match the multiple target object characteristics according to the target mapping rules to obtain the target classification identifier.

[0016] In one possible implementation, the number of the at least one target mapping rule is multiple; the execution module is specifically used to determine the field information under the conclusion field of the multiple target mapping rules to obtain multiple classification identifiers to be selected; the execution module is also specifically used to determine the multiple priorities corresponding to the multiple target mapping rules, and determine the target classification identifier from the multiple classification identifiers to be selected according to the multiple priorities; or, the execution module is also specifically used to determine the historical matching records from the target database, and determine the target classification identifier from the multiple classification identifiers to be selected according to the historical matching records.

[0017] In a possible implementation, the device also includes: a generation module, used to determine multiple sample data and multiple label classification identifiers corresponding to multiple sample objects; the generation module is also used to extract multiple target object features of the target object from the multiple sample data; the generation module is also used to input the multiple target object features and the label classification identifiers into a relational extraction model to obtain the multiple mapping rules; the generation module is also used to store the multiple mapping rules in a target database.

[0018] In a possible implementation, the device also includes: a construction module, used to determine multiple information gain values ​​corresponding to the multiple mapping rules; the construction module is also used to determine the mapping rule with the largest information gain value as the root node; the construction module is also used to construct a decision tree of the multiple mapping rules starting from the root node according to the multiple information gain values; the construction module is also used to store the decision tree in the database.

[0019] In a possible implementation, the device further includes: a display module, used to generate multiple modification controls based on multiple mapping rules, and the multiple modification controls are used to modify the multiple mapping rules; the display module is also used to generate a visualization page based on the decision tree and the multiple modification controls.

[0020] In a third aspect, an embodiment of the present application provides an object classification device, including: a memory, a processor;

[0021] The memory stores computer-executable instructions;

[0022] The processor executes the computer-executable instructions stored in the memory, so that the processor executes the above first aspect and / or various possible implementations of the first aspect.

[0023] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the first aspect above and / or various possible implementations of the first aspect.

[0024] In a fifth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the above first aspect and / or various possible implementation methods of the first aspect.

[0025] The object classification method, device, electronic device, storage medium and program product provided by the embodiments of the present application include: receiving a classification request, the classification request includes multimodal data of a target object; extracting multiple target object features of the target object from the multimodal data according to the classification request; determining a target database, determining multiple mapping rules from the target database, the multiple mapping rules being obtained through model extraction; matching the target object features through the multiple mapping rules to obtain a target classification identifier of the target object. In the above scheme, the multiple mapping rules extracted in advance can cover the mapping relationship between multiple object features and classifications, and classification can be quickly performed through multiple mapping rules, avoiding the high complexity of manual classification, thereby improving classification efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0027] Figure 1 A schematic diagram of an application scenario of an object classification method provided in an embodiment of the present application;

[0028] Figure 2A flowchart of an object classification method provided in an embodiment of the present application;

[0029] Figure 3 A flowchart of an object classification method provided in an embodiment of the present application;

[0030] Figure 4 A schematic diagram of extracting object features provided in an embodiment of the present application;

[0031] Figure 5 A schematic diagram of generating a decision tree provided in an embodiment of the present application;

[0032] Figure 6 A schematic diagram of the matching process provided in the embodiment of the present application;

[0033] Figure 7 A schematic diagram of the structure of an object classification device provided in an embodiment of the present application;

[0034] Figure 8 A schematic diagram of the structure of an object classification device provided in an embodiment of the present application;

[0035] Fig. 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0036] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0037] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0038] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant countries and regions, take necessary confidentiality measures, do not violate public order and good customs, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0039] In addition, this application involves conducting big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.), and using artificial intelligence technology to make automated decisions, and making technical solutions that have a significant impact on personal rights and interests based on the results of automated decisions. The application provides users with corresponding operation entrances for them to choose to agree or reject the results of automated decisions; if the user chooses to reject, the expert decision-making process will be entered.

