Material label processing method, device, electronic equipment and medium

Abstract

The present disclosure provides a material label processing method, device, electronic device and medium, which relate to the field of artificial intelligence, especially to technical fields such as big data, intelligent recommendation, and natural language processing. The specific implementation scheme is as follows: obtaining multiple candidate material labels associated with multiple materials; generating a heterogeneous graph based on multiple materials and multiple candidate material labels associated with multiple materials, the heterogeneous graph is used to characterize the association relationship between the candidate material labels and the corresponding materials; based on the heterogeneous graph, determining a first preset number of candidate material labels associated with each material from the multiple candidate material labels associated with the multiple materials; for each material, determining the first category information corresponding to each of the first preset number of candidate material labels; and processing the first preset number of candidate material labels based on the first category information to obtain a first material label set that matches the material.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, particularly to artificial intelligence, big data, intelligent recommendation, natural language processing, and other technical fields. Specifically, it relates to a material label processing method and device, electronic equipment, storage medium, and computer program product. Background Art

[0002] Material tags describe key information about video materials. Users can use these tags to search for relevant video materials from a library and use them for video editing. However, the material tags provided in related technologies lack diversity and relevance to video materials. This inadequately meets user search requirements and reduces user experience. Summary of the Invention

[0003] The present disclosure provides a material tag processing method and device, an electronic device, a storage medium, and a computer program product.

[0004] According to one aspect of the present disclosure, a material label processing method is provided, including: obtaining multiple candidate material labels associated with multiple materials; generating a heterogeneous graph based on the multiple materials and the multiple candidate material labels associated with the multiple materials, the heterogeneous graph being used to represent the association relationship between the candidate material labels and the corresponding materials; determining, based on the heterogeneous graph, a first preset number of candidate material labels associated with each material from the multiple candidate material labels associated with the multiple materials; determining, for each material, first category information corresponding to each of the first preset number of candidate material labels; and processing the first preset number of candidate material labels based on the first category information to obtain a first material label set that matches the material.

[0005] According to another aspect of the present disclosure, a material label processing device is provided, including: an acquisition module for acquiring multiple candidate material labels associated with multiple materials; a first generation module for generating a heterogeneous graph based on the multiple materials and the multiple candidate material labels associated with the multiple materials, the heterogeneous graph being used to characterize the association relationship between the candidate material labels and the corresponding materials; a first determination module for determining, based on the heterogeneous graph, a first preset number of candidate material labels associated with each material from the multiple candidate material labels associated with the multiple materials; a second determination module for determining, for each material, first category information corresponding to each of the first preset number of candidate material labels; and a first processing module for processing the first preset number of candidate material labels based on the first category information to obtain a first material label set that matches the material.

[0006] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method provided according to the present disclosure.

[0007] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided. The computer instructions are used to cause a computer to execute the method provided according to the present disclosure.

[0008] According to another aspect of the present disclosure, a computer program product is provided, comprising a computer program, which implements the method provided according to the present disclosure when executed by a processor.

[0009] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0011] Figure 1 is a schematic diagram of an exemplary system architecture to which the material label processing method and apparatus according to an embodiment of the present disclosure can be applied;

[0012] Figure 2 is a flow chart of a material tag processing method according to an embodiment of the present disclosure;

[0013] Figure 3 is a schematic diagram of a method for obtaining multiple candidate material labels according to an embodiment of the present disclosure;

[0014] Figures 4A to 4C is a schematic diagram of a method for obtaining a first material tag set according to an embodiment of the present disclosure;

[0015] Figure 5 is a schematic diagram of a process for determining a feature vector of a newly added material according to an embodiment of the present disclosure;

[0016] Figure 6 is a block diagram of a material tag processing device according to an embodiment of the present disclosure; and

[0017] Figure 7 4 is a block diagram of an electronic device for implementing the material tag processing method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0018] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0019] Figure 1 is a schematic diagram of an exemplary system architecture to which the material tag processing method and apparatus according to an embodiment of the present disclosure can be applied. Figure 1 The examples shown are merely examples of system architectures to which the embodiments of the present disclosure may be applied, to help those skilled in the art understand the technical content of the present disclosure, but do not mean that the embodiments of the present disclosure may not be used in other devices, systems, environments or scenarios.

