Data processing method and device

By using the model in the first published information classification tree to classify and annotate the training text, a sample set is generated for training the second published information classification tree, which solves the problem of time-consuming and labor-consuming manual labeling and realizes the rapidity of automatic labeling Efficiency, adapt to changes in the classification rules for publishing information.

CN112015893BActive Publication Date: 2024-11-29DOUYIN VISION CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202010807711.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-08-12
Publication Date
2024-11-29
Estimated Expiration
2040-11-29

AI Technical Summary

Technical Problem

In the prior art, manual labeling of training sample sets is time-consuming and labor-intensive, and it is impossible to quickly adapt to changes in the classification rules for publishing information.

Method used

The training text is classified and annotated through the release information classification model in the first release information classification tree, and a set of label samples for training the second release information classification tree is generated, so that automatic labeling is realized to speed up the labeling process.

Benefits of technology

It improves the labeling speed and efficiency, reduces the time required to re-label the training text due to changes in classification rules, and improves the flexibility and efficiency of data processing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112015893B_ABST
    Figure CN112015893B_ABST
Patent Text Reader

Abstract

The present disclosure discloses a data processing method, device, electronic device and computer-readable storage medium. The method includes: in response to the change operation of the first release information classification tree, obtaining the second release information classification tree; wherein the first release information classification tree and the second release information classification tree are both composed of multiple category nodes, and each category node contains a release information classification model; obtaining training text from the landing page of the release information; classifying the training text with the release information classification model of each category node in the first release information classification tree, and marking the release information category of each training text according to the classification result; determining the training text sample set for training the release information classification model of each category node in the second release information classification tree according to the marked training text. The embodiment of the present disclosure can accelerate the annotation speed and improve the annotation efficiency by automatically annotating the training text.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular to a data processing method, device and computer-readable storage medium. Background Art

[0002] Release information classification refers to the classification of release information into different types according to different purposes. In this way, the appropriate release information type can be selected for push according to the different characteristics of users to achieve the best effect.

[0003] The classification task is usually performed through a release information classification model, and training the release information classification model requires training text for training. The usually adopted method is: obtain a training sample set by manually annotating the training text, and then use the training sample set to train the release information classification model corresponding to each category node in the release information classification tree. Subsequently, the release information classification tree is used to classify the training text into release information rows. However, when the release information classification rules change, all training texts need to be re-annotated to obtain a new training sample set, which is very time-consuming and labor-intensive. Summary of the invention

[0004] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0005] The technical problem solved by the present disclosure is to provide a data processing method to at least partially solve the technical problem of manually labeling training sample sets in the prior art, which is very time-consuming and labor-intensive. In addition, a data processing device, a data processing hardware device, a computer-readable storage medium and a data processing terminal are also provided.

[0006] In order to achieve the above object, according to one aspect of the present disclosure, the following technical solution is provided:

[0007] A data processing method, comprising:

[0008] In response to the change operation of the first release information classification tree, a second release information classification tree is obtained; wherein the first release information classification tree and the second release information classification tree are both composed of a plurality of category nodes, and each category node contains a release information classification model;

[0009] Obtain training text from the landing page of the published information;

[0010] Classifying the training texts into published information using the published information classification model of each category node in the first published information classification tree, and marking the published information category of each training text according to the classification result;

[0011] A training text sample set for training a release information classification model for each category node in the second release information classification tree is determined according to the annotated training text.

[0012] In order to achieve the above object, according to one aspect of the present disclosure, the following technical solution is provided:

[0013] A data processing device, comprising:

[0014] A change response module, used to obtain a second release information classification tree in response to a change operation of the first release information classification tree; wherein the first release information classification tree and the second release information classification tree are both composed of a plurality of category nodes, and each category node contains a release information classification model;

[0015] A text acquisition module, used to acquire training text from the landing page of the published information;

[0016] A text annotation module, used to classify the training text according to the published information classification model of each category node in the first published information classification tree, and annotate the published information category of each training text according to the classification result;

[0017] The sample set determination module is used to determine a training text sample set for training a release information classification model for each category node in the second release information classification tree according to the marked training text.

[0018] In order to achieve the above object, according to one aspect of the present disclosure, the following technical solution is provided:

[0019] An electronic device, comprising:

[0020] a memory for storing non-transitory computer-readable instructions; and

[0021] A processor is used to run the computer-readable instructions so that the processor implements any of the data processing methods described above when executing the instructions.

[0022] In order to achieve the above object, according to one aspect of the present disclosure, the following technical solution is provided:

[0023] A computer-readable storage medium is used to store non-transitory computer-readable instructions. When the non-transitory computer-readable instructions are executed by a computer, the computer is caused to execute any of the data processing methods described above.

[0024] In order to achieve the above object, according to another aspect of the present disclosure, the following technical solution is also provided:

[0025] A data processing terminal comprises any one of the above data processing devices.

[0026] The disclosed embodiment classifies the published information of the training text through the published information classification model of each category node in the first published information classification tree, and labels the published information category of each training text according to the classification result, and determines the training text sample set used to train the published information classification model of each category node in the second published information classification tree according to the labeled training text. By automatically labeling the training text, the labeling speed can be accelerated and the labeling efficiency can be improved.