[0040] It should be noted that the object classification method, device, electronic device, storage medium and program product of the present application can be used in the field of artificial intelligence technology, and can also be used in any field other than artificial intelligence. The application field of the object classification method, device, electronic device, storage medium and program product of the present application is not limited.

[0041] Figure 1 A schematic diagram of an application scenario of an object classification method provided in an embodiment of the present application is used as an example in combination with the illustrated scenario: extracting object features from an object, which can reflect the attribute characteristics of the object or the characteristics of the application scenario, etc., and can represent the object. Classification processing is performed according to the object features to obtain the target classification to which the object belongs, and the use of the object is guided by the target classification.

[0042] Optionally, the object may be a product, and the target classification identifier may be a specific type of the product.

[0043] Combined with the scenario example description, taking the product as an example, after determining the specific type of the product, you can find the corresponding user manual and / or usage scenario according to the type. The user manual and / or usage scenario can accurately guide the use of the product to correctly achieve the expected functional effects of the product.

[0044] In the related art, the classification process is performed manually to obtain the target category to which the object belongs. However, when faced with objects of many types, complex types, or constantly changing types, the classification process performed manually has the problem of high complexity, resulting in low classification efficiency.

[0045] The object classification method provided in this application is intended to solve the above technical problems in the prior art.

[0046] The technical solution of the present application and how the technical solution of the present application solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0047] Figure 2A schematic diagram of a flow chart of an object classification method provided in an embodiment of the present application, the method comprising the following steps:

[0048] S201: Receive a classification request, where the classification request includes multimodal data of a target object.

[0049] The target object is an object that needs to be classified, and the classification request is used to trigger object classification.

[0050] Optionally, before using the object, a classification request is automatically triggered, thereby automatically triggering the object classification, which can effectively reduce the operation of manually triggering the object classification, thereby improving the efficiency of classification.

[0051] Optionally, the multimodal data is description data of the target object. For example, the multimodal data may be introduction information, description information, remarks information, or promotional information of the product.

[0052] Optionally, multimodal data can be represented by multiple data types, including but not limited to: text, image, audio, or video.

[0053] S202: extract multiple target object features of the target object from the multimodal data according to the classification request.

[0054] Optionally, the target object feature is a numerical value or vector representation representing an essential characteristic of the target object, and multiple target object features represent the target object from multiple different dimensions.

[0055] Combined with the scenario example, take the target object as a product, and the multimodal data as a promotional picture of the product. The promotional picture may include what specific product the target object is, the working conditions of the product, the specifications of the product, etc. These information are multiple target object features. In addition, the promotional picture may also include non-functional descriptions of the product that are only used for promotion. These descriptions cannot provide help for classification. Extracting the target object features before matching can avoid the interference of non-functional descriptions on classification, thereby improving the accuracy of classification.

[0056] S203: Determine a target database, and determine a plurality of mapping rules from the target database, wherein the plurality of mapping rules are obtained by extracting the model.

[0057] Optionally, the mapping rule is in the form of an IF-THEN statement, which includes a condition field and a conclusion field. The condition field includes at least one object feature, and the conclusion field includes a classification identifier. The mapping rule can be used to quickly perform matching processing.

[0058] Illustrated with a scenario example. For instance, any one mapping rule is: IF (object feature A) AND (object feature B) THEN (category A). If multiple target objects have object feature A and object feature B, then the target objects belong to category A.

[0059] Optionally, through the model extraction method, mapping rules are obtained by acquiring the complex patterns and relationships between object features and categories. The mapping rules can be reused, and no manual operation is required when using the mapping rules, thereby improving the efficiency of classification.

[0060] Optionally, if the relationship between object features and categories changes, then re - perform model matching to obtain updated mapping rules and update the target database to enhance the flexibility of the mapping rules.

[0061] Illustrated with a scenario example, compared with manually writing and maintaining a large number of mapping rules, using the model to automatically extract mapping rules can significantly reduce the workload of manual intervention. In addition, the automated process also reduces the possibility of human errors.