[0020] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0021] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103. For example, web browser applications, search applications, instant messaging tools, email clients, or social platform software, etc. (only as examples).

[0022] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0023] The server 105 may be a server that provides various services, such as a background management server (for example only) that supports websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process received data such as user requests, and feed back processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0024] For example, server 105 can obtain multiple materials and multiple candidate material tags associated with each of the multiple materials from terminal devices 101, 102, and 103 via network 104. Subsequently, a heterogeneous graph is generated based on the multiple materials and the multiple candidate material tags associated with each of the multiple materials. Subsequently, based on the heterogeneous graph, a first preset number of candidate material tags associated with each of the multiple materials is determined from the multiple candidate material tags associated with each of the multiple materials, and first category information corresponding to each of the first preset number of candidate material tags is determined for each of the materials. Then, based on the first category information, the first preset number of candidate material tags are processed to obtain a first material tag set that matches the materials.

[0025] It should be noted that the material tag processing method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the material tag processing device provided in the embodiment of the present disclosure can generally be set in the server 105.

[0026] Alternatively, the material tag processing method provided in the embodiment of the present disclosure may also be executed by a server or server cluster that is different from the server 105 and that is capable of communicating with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the material tag processing apparatus provided in the embodiment of the present disclosure may also be set in a server or server cluster that is different from the server 105 and that is capable of communicating with the terminal devices 101, 102, 103 and / or the server 105.

[0027] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0028] It should be noted that the sequence numbers of the operations in the following method are only used to indicate the operation for the purpose of description, and should not be regarded as indicating the order in which the operations should be performed. Unless explicitly stated, the method does not need to be performed in the order shown.

[0029] Figure 2 is a flowchart of a material tag processing method according to an embodiment of the present disclosure.

[0030] like Figure 2 As shown, the material tag processing method 200 may include operations S210 to S250.

[0031] In operation S210 , a plurality of candidate material tags associated with respective ones of the plurality of materials are acquired.

[0032] In operation S220 , a heterogeneous graph is generated based on the multiple materials and the multiple candidate material tags associated with each of the multiple materials. The heterogeneous graph is used to represent the association relationship between the candidate material tags and the corresponding materials.

[0033] In operation S230 , based on the heterogeneous graph, a first preset number of candidate material labels associated with each material is determined from a plurality of candidate material labels associated with each of the plurality of materials.

[0034] In operation S240 , for each material, first category information corresponding to each of a first preset number of candidate material tags is determined.

[0035] In operation S250 , a first preset number of candidate material labels are processed based on the first category information to obtain a first material label set that matches the material.

[0036] According to an embodiment of the present disclosure, the material library may include a large amount of materials, such as video clips or video frame images obtained by intercepting video content corresponding to video data.

[0037] The video content corresponding to the video data is usually represented by description information and video tags. In this way, the video content corresponding to the corresponding video data can be quickly grasped through the description information and video tags.

[0038] Here, the description information may include, for example, a title associated with the video data, source information of the video data, and descriptive text related to the video content. The description text is relatively long. The description text includes multiple words, and based on the meanings of the multiple words and the part of speech of each word in the description text, a comprehensive description of the video content of the video data can be completed.

[0039] A video can have one or more video tags. Each video tag is relatively short, independent of the order or meaning of the tags. Video tags can be used to describe entities within the video content, such as people, events, and locations.

[0040] Since the materials in the material library are obtained by intercepting the video content corresponding to the video data, the candidate material tags of each material can be obtained based on the mapping relationship between the video data and the materials using the description information of the video data and the video tags.

[0041] In an embodiment of the present disclosure, a heterogeneous graph may be generated based on multiple materials and multiple candidate material tags corresponding to each of the multiple materials. The heterogeneous graph may be used to represent the association relationship between the candidate material tags and the corresponding materials.

[0042] For example, each material may be used as a material node, each candidate material label may be used as a candidate material label node, and each node may be connected based on the relationship between each material and each candidate material label, thereby obtaining a heterogeneous graph.