[0027] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the following specifically cites the preferred embodiments and describes them in detail with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0029] Figure 1a is a flowchart of a data processing method according to an embodiment of the present disclosure;

[0030] Figure 1b A schematic diagram of a first release information classification tree in a data processing method according to an embodiment of the present disclosure;

[0031] Figure 1c A schematic diagram of a second release information classification tree in a data processing method according to an embodiment of the present disclosure;

[0032] Figure 1d A schematic diagram of a splitting operation in a data processing method according to an embodiment of the present disclosure;

[0033] Figure 1e A schematic diagram of an adding operation in a data processing method according to an embodiment of the present disclosure;

[0034] Figure 1f A schematic diagram of a deletion operation in a data processing method according to an embodiment of the present disclosure;

[0035] Figure 1g A schematic diagram of a moving operation in a data processing method according to an embodiment of the present disclosure;

[0036] Figure 2 is a flowchart of a data processing device according to an embodiment of the present disclosure;

[0037] Figure 3 The figure is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0038] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, which are instead provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.

[0039] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders or in parallel. In addition, the method embodiments may include additional steps or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0040] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0041] Embodiment 1

[0042] In order to solve the technical problem of manually labeling training sample sets, which is very time-consuming and labor-intensive, the present disclosure provides a data processing method. Figure 1a As shown, the data processing method mainly includes the following steps S11 to S14.

[0043] Step S11: In response to the change operation of the first release information classification tree, a second release information classification tree is obtained; wherein the first release information classification tree and the second release information classification tree are both composed of multiple category nodes, and each category node contains a release information classification model.

[0044] The first release information classification tree and the second release information classification tree are both tree structures, and both the first release information classification tree and the second release information classification tree contain multiple levels of release information categories, and each level of release information categories is composed of multiple category nodes, and each category node contains a release information classification model. Figure 1bThe figure is an example of a first release information classification tree. By performing a change operation on the basis of the first release information classification tree, a second release information classification tree can be obtained, that is, the second release information classification tree is a deformation of the first release information classification tree, such as Figure 1c As shown, it is an example of the second release information classification tree, that is, Figure 1b A category node is added to the root node of the first published information classification tree shown.

[0045] The change operation may include adding a category node, splitting a category node, deleting a category node, moving a category node, etc.

[0046] The information released can be advertisements, news, personal dynamics, etc. For example, according to the content, the information released can be classified into product release information, enterprise release information, service release information, concept release information, etc. The categories corresponding to the product release information include 3C products, smart terminal products, digital products, etc., where 3C products are a general term for computer, communication and consumer electronic products. The categories corresponding to smart terminal products include smart phones, smart speakers, smart bracelets, etc. Based on the above classification, the first user intention release information classification tree digital products 3C products can be constructed.

[0047] Step S12: Obtain training text from the landing page of the published information.

[0048] Specifically, the title and text content (including direct text or text obtained through image text recognition processing, etc.) on the landing page can be obtained as training text.

[0049] Step S13: classifying the training text according to the release information classification model of each category node in the first release information classification tree, and marking the release information category of each training text according to the classification result.

[0050] Since the release information classification model of each category node in the first release information classification tree is a pre-trained release information classification model, the first release information classification tree can be used to perform preliminary annotation on the training text.

[0051] Step S14: determining a set of training text samples for training a release information classification model for each category node in the second release information classification tree according to the annotated training text.

[0052] Since the structure of the second release information classification tree has changed, the training text annotated by the first release information classification tree cannot be directly used as a training text sample set for training the release information classification model of each category node in the second release information classification tree, and further annotation is required.

[0053] This embodiment classifies the published information of the training text through the published information classification model of each category node in the first published information classification tree, and labels the published information category of each training text according to the classification result. The training text sample set used to train the published information classification model of each category node in the second published information classification tree is determined according to the labeled training text. By automatically labeling the training text, the labeling speed can be accelerated and the labeling efficiency can be improved.

[0054] In a first optional embodiment, step S14 specifically includes:

[0055] 1. Determine a set of training text samples for training the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next-level category node of the second category node in the second release information classification tree based on the training text annotated by the release information classification model of the first category node in the first release information classification tree; wherein the first category node is a category node that may be incorrectly annotated; and the second category node is the same category node as the first category node in the second release information classification tree.

[0056] 2. Use the training text annotated by the release information classification model of the third category node in the first release information classification tree as a training text sample set for training the release information classification model of the fourth category node in the second release information classification tree; the third category node is a correctly annotated category node; the fourth category node is the same category node as the third category node in the second release information classification tree.

[0057] Specifically, the category nodes whose semantics have changed and the category nodes whose semantics have not changed in the first release information classification tree can be determined according to the change operation; the category nodes whose semantics have changed are regarded as category nodes that may be incorrectly labeled, namely the first category nodes, and the category nodes whose semantics have not changed are regarded as category nodes that are correctly labeled, namely the third category nodes; at the same time, the category node identical to the first category node is found in the second release information classification tree as the second category node and the next-level category node of the second category node is found, and the category node identical to the third category node is found in the second release information classification tree as the fourth category node. Since the second release information classification tree is obtained on the basis of the first release information classification tree, when the category nodes in the first release information classification tree undergo semantic changes, the release information recognition model of the corresponding category node in the second release information classification tree and the release information recognition model of the next-level category node may change, so they both need to be retrained.

[0058] In a second optional embodiment, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category node is the plurality of subcategory nodes; and step S141 specifically includes:

[0059] 1. Obtain a set of positive training text samples annotated by the release information classification model of the first category of nodes.