[0062] S204. Perform matching processing on the target object features through multiple mapping rules to obtain the target classification identifier of the target object.

[0063] Optionally, the target classification identifier is the unique identifier of the category corresponding to the target object. Through the target classification identifier, the category to which the target belongs can be uniquely determined.

[0064] Optionally, determine the target mapping rule from multiple mapping rules. The condition field of the target mapping rule includes at least each target object feature, and determine the information in the conclusion field of the target mapping rule as the target classification identifier.

[0065] The object classification method provided by the embodiments of the present application receives a classification request, where the classification request includes multi - modal data of a target object; according to the classification request, extract multiple target object features of the target object from the multi - modal data; determine a target database, and determine multiple mapping rules from the target database, where the multiple mapping rules are obtained through model extraction; perform matching processing on the target object features through the multiple mapping rules to obtain the target classification identifier of the target object. In the above solution, the multiple pre - extracted mapping rules can cover various mapping relationships between object features and categories, and classification can be quickly performed through the multiple mapping rules, avoiding the problem of high complexity in manual classification, thereby improving the classification efficiency.

[0066] Based on any one of the above embodiments, below, in combination with Figure 3 , the detailed process of object classification will be described.

[0067] Figure 3The following is a flow chart of an object classification method provided in an embodiment of the present application. Figure 3 As shown, the method includes:

[0068] S301: Receive a classification request, where the classification request includes multimodal data of a target object.

[0069] It should be noted that the execution process of S301 refers to S201 and will not be repeated here.

[0070] S302: extract multiple target object features of the target object from the multimodal data according to the classification request.

[0071] A feasible implementation method is to extract target object features through the following method: determine the data type of the multimodal data, the data type is text or image; if the data type is text, extract multiple target object features from the multimodal data through natural language processing technology; if the data type is image, extract multiple target object features from the multimodal data through computer vision technology and natural language processing technology.

[0072] Optionally, for text data, the text is in natural language form, and the BERT model can be used to implement natural language processing technology. The BERT model extracts multiple target object features from the text.

[0073] Optionally, for text data, a bidirectional long short-term memory (BiLSTM) model can be used to extract multiple target object features from the text. The BiLSTM model processes sequence data from both the forward and backward directions to better capture contextual information.

[0074] Optionally, for image data, after receiving a sample image, computer vision technology is used to detect the location of text from the image. Further, the location of the text is sent to a BERT model or a BiLSTM model to extract multiple target object features.

[0075] Optionally, before obtaining the area where the text is located, the image is noise-removed and normalized to improve the accuracy of extraction.

[0076] Next, combine Figure 4 The extracted object features are described.

[0077] Figure 4 Schematic diagram of extracting object features provided in the embodiment of the present application. Figure 4As shown, the data type of the multimodal data is determined. If the data type is text, multiple target object features are extracted from the multimodal data using natural language processing technology. If the data type is an image, the text part is first extracted from the image, and then multiple target object features are extracted from the text using natural language processing technology.

[0078] In this feasible implementation, adopting a suitable extraction scheme for automated extraction according to the data type can reduce manual operations and thus improve classification efficiency.

[0079] S303: Determine a target database, and determine multiple mapping rules from the target database, where the multiple mapping rules are extracted through a model.

[0080] A feasible implementation method can generate multiple mapping rules by the following method: determine multiple sample data and multiple label classification identifiers corresponding to multiple sample objects; extract multiple target object features of the target object from the multiple sample data; input the multiple target object features and label classification identifiers into the relational extraction model to obtain multiple mapping rules; and store the multiple mapping rules in the target database.

[0081] The sample object is an object used as a training sample, the multiple sample data are description data corresponding to the sample object, and the label classification identifier is a determined classification identifier of the sample object.

[0082] Combined with the scenario example, the relationship extraction model learns the association between multiple target object features and label classification identifiers, and generates multiple mapping rules based on the association. The multiple mapping rules are stored in the target database to enable accurate classification through the target database in the future.