[0043] It can be understood that since the material is obtained by intercepting the video content corresponding to the video data, the candidate material labels obtained using the description information of the video data and the video label usually have a certain deviation from the material content corresponding to the material, that is, the correlation between these candidate material labels and the material has a certain deviation. In other words, in the heterogeneous graph constructed based on these candidate material labels and the material, there may be candidate material label nodes that are not related to the material node but actually have established a connection relationship with the material node, or candidate material label nodes that have a correlation with the material node have not established a connection relationship with the material node. Therefore, it is necessary to further process the node data (including material, candidate material labels) corresponding to each node in the heterogeneous graph in the future to improve the correlation between each node.

[0044] In the embodiment of the present disclosure, based on the heterogeneous graph, a first preset number of candidate material labels associated with each material may be determined from a plurality of candidate material labels associated with each of the plurality of materials.

[0045] The first preset number of candidate material tags can be considered as a set of candidate material tags with potential associations with the material, obtained by searching among the candidate material tags corresponding to all materials. This avoids missing candidate material tags with potential associations with the material, thereby ensuring the diversity and accuracy of the material tags corresponding to the material.

[0046] Afterwards, for each material, first category information corresponding to each of a first preset number of candidate material tags may be determined. The first category information is used to characterize the category attributes of the candidate material tags, such as person, event, food, organization, etc.

[0047] For example, in the embodiment of the present disclosure, a pre-trained named entity recognition model can be used to determine the first category information corresponding to each of the first preset number of candidate material tags. The pre-trained named entity recognition model can be selected according to actual conditions and is not limited here.

[0048] Next, based on the first category information, a first preset number of candidate material tags are further screened to obtain a first material tag set that matches the material. The first material tag set includes multiple first material tags. Each first material tag has a real association with the material.

[0049] According to an embodiment of the present disclosure, by constructing a heterogeneous graph using multiple materials and multiple candidate material labels associated with each of the multiple materials, a preliminary association relationship between each material and each candidate material label can be established. Subsequently, based on the heterogeneous graph, a first preset number of candidate material labels associated with each material are determined from the multiple candidate material labels associated with each of the multiple materials, thereby ensuring the diversity and accuracy of the material labels corresponding to the materials. Subsequently, based on the first preset number of candidate material labels, these candidate material labels are finely screened based on the first type of information, thereby improving the accuracy and precision of the material labels corresponding to the materials.

[0050] In an embodiment of the present disclosure, multiple candidate material tags may be acquired in the following manner.

[0051] For example, in one example, for each material, at least one first candidate material tag associated with the material may be determined from a plurality of to-be-processed material tags contained in a material library, and each of the at least one first candidate material tag may be determined as a candidate material tag.

[0052] In the embodiment of the present disclosure, the multiple to-be-processed material tags contained in the material library refer to potential material tags corresponding to the material obtained by using the description information of the video data and the video tags.

[0053] For example, the video tag corresponding to the video data can be used as the material tag to be processed. For example, the description information of the video data can be segmented, and the keywords obtained by the segmentation can be determined as the material tag to be processed.

[0054] As previously described, since the candidate material tags obtained using the description information of the video data and the video tags usually have a certain deviation from the material content corresponding to the material, the relevance between the above-mentioned multiple material tags to be processed and the material has a certain deviation.

[0055] To improve the accuracy of the candidate material labels, in the disclosed embodiments, a multimodal pre-trained model can be used to determine the similarity between the material and each material label to be processed. The material labels to be processed are then sorted based on the similarity, and at least one first candidate material label associated with the material is obtained based on the similarity sorting results. This at least one first candidate material label is then used as the candidate material label.

[0056] In another example, in order to further enrich the candidate material tags and improve the diversity of the candidate material tags, a tag library can also be used to obtain multiple candidate material tags.

[0057] Figure 3Schematic diagram of a method for obtaining multiple candidate material labels according to an embodiment of the present disclosure. Figure 3 An example is given to illustrate the process of obtaining multiple candidate material labels.