[0060] This embodiment splits the first category node into finer classifications, namely, into multiple subcategory nodes, so the positive training text sample set used by the release information classification model of the multiple subcategory nodes is a subset of the positive training text sample set used by the release information classification model of the first category node. Therefore, the positive training text sample set classified and labeled by the release information classification model of the first category node cannot be determined as the positive training text sample set of the release information classification model of the second category node and the release information classification model of each subcategory node in the second release information classification tree. Therefore, it is necessary to manually re-label the training text in the above-mentioned positive training text sample set according to the above-mentioned multiple subcategory nodes, and finally determine whether it is the positive training text sample set of the release information classification model of the second category node.

[0061] 2. In response to the labeling operation of the training text in the training text sample set, label the training samples.

[0062] 3. Re-determine the positive training text sample set and the negative training text sample set for training the release information classification model of the second category node and the release information classification model of each subcategory node in the second release information classification tree according to the labeling results.

[0063] by Figure 1dFor example, the category of smart terminal products is split into multiple finer categories, namely "smart phones", "smart speakers" and "smart bracelets". In this way, the semantics of the original "smart terminal products" node, i.e., the first category node, changes, and the positive training text sample set annotated by the original "smart terminal products" node's published information classification model is uncertain. It is necessary to manually annotate to determine the positive training text sample set and negative training text sample set of the published information classification model of the above-mentioned "smart phone" node, "smart speaker" node and "smart bracelet" node. According to the positive training text sample set of the published information classification model of the above-mentioned "smart phone" node, "smart speaker" node and "smart bracelet" node, the positive training text sample set of the published information classification model of the new "smart terminal products" node, i.e., the second category node, can be determined. That is, the positive training text sample set of the published information classification model of the above-mentioned "smart phone" node, "smart speaker" node and "smart bracelet" node is the positive training text sample set of the published information classification model of the new "smart terminal products" node.

[0064] In a third optional embodiment, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category node is the plurality of subcategory nodes; and step S141 specifically includes:

[0065] 1. Obtain a set of negative training text samples annotated by the release information classification model of the first category of nodes.

[0066] This embodiment splits the first category node into finer categories, namely, into multiple subcategory nodes. Therefore, the set of negative training text samples classified and labeled by the publication information classification model of the first category node must be the set of negative training text samples of the publication information classification model of the second category node and the publication information classification model of each subcategory node in the second publication information classification tree. Therefore, no manual labeling is required.

[0067] 2. Using the negative training text sample set as a negative training text sample set for training the release information classification model of the second category node and the release information classification model of each subcategory node in the second release information classification tree.

[0068] With the above Figure 1dFor example, the split in , the smart terminal product category is split into multiple finer categories, "smart phones", "smart speakers" and "smart bracelets". In this way, the semantics of the original "smart terminal products" node, that is, the first category, changes. The negative training text sample set annotated by the release information classification model of the original "smart terminal products" node is also the negative training text sample set of the release information classification model of the new "smart terminal products" node, that is, the second category node, and also the negative training text sample set of the release information classification model of the "smart phone" subcategory node, the release information classification model of the "smart speaker" subcategory node, and the release information classification model of the "smart bracelet" subcategory node.

[0069] In a fourth optional embodiment, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node; and step S141 specifically includes:

[0070] 1. Obtain a set of negative training text samples annotated by the release information classification model of the first category of nodes.

[0071] When a new release information category, i.e., a subcategory node, is added to the first category node, the semantics of the first category node changes. The negative training text sample set annotated by the release information classification model of the first category node is uncertain (because it may be the positive training text sample set of the subcategory node), and manual re-annotation is required to determine the negative training text sample set of the release information classification model of the second category node and the positive training text sample set of the release information classification model of the subcategory node in the second release information classification tree.

[0072] 2. In response to the labeling operation of the training text in the negative training text sample set, label the training samples.

[0073] 3. Re-determine the negative training text sample set for training the release information classification model of the second category node and the positive training text sample set for training the release information classification model of the subcategory node in the second release information classification tree based on the labeling results.

[0074] by Figure 1e Taking the addition shown as an example, digital products are added to the root node, i.e., the first category node, and the semantics of the root node changes. The negative training text sample set annotated by the release information classification model of the old root node, i.e., the first category node, is uncertain (because it may be the positive of the digital product model), and manual re-annotation is required to determine the negative training text sample set of the release information classification model of the new root node, i.e., the second category node, and the positive training text sample set of the release information classification model of the digital product subcategory node.

[0075] In a fifth optional embodiment, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node; and step S141 specifically includes:

[0076] 1. Obtain a set of positive training text samples annotated by the release information classification model of the first category of nodes.

[0077] When a new release information category, ie, a subcategory node, is added to the first category node, the semantics of the first category node changes. The positive training text sample set annotated by the release information classification model of the first category node is the negative training text sample set of the added subcategory node model.

[0078] 2. In response to the labeling operation of the training text in the positive training text sample set, the positive training text sample set is used as a negative training text sample set for training the release information classification model of the subcategory node in the second release information classification tree.

[0079] by Figure 1e Taking the addition shown as an example, digital products are added to the root node, i.e., the first category node, and the semantics of the root node changes. The positive training text sample set annotated by the old root node model, i.e., the release information classification model of the first category node, is the negative training text sample set of the release information classification model of the digital product subcategory node.