[0083] In this feasible implementation, multiple mapping rules learned from a large number of samples through a relationship extraction model can quickly learn complex association relationships compared to manual methods, thereby improving the generation efficiency of multiple mapping rules.

[0084] In a feasible implementation method, the object classification method also includes: determining multiple information gain values ​​corresponding to multiple mapping rules; determining the mapping rule with the largest information gain value as the root node; starting from the root node, constructing a decision tree of multiple mapping rules according to multiple information gain values; and storing the decision tree in a database.

[0085] For example, information gain is an indicator that measures the ability of any rule to reduce uncertainty (entropy) during the classification process, and information gain can be used to select the best split node when building a decision tree.

[0086] Next, combine Figure 5 Describe the generation of a decision tree.

[0087] Figure 5 Schematic diagram of generating a decision tree provided in an embodiment of the present application. Figure 5 As shown in FIG. 1 , according to multiple information gain values, the mapping rule with the largest information gain value is selected as the root node. Starting from the root node, multiple information gain values ​​are recursively divided, and each time the mapping rule with the largest information gain among the currently remaining mapping rules is selected as the split point. When all mapping rules are correctly classified, the recursion stops and a decision tree is obtained.

[0088] Optionally, a complete decision tree structure is gradually constructed through a recursive process.

[0089] Optionally, the feature that is most helpful for classification is selected from multiple mapping rules as the root node of the decision tree. Starting from the root node, the best feature is selected for segmentation, and leaf nodes are generated according to the value of the feature. Then, this process is repeated for each leaf node until the stopping condition is met.

[0090] Optionally, after obtaining the decision tree, start from the root node and traverse the entire decision tree. During the traversal process, record the path from the root node to each leaf node. Each path from the root node to the leaf node can be converted into a mapping rule. Optimize the extracted rules as needed, such as merging similar rules, deleting redundant rules, etc., so as to gradually improve the decision tree.

[0091] Combined with the scenario example, the decision tree is stored in the database so that the decision tree can be obtained through the target database for classification later.

[0092] In this feasible implementation, by selecting the optimal split point, it is ensured that each split can minimize uncertainty, thereby improving the reliability of the decision tree.

[0093] In a feasible implementation manner, the object classification method further includes: generating multiple modification controls according to multiple mapping rules, the multiple modification controls are used to modify the multiple mapping rules; and generating a visualization page according to the decision tree and the multiple modification controls.

[0094] Optionally, the visualization page is used to display the structure of the decision tree, and the modification control is an element in the visualization page, which is used by the user to modify the mapping rules in the decision tree.

[0095] Optionally, multiple mapping rules are associated with multiple modification controls, and the multiple mapping rules correspond to the multiple modification controls one by one. In the visualization page, the mapping rules and modification controls that have an associated relationship are located in adjacent positions, so that the user can accurately modify the mapping rules through the modification controls.

[0096] Combined with the scenario example, if the mapping rule changes or the user finds from the visualization page that the mapping rule needs to be modified, the mapping rule can be modified by modifying the control.

[0097] In this feasible implementation, the visualization page enables the user to intuitively view and modify complex mapping rules, thereby improving the user experience.

[0098] S304: Determine at least one target mapping rule from the multiple mapping rules, wherein a condition field of any target mapping rule includes multiple target object features.

[0099] In combination with a scenario example, for example, if the multiple target object features are object feature A and object feature B, then the condition field of the target mapping rule determined from the multiple mapping rules should include at least object feature A and object feature B.

[0100] Optionally, if the number of target mapping rules is 1, the target mapping rule is used to perform matching processing to obtain a target classification identifier.

[0101] Optionally, a confidence level is set for each mapping rule. If there are multiple target mapping rules, a mapping rule with the highest confidence level among the multiple target mapping rules is used for matching processing to obtain a target classification identifier.

[0102] S305: If the number of at least one target mapping rule is multiple, determine the field information under the conclusion field of the multiple target mapping rules to obtain multiple classification identifiers to be selected.