[0058] like Figure 3 As shown, for each material 301 , at least one first candidate material tag 304 associated with the material 301 is determined from a plurality of to-be-processed material tags 302 contained in the material library.

[0059] For example, the similarity between the material 301 and each material tag to be processed is determined, the material tags to be processed are sorted based on the similarity, and at least one first candidate material tag 304 associated with the material 301 is obtained based on the similarity sorting result 303 .

[0060] In addition, description information 305 associated with the material 301 may be obtained, and word segmentation processing may be performed on the description information 305 to obtain a plurality of keywords 306 .

[0061] For example, the title associated with the video data and the description text related to the video content included in the description information 305 can be segmented to obtain multiple keywords 306. These keywords have a potential association with the material content, and candidate material tags can be screened from these keywords.

[0062] Afterwards, multiple keywords 306 are matched with a preset tag library 307. If a keyword is matched in the tag library 307, the keyword is determined as a second candidate material tag 308 associated with the material 301. If a keyword is not matched in the tag library 307, the keyword is discarded.

[0063] The preset tag library 307 includes tag sets of multiple tag types, such as a people tag set and a food tag set. These tag sets can be selected and configured based on actual application needs. Each tag set includes most of the material tags associated with that tag type. It can be assumed that all material tags in the tag library 307 encompass all the material tag types desired by the user.

[0064] In the embodiment of the present disclosure, in order to improve the relevance of the candidate material tags, multiple keywords 306 may be matched with the material tags in the tag library 307 to determine whether each keyword can be supplemented as a potential candidate material tag for the material 301 .

[0065] Next, at least one first candidate material tag 304 and each second candidate material tag 308 are determined as a plurality of candidate material tags 309. In this way, the candidate material tags can be further enriched and the diversity of the candidate material tags can be improved.

[0066] According to an embodiment of the present disclosure, based on the first category information, multiple strategies can be used to process the first preset number of candidate material tags to obtain a first material tag set that matches the material. Figures 4A to 4C The process of obtaining the first material tag set is described with an example.

[0067] Figures 4A to 4C 3 is a schematic diagram of a method for obtaining a first material tag set according to an embodiment of the present disclosure.

[0068] In one example, if it is determined that the first category information belongs to the first category attribute, and the first category attribute includes, for example, non-person category attributes such as food, organization, and event, then a first strategy can be used to process a first preset number of candidate material labels to obtain a first material label set that matches the material.

[0069] like Figure 4A As shown, for example, based on the first strategy, the matching degree 411 between the material 401 and the first preset number of candidate material tags 410 can be determined. Then, the first preset number of candidate material tags 410 are sorted according to the matching degree 411, and a first material tag set 4131 is obtained based on the sorting result 412.

[0070] In another example, if it is determined that the first category information belongs to the second category attribute, and the second category attribute includes the category attribute of the person class, a second strategy can be used to process the first preset number of candidate material tags to obtain a first material tag set that matches the material.

[0071] like Figure 4B and Figure 4C As shown, for example, based on the second strategy, the target object in the material 401 is identified to obtain a recognition result 414 and a recognition confidence level (not shown in the figure) for the target object. For example, a target recognition algorithm can be used to identify the target object in the material 401 to obtain a recognition result 414 and a recognition confidence level for the target object. The recognition confidence level is used to represent the confidence level corresponding to the recognition result 414 when identifying the target object.

[0072] If it is determined that the recognition confidence is less than the first preset confidence, the recognition result 414 for the target object is matched with the first preset number of candidate material tags 410. If it is determined that the recognition result 414 matches at least one candidate material tag 415 in the first preset number of candidate material tags 410, a first material tag set 4132 (e.g., Figure 4BThe first preset reliability can be set according to specific circumstances, and this disclosure does not limit this.

[0073] If it is determined that the recognition confidence is greater than or equal to the first preset confidence, a first material label set 4133 (such as Figure 4C shown).

[0074] It can be understood that if the recognition confidence level is greater than or equal to the first preset confidence level, then recognition result 414 for the target object is relatively accurate, and the first preset number of candidate material labels 410 are selected based on material 401 from multiple candidate material labels associated with multiple materials. Therefore, it can be assumed that if recognition result 414 is relatively accurate, these candidate material labels are also accurate, and recognition result 414 and the first preset number of candidate material labels 410 can be directly added to first material label set 4133.