[0080] In a sixth optional embodiment, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node; and step S141 specifically includes:

[0081] 1. Obtain a set of positive training text samples annotated by the release information classification model of the first category of nodes.

[0082] When a new release information category, i.e., a subcategory node, is added to the first category node, the semantics of the first category node changes. If the first category node still has other subcategory nodes, the positive training text sample set annotated by the release information classification model of the first category node is the positive training text sample set of the release information classification model of the second category node in the second release information classification tree.

[0083] 2. Using the positive training text sample set as a positive training text sample set for training a release information classification model of the second category node in the second release information classification tree.

[0084] by Figure 1eTaking the addition shown as an example, a digital product is added to the root node, i.e., the first category node, and the semantics of the root node changes. The positive training text sample set annotated by the release information classification model of the old root node, i.e., the first category node, becomes the positive training text sample set of the release information classification model of the new root node, i.e., the second category node.

[0085] In a seventh optional embodiment, the change operation is to delete a subcategory node of the first category node, and the corresponding next-level category node is empty; and step S141 specifically includes:

[0086] 1. Obtain a set of positive training text samples annotated by the release information classification model of the first category of nodes.

[0087] Specifically, from the subcategory node under the first category node, the semantics of the first category node changes, and the set of positive training text samples annotated by the release information classification model of the first category node is uncertain, and it needs to be re-annotated, because the positive training text samples annotated by the release information classification model of the subcategory node that has been deleted may be negative training text samples for the release information classification model of the second category node in the second release information classification tree.

[0088] by Figure 1f Taking the deletion shown in as an example, if the "3C product" sub-category node is deleted from the root node, that is, the first category node, the root node semantics changes, and the negative training text sample set predicted by the old root node model is still the negative training text sample set of the new root node model; the positive training text sample set predicted by the old root node needs to be re-labeled (because it may be the positive training text sample set of the 3C product model that has been deleted, and it is the negative training text sample set for the new root node model); the positive training text sample set and the negative training text sample set predicted by the "3C product" model are both the negative training text sample sets of the new root node model.

[0089] 2. In response to the labeling operation of the training text in the training text sample set, label the training samples.

[0090] 3. Re-determine a training text sample set for training the release information classification model of the second category node in the second release information classification tree according to the labeling result.

[0091] In an eighth optional embodiment, the change operation is to delete a subcategory node of the first category node, and the corresponding next-level category node is empty; step S141 specifically includes:

[0092] 1. Obtain a positive training text sample set and a negative training text sample set annotated by a release information classification model of the subcategory node.

[0093] Specifically, from the subcategory node under the first category node, the semantics of the first category node changes, and the positive training text sample set and the negative training text sample set annotated by the release information classification model of the deleted subcategory node are both the negative training text sample set of the release information classification model of the second category node in the second release information classification tree.

[0094] 2. Using both the positive training text sample set and the negative training text sample set as negative training text sample sets for training the release information classification model of the second category node in the second release information classification tree.

[0095] In a ninth optional embodiment, the change operation is to delete a subcategory node of the first category node, and the corresponding next-level category node is empty; step S141 specifically includes:

[0096] 1. Obtain a set of negative training text samples annotated by the release information classification model of the first category of nodes.

[0097] Specifically, from the subcategory node under the first category node, the semantics of the first category node changes, and the negative training text sample set annotated by the release information classification model of the first category node is the negative training text sample set of the release information classification model of the second category node in the second release information classification tree.

[0098] 2. Using the negative training text sample set as a negative training text sample set for training a release information classification model of the second category node in the second release information classification tree.

[0099] In addition, the change operation may also be the following three situations:

[0100] First, move the category node to the parent node level, that is, move the fifth category node to the first category node in the first published information classification tree as a subcategory node of the first category node. If the fifth category node has a parent node, it is equivalent to a delete operation for the parent node, and the method for determining the training text sample set of the corresponding published information classification model can be found in the seventh to ninth optional embodiments mentioned above. For the first category node, it is equivalent to an add operation, and the method for determining the training text sample set of the corresponding published information classification model can be found in the fourth to sixth optional embodiments mentioned above. Figure 1g Taking the mobile as an example, the "smartphone" node changes from the third-level classification to the second-level classification, and the semantics of the parent node of the original third-level classification changes, and its operation is the same as deletion. For the root node, that is, the first category node, its operation is the same as adding.

[0101] Second, moving to the sibling node of the parent node is equivalent to a deletion operation for the parent node of the moving party, and the method for determining the corresponding training text sample set of the release information classification model can be found in the seventh to ninth optional embodiments above. For the parent node being moved to, it is equivalent to an addition operation, and the method for determining the corresponding training text sample set of the release information classification model can be found in the fourth to sixth optional embodiments above.

[0102] Third, moving to the child node of the sibling node does not change the semantics of the parent node, which is equivalent to an add operation for the sibling node. The method for determining the training text sample set of the corresponding release information classification model can be found in the fourth to sixth optional embodiments above.

[0103] Those skilled in the art should understand that, based on the above-mentioned embodiments, obvious modifications (for example, superposition of the listed modes) or equivalent substitutions may be made.

[0104] In the above, although the various steps in the data processing method embodiment are described in the above order, those skilled in the art should be aware that the steps in the embodiments of the present disclosure are not necessarily executed in the above order, and they may also be executed in other orders such as reverse order, parallel order, and cross order. Moreover, on the basis of the above steps, those skilled in the art may also add other steps. These obvious variations or equivalent replacement methods should also be included in the protection scope of the present disclosure and will not be repeated here.