[0103] Optionally, the conclusion field of each target mapping rule includes a field of information, and multiple target mapping rules correspond one-to-one to multiple classification identifiers to be selected.

[0104] In combination with the scenario example, if there are multiple target mapping rules, multiple target mapping rules will be matched to obtain multiple different candidate classification identifiers, which need to be further screened to obtain the target classification identifier.

[0105] Next, combine Figure 6 The matching process will be described.

[0106] Figure 6 This is a schematic diagram of the matching process provided in the embodiment of the present application. Figure 6 As shown, multiple mapping rules are determined, and a target mapping rule is determined from the multiple mapping rules. If the number of target mapping rules is 1, a matching process is performed through the 1 target mapping rule to obtain a target classification identifier. If the number of target mapping rules is multiple, a matching process is performed through the multiple target mapping rules to obtain multiple candidate classification identifiers, and the target classification identifier is screened from the multiple candidate classification identifiers.

[0107] S306: Determine multiple priorities corresponding to the multiple target mapping rules, and determine a target classification identifier from a plurality of candidate classification identifiers according to the multiple priorities.

[0108] Optionally, the priority is determined based on the specific scope of conditions covered by the mapping rule. For example, a mapping rule with a narrower coverage takes precedence over a broader mapping rule.

[0109] Exemplarily, for example, multiple target mapping rules include mapping rule 1 and mapping rule 2, mapping rule 1 is IF (natural disaster) AND (rice) THEN (rice natural disaster insurance), and mapping rule 2 is IF (natural disaster) AND (rice) AND (wheat) THEN (comprehensive agricultural insurance). Among them, natural disasters, rice, and wheat are object features, and rice natural disaster insurance and comprehensive agricultural insurance are classification identifiers. In this case, the classification identifier of mapping rule 1 is more specific, and the corresponding priority of mapping rule 1 is higher, and the classification identifier of mapping rule 1 is determined as the target classification identifier.

[0110] S307: Determine historical matching records from the target database, and determine a target classification identifier from a plurality of candidate classification identifiers according to the historical matching records.

[0111] Optionally, the target classification identifier is determined according to the frequency of occurrence of each candidate classification identifier in the historical matching records. For example, the candidate classification identifier with the highest frequency of occurrence is determined as the target classification identifier.

[0112] Combined with the scenario examples, historical matching records represent successful experiences in dealing with similar problems in the past, which have been verified by users. Historical matching records can provide effective references to improve the accuracy of classification.

[0113] It should be noted that the present application does not limit the execution order of S306 and S307.

[0114] Figure 7 This is a schematic diagram of the structure of an object classification device provided in an embodiment of the present application. Figure 7 As shown, the object classification device 70 may include: a receiving module 71, an extracting module 72, a determining module 73, and a matching module 74, wherein:

[0115] The receiving module 71 is used to receive a classification request, where the classification request includes multimodal data of a target object.

[0116] The extraction module 72 is used to extract multiple target object features of the target object from the multimodal data according to the classification request.

[0117] The determination module 73 is used to determine the target database and determine a plurality of mapping rules from the target database, wherein the plurality of mapping rules are obtained by extracting the model.

[0118] The matching module 74 is used to perform matching processing on the target object features through multiple mapping rules to obtain the target classification identification of the target object.

[0119] Optionally, the receiving module 71 may execute Figure 2 S201 in the embodiment.

[0120] Optionally, the extraction module 72 may execute Figure 2 S202 in the embodiment.

[0121] Optionally, the determination module 73 may execute Figure 2 S203 in the embodiment.

[0122] Optionally, the matching module 74 may execute Figure 2 S204 in the embodiment.

[0123] It should be noted that the object classification device shown in the embodiment of the present application can execute the technical solution shown in the above method embodiment, and its implementation principle and beneficial effects are similar, which will not be repeated here.

[0124] In a possible implementation, the extraction module 72 is specifically configured to:

[0125] Determine the data type of the multimodal data, which is text or image;

[0126] If the data type is text, multiple target object features are extracted from the multimodal data using natural language processing technology;

[0127] If the data type is a picture, multiple target object features are extracted from the multimodal data through computer vision technology and natural language processing technology.