[0075] Based on the above manner, the first preset number of candidate material labels can be processed using a corresponding screening strategy based on the first category information, thereby improving the accuracy and precision of the first material label set.

[0076] In some embodiments, when it is determined that the first category information belongs to the second category attribute (person category attribute), the knowledge data may be used to expand the plurality of first material tags included in the first material tag set to increase the diversity of the first material tags.

[0077] For example, the first knowledge data may be acquired based on description information associated with the material.

[0078] In the embodiment of the present disclosure, the description information associated with the material refers to the description information of the video data corresponding to the material. The description information of the video data is the same or similar to the definition described above and will not be repeated here.

[0079] For example, first knowledge data associated with the target object in the material can be obtained based on the title associated with the video data, the source information of the video data, and the descriptive text related to the video content contained in the above description information. The first knowledge data may include, for example, the film and television works associated with the target object, the corresponding virtual characters in the film and television works, etc.

[0080] Then, based on the first knowledge data, first additional material tags corresponding to each of the plurality of first material tags are obtained, wherein the first knowledge data is associated with the first material tags.

[0081] For example, the first knowledge data can be matched with each first material tag. If a first material tag is determined to match the first knowledge data, the first knowledge data can be determined as the first additional material tag corresponding to the first material tag. For example, if the first material tag represents a person's name, and the first knowledge data includes film and television works related to the person's name, the first knowledge data can be used as the first additional material tag for the first material tag and added to the first material tag set.

[0082] In addition, additional screening may be performed on the candidate material tags in the first preset number of candidate material tags based on the first knowledge data to ensure that all candidate material tags related to the material are screened out.

[0083] For example, the first knowledge data is matched against each candidate material tag in a first preset number of candidate material tags. If a candidate material tag corresponding to the first knowledge data is matched in the first preset number of candidate material tags, a second additional material tag is determined by combining the first knowledge data with the corresponding candidate material tag. The second additional material tag is then added to the first material tag set.

[0084] In some embodiments, deduplication processing may also be performed on the first material tags in the first material tag set to remove duplicate first material tags.

[0085] In an embodiment of the present disclosure, based on the heterogeneous graph, determining a first preset number of candidate material labels associated with each material from a plurality of candidate material labels associated with each of the plurality of materials may include the following operations.

[0086] For example, feature extraction is performed on the node data of each node in the heterogeneous graph to obtain a first feature vector corresponding to each material node and a second feature vector corresponding to each candidate material label node. Exemplarily, a pre-trained graph neural network model can be used to perform feature extraction on the node data of each node in the heterogeneous graph to obtain a feature vector corresponding to each node.

[0087] Then, for each material, based on the similarity between the first feature vector and each second feature vector, a first preset number of candidate material tags associated with the material are determined from the plurality of candidate material tags associated with each of the multiple materials. This allows a set of candidate material tags potentially associated with the material to be selected from the candidate material tags corresponding to all materials, thereby avoiding omission of candidate material tags potentially associated with the material and ensuring the diversity and accuracy of the material tags associated with the material.

[0088] The materials in the material library may be updated in real time. In the embodiment of the present disclosure, material tags can also be recalled for newly added materials in the material library.

[0089] Because the number of newly added materials in the material library is very large, recalling material labels for all materials in the material library would reduce processing efficiency. To address this problem, in the disclosed embodiments, a neighbor sampling algorithm can be used to construct a relationship graph containing the newly added materials. Based on this relationship graph, the feature vectors of the newly added materials are determined. Material labels associated with the newly added materials are then recalled based on the feature vectors of the newly added materials. This reduces computational complexity and improves processing efficiency.

[0090] Figure 5 FIG. 1 is a schematic diagram of a process for determining a feature vector of a newly added material according to an embodiment of the present disclosure. Figure 5 The process of determining the feature vector of the newly added material is described by example. It should be noted that, for the convenience of description, Figure 5 In the following description, the material will be represented by its Uniform Resource Locator (URL). It should be understood that the technical solution of the present disclosure is not limited to this.