[0105] The following is an embodiment of the device disclosed in the present invention. The embodiment of the device disclosed in the present invention can be used to execute the steps implemented by the embodiment of the method disclosed in the present invention. For the convenience of explanation, only the parts related to the embodiment of the present invention are shown. For specific technical details not disclosed, please refer to the embodiment of the method disclosed in the present invention.

[0106] Embodiment 2

[0107] In order to solve the technical problem that multiple manually labeled training sample sets are very time-consuming and labor-intensive, the present disclosure provides a data processing device. The device can execute the steps in the data processing method embodiment described in the above embodiment 1. Figure 2 As shown, the device mainly includes: a change response module 21, a text acquisition module 22, a text annotation module 23 and a sample set determination module 24; wherein,

[0108] The change response module 21 is used to obtain a second release information classification tree in response to the change operation of the first release information classification tree; wherein the first release information classification tree and the second release information classification tree are both composed of a plurality of category nodes, and each category node contains a release information classification model;

[0109] The text acquisition module 22 is used to acquire training text from the landing page of the published information;

[0110] The text annotation module 23 is used to classify the training text according to the published information classification model of each category node in the first published information classification tree, and annotate the published information category of each training text according to the classification result;

[0111] The sample set determination module 24 is used to determine a training text sample set for training a release information classification model of each category node in the second release information classification tree according to the annotated training text.

[0112] Furthermore, the sample set determination module 24 is specifically used to determine a training text sample set for training the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next-level category node of the second category node in the second release information classification tree based on the training text annotated by the release information classification model of the first category node in the first release information classification tree; wherein, the first category node is a category node that may be incorrectly annotated; and the second category node is a category node in the second release information classification tree that is the same as the first category node.

[0113] The sample set determination module 24 is specifically used to use the training text annotated by the release information classification model of the third category node in the first release information classification tree as the training text sample set for training the release information classification model of the fourth category node in the second release information classification tree; the third category node is a correctly annotated category node; the fourth category node is the same category node as the third category node in the second release information classification tree.

[0114] Furthermore, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category nodes are the plurality of subcategory nodes;

[0115] Moreover, the sample set determination module 24 is specifically used to: obtain a positive training text sample set annotated by the publication information classification model of the first category node; annotate the training samples in response to the annotation operation of the training text in the positive training text sample set; and re-determine the positive training text sample set and the negative training text sample set for training the publication information classification model of the second category node and the publication information classification model of each subcategory node in the second publication information classification tree based on the annotation results.

[0116] Furthermore, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category node is the plurality of subcategory nodes;

[0117] Moreover, the sample set determination module 24 is specifically used to: obtain a negative training text sample set annotated by the publication information classification model of the first category node; and use the negative training text sample set as a negative training text sample set for training the publication information classification model of the second category node and the publication information classification model of each subcategory node in the second publication information classification tree.

[0118] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0119] Moreover, the sample set determination module 24 is specifically used to: obtain a negative training text sample set annotated by the publication information classification model of the first category node; annotate the training samples in response to the annotation operation of the training text in the negative training text sample set; and re-determine the negative training text sample set for training the publication information classification model of the second category node in the second publication information classification tree and the positive training text sample set for the publication information classification model of the subcategory node based on the annotation results.

[0120] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0121] Furthermore, the sample set determination module 24 is specifically used to: obtain a positive training text sample set annotated by the release information classification model of the first category node; in response to the annotation operation of the training text in the positive training text sample set, use the positive training text sample set as a negative training text sample set for training the release information classification model of the subcategory node in the second release information classification tree.

[0122] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0123] Furthermore, the sample set determination module 24 is specifically used to: obtain a positive training text sample set annotated by the release information classification model of the first category node; and use the positive training text sample set as a positive training text sample set for training the release information classification model of the second category node in the second release information classification tree.

[0124] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0125] Furthermore, the sample set determination module 24 is specifically used to: obtain a positive training text sample set annotated by the release information classification model of the first category node; annotate the training samples in response to the annotation operation of the training text in the positive training text sample set; and re-determine the training text sample set for training the release information classification model of the second category node in the second release information classification tree based on the annotation results.

[0126] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0127] Furthermore, the sample set determination module 24 is specifically used to: obtain a positive training text sample set and a negative training text sample set annotated by the release information classification model of the subcategory node; and use both the positive training text sample set and the negative training text sample set as a negative training text sample set for training the release information classification model of the second category node in the second release information classification tree.

[0128] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0129] Moreover, the sample set determination module 24 is specifically used to: obtain a negative training text sample set annotated by the release information classification model of the first category node; and use the negative training text sample set as a negative training text sample set for training the release information classification model of the second category node in the second release information classification tree.

[0130] For detailed descriptions on the working principle of the data processing device embodiment, the technical effects achieved, etc., please refer to the relevant descriptions in the aforementioned data processing method embodiment, which will not be repeated here.

[0131] Embodiment 3

[0132] Reference below Figure 3 , which shows a schematic diagram of the structure of an electronic device 300 suitable for implementing the embodiment of the present disclosure. The terminal device in the embodiment of the present disclosure may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0133] like Figure 3As shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0134] Typically, the following devices may be connected to the I / O interface 305: input devices 306 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 308 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 309. The communication devices 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 3 The electronic device 300 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0135] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 309, or installed from the storage device 308, or installed from the ROM 302. When the computer program is executed by the processing device 301, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0136] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any superposition of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any superposition of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable superposition of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable superposition of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable superposition of the above.