[0128] Figure 8 A schematic diagram of the structure of an object classification device provided in an embodiment of the present application. Figure 7 Based on the embodiment shown, Figure 8 As shown, the object classification device 80 further includes: an execution module 75, a generation module 76, a construction module 77, and a display module 78, wherein:

[0129] The execution module 75 is used to:

[0130] Determining at least one target mapping rule from a plurality of mapping rules, wherein a condition field of any target mapping rule includes a plurality of target object features;

[0131] According to the target mapping rules, multiple target object features are matched and processed to obtain a target classification identifier.

[0132] In a possible implementation manner, the number of at least one target mapping rule is multiple; the execution module 75 is specifically used to:

[0133] Determine the field information under the conclusion fields of multiple target mapping rules to obtain multiple candidate classification identifiers;

[0134] Determine multiple priorities corresponding to multiple target mapping rules, and determine a target classification identifier from multiple classification identifiers to be selected according to the multiple priorities; or,

[0135] A historical matching record is determined from a target database, and a target classification identifier is determined from a plurality of candidate classification identifiers according to the historical matching record.

[0136] The generating module 76 is used for:

[0137] Determine a plurality of sample data and a plurality of label classification identifiers corresponding to the plurality of sample objects;

[0138] extracting multiple target object features of the target object from the multiple sample data;

[0139] Input multiple target object features and label classification identifiers into the relation extraction model to obtain multiple mapping rules;

[0140] Store multiple mapping rules into the target database.

[0141] Building block 77 for:

[0142] Determine multiple information gain values ​​corresponding to multiple mapping rules;

[0143] The mapping rule with the largest information gain value is determined as the root node;

[0144] Starting from the root node, a decision tree of multiple mapping rules is constructed according to multiple information gain values;

[0145] Store the decision tree in a database.

[0146] Display module 78, for:

[0147] Generate multiple modification controls according to the multiple mapping rules, and the multiple modification controls are used to modify the multiple mapping rules;

[0148] Generate a visualization page based on the decision tree and multiple modification controls.

[0149] Fig. 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application is shown in FIG. Fig. 9 As shown, the electronic device includes:

[0150] The electronic device includes a processor 291 and a memory 292; it may also include a communication interface 293 and a bus 294. The processor 291, the memory 292, and the communication interface 293 may communicate with each other through the bus 294. The communication interface 293 may be used for information transmission. The processor 291 may call the logic instructions in the memory 292 to execute the method of the above embodiment.

[0151] In addition, the logic instructions in the above-mentioned memory 292 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.

[0152] The memory 292 is a computer-readable storage medium that can be used to store software programs and computer executable programs, such as program instructions / modules corresponding to the methods in the embodiments of the present application. The processor 291 executes functional applications and data processing by running the software programs, instructions, and modules stored in the memory 292, that is, implementing the methods in the above method embodiments.

[0153] The memory 292 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application required for at least one function; the data storage area may store data created according to the use of the terminal device, etc. In addition, the memory 292 may include a high-speed random access memory and may also include a non-volatile memory.

[0154] An embodiment of the present application provides a non-temporary computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the method as described in the above embodiment.

[0155] An embodiment of the present application provides a computer program product, including a computer program, which implements the method of the above embodiment when the computer program is executed by a processor.

[0156] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present application.

[0157] It should be further noted that, although the various steps in the flowchart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowchart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0158] It should be understood that the above-mentioned device embodiments are only illustrative, and the device of the present application can also be implemented in other ways. For example, the division of units / modules in the above-mentioned embodiments is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units, modules or components can be combined, or can be integrated into another system, or some features can be ignored or not executed.

[0159] In addition, unless otherwise specified, each functional unit / module in each embodiment of the present application may be integrated into one unit / module, each unit / module may exist physically separately, or two or more units / modules may be integrated together. The above-mentioned integrated unit / module may be implemented in the form of hardware or in the form of a software program module.