[0091] like Figure 5 As shown, when obtaining new material (such as new material URL-new) and at least one new material tag corresponding to the new material (such as tagA, tagB,..., tagN, just an example), at least one new material tag can be matched with the preset tag library and heterogeneous graph respectively.

[0092] For example, each new material tag (e.g., tagA, tagB, ..., tagN) in the new material tag can be matched with the candidate material tags corresponding to each candidate material tag node in the heterogeneous graph, and each new material tag can be matched with each material tag in the tag library. If a certain new material tag(s) is matched in the tag library and the same new material tag(s) is matched in the heterogeneous graph, a relationship graph containing the new material is generated based on the new material, the new material tags, and the heterogeneous graph based on the neighbor sampling algorithm.

[0093] For example, for newly added material tags tagA to tagN, if the newly added material tags tagA and tagB are matched in the tag library, and the newly added material tags tagA and tagB are matched in the heterogeneous graph, then based on the neighbor sampling algorithm, according to the newly added material (such as the newly added material URL-new), the newly added material tags (such as tagA and tagB) and the heterogeneous graph, a relationship graph containing the newly added material (such as Figure 5 shown are examples only).

[0094] For example, Figure 5 In the relationship diagram shown, URL1 through URL3 represent material nodes sampled from the heterogeneous graph and associated with the newly added material URL-new, and tag1 through tag5 represent candidate material tag nodes sampled from the heterogeneous graph and associated with URL1 through URL3, respectively. It should be noted that when generating a relationship diagram containing newly added materials based on the neighbor sampling algorithm, the number of sampling layers and the number of samples per layer can be set based on actual needs, and this disclosure does not impose any restrictions on this.

[0095] Next, the relationship graph can be processed using a neighbor aggregation algorithm with the newly added material as the target node to generate a feature vector for the newly added material.

[0096] For example, the newly added material URL-new in the relationship graph can be used as the target node, and the vectors of the target node's neighbor nodes can be aggregated using a neighbor aggregation algorithm to generate a feature vector for the target node, i.e., the feature vector of the newly added material URL-new. In the disclosed embodiments, neighbor aggregation algorithms can include, for example, average aggregation, pooling aggregation, and LSTM aggregation, and the specific algorithm can be selected based on actual needs.

[0097] After determining the feature vector of the newly added material, a second preset number of candidate material labels associated with the newly added material can be determined from multiple candidate material labels associated with multiple materials based on the similarity between the feature vector of the newly added material and the second feature vector corresponding to each candidate material label node in the heterogeneous graph.

[0098] Next, the second category information corresponding to each of the second preset number of candidate material tags is determined. The second category information has the same or similar definition as the first category information, and will not be described in detail here.

[0099] Afterwards, based on the second category information, a second preset number of candidate material labels are processed to obtain a second material label set that matches the newly added material.

[0100] In the embodiment of the present disclosure, the process of processing the second preset number of candidate material labels based on the second category information is the same as or similar to the process of processing the first preset number of candidate material labels based on the first category information.

[0101] For example, in one example, if it is determined that the second category information belongs to the first category attributes, and the first category attributes include, for example, non-person category attributes such as food, organization, event, etc., then the first strategy can be used to process the second preset number of candidate material labels to obtain a second material label set that matches the newly added material.

[0102] For example, based on the first strategy, the matching degree between the newly added material and the second preset number of candidate material labels can be determined respectively. Then, the second preset number of candidate material labels are sorted according to the matching degree, and a second material label set is obtained based on the sorting result.

[0103] In another example, if it is determined that the second category information belongs to the second category attribute, and the second category attribute includes the category attribute of the person class, a second strategy can be used to process a second preset number of candidate material tags to obtain a second material tag set that matches the newly added material.

[0104] For example, based on the second strategy, the target object in the newly added material is identified to obtain a recognition result and a recognition confidence level for the target object.