[0137] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with 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"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0138] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0139] The above-mentioned computer-readable medium carries one or more programs. When the above-mentioned one or more programs are executed by the electronic device, the electronic device: obtains a second release information classification tree in response to the change operation of the first release information classification tree; wherein the first release information classification tree and the second release information classification tree are both composed of multiple category nodes, and each category node contains a release information classification model; obtains training text from the landing page of the release information; classifies the training text into release information through the release information classification model of each category node in the first release information classification tree, and labels the release information category of each training text according to the classification result; determines a training text sample set for training the release information classification model of each category node in the second release information classification tree according to the labeled training text.

[0140] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or superpositions thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0141] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the superposition of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with the superposition of special hardware and computer instructions.

[0142] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a unit does not limit the unit itself in some cases. For example, the first acquisition unit may also be described as a "unit for acquiring at least two Internet Protocol addresses".

[0143] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0144] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable superposition of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable superposition of the foregoing.

[0145] According to one or more embodiments of the present disclosure, a data processing method is provided, including:

[0146] In response to the change operation of the first release information classification tree, a second release information classification tree is obtained; wherein the first release information classification tree and the second release information classification tree are both composed of a plurality of category nodes, and each category node contains a release information classification model;

[0147] Obtain training text from the landing page of the published information;

[0148] Classifying the training texts into published information using the published information classification model of each category node in the first published information classification tree, and marking the published information category of each training text according to the classification result;

[0149] A training text sample set for training a release information classification model for each category node in the second release information classification tree is determined according to the annotated training text.

[0150] Furthermore, the step of determining a set of training text samples for training a release information classification model for each category node in the second release information classification tree based on the annotated training texts includes:

[0151] A set of training text samples is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree to be used for training the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next-level category node of the second category node in the second release information classification tree; wherein the first category node is a category node that may be incorrectly annotated; and the second category node is a category node in the second release information classification tree that is the same as the first category node.

[0152] The training text annotated by the release information classification model of the third category node in the first release information classification tree is used as a training text sample set for training the release information classification model of the fourth category node in the second release information classification tree; the third category node is a correctly annotated category node; the fourth category node is the same category node as the third category node in the second release information classification tree.

[0153] Furthermore, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category nodes are the plurality of subcategory nodes;

[0154] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0155] Obtaining a positive training text sample set annotated by a release information classification model of the first category of nodes;

[0156] In response to the labeling operation of the training text in the training text sample set, labeling the training sample;

[0157] The positive training text sample set and the negative training text sample set used to train the release information classification model of the second category node and the release information classification model of each subcategory node in the second release information classification tree are re-determined according to the labeling result.

[0158] Furthermore, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category nodes are the plurality of subcategory nodes;

[0159] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0160] Obtaining a set of negative training text samples annotated by the release information classification model of the first category of nodes;

[0161] The negative training text sample set is used as a negative training text sample set for training the release information classification model of the second category node and the release information classification model of each subcategory node in the second release information classification tree.

[0162] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0163] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0164] Obtaining a set of negative training text samples annotated by the release information classification model of the first category of nodes;

[0165] In response to the labeling operation of the training text in the negative training text sample set, labeling the training sample;

[0166] The negative training text sample set for training the release information classification model of the second category node and the positive training text sample set for training the release information classification model of the subcategory node in the second release information classification tree are re-determined according to the labeling result.

[0167] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0168] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0169] Obtaining a positive training text sample set annotated by a release information classification model of the first category of nodes;

[0170] In response to the labeling operation of the training text in the positive training text sample set, the positive training text sample set is used as a negative training text sample set for training the release information classification model of the subcategory node in the second release information classification tree.

[0171] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0172] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0173] Obtaining a positive training text sample set annotated by a release information classification model of the first category of nodes;

[0174] The positive training text sample set is used as a positive training text sample set for training a release information classification model of the second category node in the second release information classification tree.

[0175] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0176] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0177] Obtaining a positive training text sample set annotated by a release information classification model of the first category of nodes;

[0178] In response to the labeling operation of the training text in the training text sample set, labeling the training sample;

[0179] A training text sample set for training the release information classification model of the second category node in the second release information classification tree is re-determined according to the labeling result.

[0180] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0181] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0182] Obtaining a positive training text sample set and a negative training text sample set annotated by a release information classification model of the subcategory node;

[0183] The positive training text sample set and the negative training text sample set are both used as negative training text sample sets for training the release information classification model of the second category node in the second release information classification tree.

[0184] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0185] Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including:

[0186] Obtaining a set of negative training text samples annotated by the release information classification model of the first category of nodes;

[0187] The negative training text sample set is used as a negative training text sample set for training a release information classification model of the second category node in the second release information classification tree.

[0188] According to one or more embodiments of the present disclosure, there is provided a data processing device, including:

[0189] A change response module, used to obtain a second release information classification tree in response to a change operation of the first release information classification tree; wherein the first release information classification tree and the second release information classification tree are both composed of a plurality of category nodes, and each category node contains a release information classification model;

[0190] A text acquisition module, used to acquire training text from the landing page of the published information;

[0191] A text annotation module, used to classify the training text according to the published information classification model of each category node in the first published information classification tree, and annotate the published information category of each training text according to the classification result;

[0192] The sample set determination module is used to determine a training text sample set for training a release information classification model for each category node in the second release information classification tree according to the marked training text.