[0160] If the integrated unit / module is implemented in the form of hardware, the hardware may be a digital circuit, an analog circuit, etc. The physical implementation of the hardware structure includes but is not limited to transistors, memristors, etc. The processor may be any appropriate hardware processor, such as CPU, GPU, FPGA, DSP, and ASIC, etc. The storage unit may be any appropriate magnetic storage medium or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc.

[0161] If the integrated unit / module is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a memory, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned memory includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0162] In the above embodiments, the description of each embodiment has its own emphasis. For the part not described in detail in a certain embodiment, please refer to the relevant description of other embodiments. The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0163] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0164] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A method for object classification, characterized in that: include: receiving a classification request, the classification request including multimodal data of a target object; extracting a plurality of target object features of the target object from the multimodal data according to the classification request; Determine a target database, and determine a plurality of mapping rules from the target database, wherein the plurality of mapping rules are obtained by extracting a model; The target object features are matched using the multiple mapping rules to obtain a target classification identifier of the target object.

2. The method according to claim 1, characterized in that: Extracting a plurality of target object features of the target object from the multimodal data includes: Determine a data type of the multimodal data, the data type being text or image; If the data type is text, extracting the plurality of target object features from the multimodal data by using natural language processing technology; If the data type is a picture, the multiple target object features are extracted from the multimodal data by using computer vision technology and the natural language processing technology.

3. The method according to claim 1 or 2, characterized in that: Matching the multiple target object features through the multiple mapping rules to obtain a target classification identifier of the target object includes: Determine at least one target mapping rule from the plurality of mapping rules, wherein a condition field of any target mapping rule includes the plurality of target object features; The target object features are matched according to the target mapping rule to obtain the target classification identifier.

4. The method according to claim 3, characterized in that: The at least one target mapping rule is multiple; matching processing is performed on the multiple target object features according to the target mapping rule to obtain the target classification identifier, including: Determine the field information under the conclusion fields of multiple target mapping rules to obtain multiple candidate classification identifiers; Determine multiple priorities corresponding to multiple target mapping rules, and determine the target classification identifier from the multiple classification identifiers to be selected according to the multiple priorities; or, A historical matching record is determined from the target database, and the target classification identifier is determined from the multiple classification identifiers to be selected according to the historical matching record.

5. The method according to any one of claims 1 to 4, characterized in that The method further comprises: Determine a plurality of sample data and a plurality of label classification identifiers corresponding to the plurality of sample objects; Extracting a plurality of target object features of the target object from the plurality of sample data; Inputting the plurality of target object features and the label classification identifier into a relation extraction model to obtain the plurality of mapping rules; The plurality of mapping rules are stored in a target database.

6. The method according to claim 5, characterized in that The method further comprises: Determining a plurality of information gain values ​​corresponding to the plurality of mapping rules; Determine the mapping rule with the largest information gain value as the root node; Starting from a root node, constructing a decision tree of the plurality of mapping rules according to the plurality of information gain values; The decision tree is stored in the database.

7. The method according to claim 6, characterized in that The method further comprises: Generate a plurality of modification controls according to a plurality of mapping rules, wherein the plurality of modification controls are used to modify the plurality of mapping rules; A visualization page is generated according to the decision tree and the multiple modification controls.

8. An object classification device, characterized in that: include: A receiving module, configured to receive a classification request, wherein the classification request includes multimodal data of a target object; an extraction module, configured to extract a plurality of target object features of the target object from the multimodal data according to the classification request; A determination module, used to determine a target database, and determine a plurality of mapping rules from the target database, wherein the plurality of mapping rules are obtained by extracting a model; The matching module is used to perform matching processing on the target object features through the multiple mapping rules to obtain the target classification identifier of the target object.

9. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the 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-executable instructions, which are used to implement the method according to any one of claims 1 to 7 when executed by a processor.

11. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 7 when being executed by a processor.

Images