[0105] If it is determined that the recognition confidence is less than the second preset confidence level, the recognition result for the target object is matched against a second preset number of candidate material labels. If it is determined that the recognition result matches at least one candidate material label from the second preset number of candidate material labels, a second material label set is obtained based on the recognition result and the at least one candidate material label. The second preset confidence level may be the same as or different from the first preset confidence level. The second preset confidence level may be set based on actual circumstances and is not limited in this disclosure.

[0106] If it is determined that the recognition confidence is greater than or equal to the second preset confidence, a second material label set is obtained according to the recognition result of the target object and a second preset number of candidate material labels.

[0107] Based on the above manner, the second preset number of candidate material labels can be processed using a corresponding screening strategy based on the second category information, thereby improving the accuracy and precision of the second material label set.

[0108] In some embodiments, when it is determined that the second category information belongs to the second category attribute (person category attribute), the knowledge data may be used to expand the plurality of second material tags included in the second material tag set to increase the diversity of the second material tags.

[0109] For example, the second knowledge data may be acquired based on description information associated with the newly added material.

[0110] In the embodiment of the present disclosure, the description information associated with the newly added material is the same as or similar to the definition described above and will not be repeated here.

[0111] For example, second knowledge data associated with the target object in the newly added material can be obtained based on the description information. The second knowledge data may include, for example, film and television works associated with the target object, corresponding virtual characters in the film and television works, and the like.

[0112] Then, based on the second knowledge data, a third additional material tag corresponding to each of the plurality of second material tags is obtained, and the third additional material tag is added to the second material tag set, wherein the second knowledge data is associated with the second material tag.

[0113] It should be noted that the process of obtaining the third additional material tag is similar to the process of obtaining the first additional material tag, and will not be repeated here.

[0114] In addition, based on the second knowledge data, additional screening may be performed on the candidate material tags in the second preset number of candidate material tags to ensure that all candidate material tags related to the newly added material are screened out.

[0115] For example, the second knowledge data is matched against each candidate material tag in a second preset number of candidate material tags. If a candidate material tag corresponding to the second knowledge data is matched in the second preset number of candidate material tags, a fourth additional material tag is determined by combining the second knowledge data with the corresponding candidate material tag. The fourth additional material tag is then added to the second material tag set.

[0116] In some embodiments, deduplication processing may also be performed on the second material tags in the second material tag set to remove duplicate second material tags.

[0117] Figure 6 is a block diagram of a material tag processing device according to an embodiment of the present disclosure.

[0118] like Figure 6 As shown, the material tag processing device 600 includes: an acquisition module 610 , a first generation module 620 , a first determination module 630 , a second determination module 640 and a first processing module 650 .

[0119] The acquisition module 610 is configured to acquire a plurality of candidate material tags associated with each of the plurality of materials.

[0120] The first generating module 620 is configured to generate a heterogeneous graph based on a plurality of materials and a plurality of candidate material tags associated with each of the plurality of materials. The heterogeneous graph is configured to represent an association relationship between the candidate material tags and the corresponding materials.

[0121] The first determining module 630 is configured to determine, based on the heterogeneous graph, a first preset number of candidate material labels associated with each material from a plurality of candidate material labels associated with each of the plurality of materials.

[0122] The second determining module 640 is configured to determine, for each material, first category information corresponding to each of a first preset number of candidate material labels.

[0123] The first processing module 650 is configured to process a first preset number of candidate material labels based on the first category information to obtain a first material label set that matches the material.

[0124] According to an embodiment of the present disclosure, the first processing module 650 includes: a first determining unit and a sorting unit. The first determining unit is configured to, in response to determining that the first category information belongs to the first category attribute, determine the degree of match between the material and a first preset number of candidate material labels; and the sorting unit is configured to sort the first preset number of candidate material labels according to the degree of match, and obtain a first material label set based on the sorting result.

[0125] According to an embodiment of the present disclosure, the first processing module 650 includes: an identification unit, a first matching unit, and a second determination unit. The identification unit is configured to, in response to determining that the first category information belongs to the second category attribute, identify the target object in the material and obtain an identification result and an identification confidence for the target object; the first matching unit is configured to, in response to determining that the identification confidence is less than a preset confidence, match the identification result for the target object with a first preset number of candidate material labels; and the second determination unit is configured to, in response to determining that the identification result matches at least one candidate material label from the first preset number of candidate material labels, obtain a first material label set based on the identification result and the at least one candidate material label.