[0193] Furthermore, the sample set determination module is specifically used to determine a training text sample set for training the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next-level category node of the second category node in the second release information classification tree based on the training text annotated by the release information classification model of the first category node in the first release information classification tree; wherein, the first category node is a category node that may be incorrectly annotated; and the second category node is a category node in the second release information classification tree that is the same as the first category node.

[0194] The sample set determination module is specifically used to use the training text annotated by the release information classification model of the third category node in the first release information classification tree as the training text sample set for training the release information classification model of the fourth category node in the second release information classification tree; the third category node is a correctly annotated category node; the fourth category node is the same category node as the third category node in the second release information classification tree.

[0195] Furthermore, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category node is the plurality of subcategory nodes;

[0196] Moreover, the sample set determination module is specifically used to: obtain a positive training text sample set annotated by the publication information classification model of the first category node; annotate the training samples in response to the annotation operation of the training text in the positive training text sample set; and re-determine the positive training text sample set and the negative training text sample set for training the publication information classification model of the second category node and the publication information classification model of each subcategory node in the second publication information classification tree based on the annotation results.

[0197] Furthermore, the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category node is the plurality of subcategory nodes;

[0198] Moreover, the sample set determination module is specifically used to: obtain a negative training text sample set annotated by the publication information classification model of the first category node; and use the negative training text sample set as a negative training text sample set for training the publication information classification model of the second category node and the publication information classification model of each subcategory node in the second publication information classification tree.

[0199] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0200] Moreover, the sample set determination module is specifically used to: obtain a negative training text sample set annotated by the publication information classification model of the first category node; annotate the training samples in response to the annotation operation of the training text in the negative training text sample set; and re-determine the negative training text sample set for training the publication information classification model of the second category node in the second publication information classification tree and the positive training text sample set for the publication information classification model of the subcategory node based on the annotation results.

[0201] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0202] Furthermore, the sample set determination module is specifically used to: obtain a positive training text sample set annotated by the release information classification model of the first category node; in response to the annotation operation of the training text in the positive training text sample set, use the positive training text sample set as a negative training text sample set for training the release information classification model of the subcategory node in the second release information classification tree.

[0203] Furthermore, the change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node;

[0204] Moreover, the sample set determination module is specifically used to: obtain a positive training text sample set annotated by the release information classification model of the first category node; and use the positive training text sample set as a positive training text sample set for training the release information classification model of the second category node in the second release information classification tree.

[0205] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0206] Moreover, the sample set determination module is specifically used to: obtain a positive training text sample set annotated by the release information classification model of the first category node; annotate the training samples in response to the annotation operation of the training text in the positive training text sample set; and re-determine the training text sample set for training the release information classification model of the second category node in the second release information classification tree based on the annotation results.

[0207] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0208] Furthermore, the sample set determination module is specifically used to: obtain a positive training text sample set and a negative training text sample set annotated by the release information classification model of the subcategory node; and use both the positive training text sample set and the negative training text sample set as a negative training text sample set for training the release information classification model of the second category node in the second release information classification tree.

[0209] Furthermore, the change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty;

[0210] Moreover, the sample set determination module is specifically used to: obtain a negative training text sample set annotated by the release information classification model of the first category node; and use the negative training text sample set as a negative training text sample set for training the release information classification model of the second category node in the second release information classification tree.

[0211] According to one or more embodiments of the present disclosure, there is provided an electronic device, including:

[0212] a memory for storing non-transitory computer-readable instructions; and

[0213] The processor is used to run the computer-readable instructions so that the processor implements the above-mentioned data processing method when executing.

[0214] According to one or more embodiments of the present disclosure, a computer-readable storage medium is provided for storing non-transitory computer-readable instructions. When the non-transitory computer-readable instructions are executed by a computer, the computer is caused to execute the above-mentioned data processing method.

[0215] The above description is only a preferred embodiment of the present disclosure and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present disclosure is not limited to the technical solution formed by the specific superposition of the above-mentioned technical features, but should also cover other technical solutions formed by any superposition of the above-mentioned technical features or their equivalent features without departing from the above-mentioned disclosure concept. For example, the above-mentioned features are replaced with the technical features with similar functions disclosed in the present disclosure (but not limited to) to form a technical solution.

[0216] In addition, although each operation is described in a specific order, this should not be understood as requiring these operations to be performed in the specific order shown or in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be interpreted as limiting the scope of the present disclosure. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in a superimposed manner. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-superimposed mode.

[0217] Although the subject matter has been described in language specific to structural features or methodological logical actions, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Instead, the specific features and actions described above are merely example forms of implementing the claims.

Claims

1. A data processing method, characterized in that: include: In response to the change operation of the first release information classification tree, a second release information classification tree is obtained; wherein the first release information classification tree and the second release information classification tree are both composed of a plurality of category nodes, and each category node contains a release information classification model; Obtain training text from the landing page of the published information; Classifying the training texts into published information using the published information classification model of each category node in the first published information classification tree, and marking the published information category of each training text according to the classification result; Determine a set of training text samples for training the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree based on the training text annotated by the release information classification model of the first category node in the first release information classification tree; wherein the first category node is a category node that may be incorrectly annotated; and the second category node is a category node in the second release information classification tree that is the same as the first category node; Wherein, when the change operation is to split the first category node into multiple subcategory nodes, and the corresponding next-level category node is the multiple subcategory nodes, the training text labeled according to the release information classification model of the first category node in the first release information classification tree determines the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next-level category node of the second category node in the second release information classification tree, including: Obtain a positive training text sample set annotated by the publication information classification model of the first category node; annotate the training text in response to the annotation operation of the training text in the positive training text sample set; and re-determine the positive training text sample set and the negative training text sample set for training the publication information classification model of the second category node and the publication information classification model of each subcategory node in the second publication information classification tree based on the annotation results.

2. The method according to claim 1, characterized in that: The step of determining a set of training text samples for training a release information classification model for each category node in the second release information classification tree according to the annotated training texts includes: The training text annotated by the release information classification model of the third category node in the first release information classification tree is used as a training text sample set for training the release information classification model of the fourth category node in the second release information classification tree; the third category node is a correctly annotated category node; the fourth category node is the same category node as the third category node in the second release information classification tree.

3. The method according to claim 1, characterized in that: The change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category node is the plurality of subcategory nodes; Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including: Obtaining a set of negative training text samples annotated by the release information classification model of the first category of nodes; The negative training text sample set is used as a negative training text sample set for training the release information classification model of the second category node and the release information classification model of each subcategory node in the second release information classification tree.

4. The method according to claim 1, characterized in that The change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node; Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including: Obtaining a set of negative training text samples annotated by the release information classification model of the first category of nodes; In response to the labeling operation of the training text in the negative training text sample set, labeling the training sample; The negative training text sample set for training the release information classification model of the second category node and the positive training text sample set for training the release information classification model of the subcategory node in the second release information classification tree are re-determined according to the labeling result.

5. The method according to claim 1, characterized in that The change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node; Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including: Obtaining a positive training text sample set annotated by a release information classification model of the first category of nodes; In response to the labeling operation of the training text in the positive training text sample set, the positive training text sample set is used as a negative training text sample set for training the release information classification model of the subcategory node in the second release information classification tree.

6. The method according to claim 1, characterized in that The change operation is to add a subcategory node by taking the first category node as a parent node, and the corresponding next-level category node is the subcategory node; Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including: Obtaining a positive training text sample set annotated by a release information classification model of the first category of nodes; The positive training text sample set is used as a positive training text sample set for training a release information classification model of the second category node in the second release information classification tree.

7. The method according to claim 1, characterized in that The change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty; Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including: Obtaining a positive training text sample set annotated by a release information classification model of the first category of nodes; In response to the labeling operation of the training text in the training text sample set, labeling the training sample; A training text sample set for training the release information classification model of the second category node in the second release information classification tree is re-determined according to the labeling result.

8. The method according to claim 1, characterized in that The change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty; Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including: Obtaining a positive training text sample set and a negative training text sample set annotated by a release information classification model of the subcategory node; The positive training text sample set and the negative training text sample set are both used as negative training text sample sets for training the release information classification model of the second category node in the second release information classification tree.

9. The method according to claim 1, characterized in that: The change operation is to delete the subcategory node of the first category node, and the corresponding next-level category node is empty; Furthermore, the training text sample set used to train the release information classification model of the second category node in the second release information classification tree and the release information classification model of the next level category node of the second category node in the second release information classification tree is determined based on the training text annotated by the release information classification model of the first category node in the first release information classification tree, including: Obtaining a set of negative training text samples annotated by the release information classification model of the first category of nodes; The negative training text sample set is used as a negative training text sample set for training a release information classification model of the second category node in the second release information classification tree.

10. A data processing device, characterized in that: include: A change response module, used to obtain a second release information classification tree in response to a change operation of the first release information classification tree; wherein the first release information classification tree and the second release information classification tree are both composed of a plurality of category nodes, and each category node contains a release information classification model; A text acquisition module, used to acquire training text from the landing page of the published information; A text annotation module, used to classify the training text according to the published information classification model of each category node in the first published information classification tree, and annotate the published information category of each training text according to the classification result; A sample set determination module, used to determine a training text sample set for training a release information classification model of a second category node in the second release information classification tree and a release information classification model of a next-level category node of the second category node in the second release information classification tree based on training text annotated by the release information classification model of the first category node in the first release information classification tree; wherein the first category node is a category node that may be incorrectly annotated; and the second category node is a category node in the second release information classification tree that is the same as the first category node; Wherein, when the change operation is to split the first category node into a plurality of subcategory nodes, and the corresponding next-level category node is the plurality of subcategory nodes, the sample set determination module is specifically used to: Obtain a positive training text sample set annotated by the publication information classification model of the first category node; annotate the training text in response to the annotation operation of the training text in the positive training text sample set; and re-determine the positive training text sample set and the negative training text sample set for training the publication information classification model of the second category node and the publication information classification model of each subcategory node in the second publication information classification tree based on the annotation results.

11. An electronic device, comprising: a memory for storing non-transitory computer-readable instructions; as well as A processor is used to run the computer-readable instructions so that the processor implements the data processing method according to any one of claims 1 to 9 when executing.

12. A computer-readable storage medium for storing non-transitory computer-readable instructions, which, when executed by a computer, causes the computer to execute the data processing method according to any one of claims 1 to 9.

Citation Information

Patent Citations

  • Data classification method, device and equipment based on multi-tree

    CN107330021A