[0126] According to an embodiment of the present disclosure, the first processing module 650 further includes a third determining unit configured to, in response to determining that the recognition confidence is greater than or equal to a preset confidence, obtain a first material label set based on the recognition result for the target object and a first preset number of candidate material labels.

[0127] According to an embodiment of the present disclosure, the first material tag set includes multiple first material tags; the first processing module 650 further includes: a first acquisition unit, a second acquisition unit, a third acquisition unit, and an adding unit. The first acquisition unit is used to acquire first knowledge data; wherein the first knowledge data is acquired based on descriptive information associated with the material; the second acquisition unit is used to acquire, based on the first knowledge data, a first additional material tag corresponding to each of the multiple first material tags; wherein the first knowledge data is associated with the first material tag; the third acquisition unit is used to acquire, based on the first knowledge data, a second additional material tag corresponding to each of a first preset number of candidate material tags; and the adding unit is used to add each first additional material tag and each second additional material tag to the first material tag set.

[0128] According to an embodiment of the present disclosure, the first determination module 630 includes: an extraction unit and a fourth determination unit. The extraction unit is configured to perform feature extraction on node data of each node in the heterogeneous graph to obtain a first feature vector corresponding to each material node and a second feature vector corresponding to each candidate material label node; and the fourth determination unit is configured to determine, for each material, a first preset number of candidate material labels associated with the material from a plurality of candidate material labels associated with each of the multiple materials based on the similarity between the first feature vector and each second feature vector.

[0129] According to an embodiment of the present disclosure, the material label processing device 600 further includes: a matching module, a second generation module, an aggregation module, a third determination module, a fourth determination module, and a second processing module. The matching module is configured to, in response to obtaining a new material and a new material label corresponding to the new material, match the new material label with a preset label library and a heterogeneous graph, respectively; the second generation module is configured to, in response to matching the new material label in the label library and the heterogeneous graph, generate a relationship graph containing the new material based on the new material, the new material label, and the heterogeneous graph using a neighbor sampling algorithm; the aggregation module is configured to use the new material as a target node and process the relationship graph using a neighbor aggregation algorithm to generate a feature vector for the new material; the third determination module is configured to determine, based on the similarity between the feature vector of the new material and each second feature vector, a second preset number of candidate material labels associated with the new material from a plurality of candidate material labels associated with each of the multiple materials; the fourth determination module is configured to determine second category information corresponding to each of the second preset number of candidate material labels; and the second processing module is configured to process the second preset number of candidate material labels based on the second category information to obtain a second material label set matching the new material.

[0130] According to an embodiment of the present disclosure, the acquisition module 610 includes: a fifth determination unit and a sixth determination unit. The fifth determination unit is configured to determine, for each material, at least one first candidate material tag associated with the material from a plurality of to-be-processed material tags contained in the material library; and the sixth determination unit is configured to determine each of the at least one first candidate material tags as a candidate material tag.

[0131] According to an embodiment of the present disclosure, the acquisition module 610 further includes: a fourth acquisition unit, a word segmentation unit, a second matching unit, a seventh determination unit, and an eighth determination unit. The fourth acquisition unit is configured to acquire, for each material, description information associated with the material; the word segmentation unit is configured to perform word segmentation processing on the description information to obtain multiple keywords; the second matching unit is configured to match the multiple keywords with a preset tag library; the seventh determination unit is configured to, in response to matching a keyword in the tag library, determine the keyword as a second candidate material tag associated with the material; and the eighth determination unit is configured to determine the at least one first candidate material tag and each second candidate material tag as a plurality of candidate material tags.

[0132] It should be noted that the implementation methods, technical problems solved, functions implemented, and technical effects achieved of each module / unit / sub-unit in the device part embodiment are the same or similar to the implementation methods, technical problems solved, functions implemented, and technical effects achieved of each corresponding step in the method part embodiment, and will not be repeated here.

[0133] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision, disclosure and application of the data involved (for example, including but not limited to user personal information) shall comply with the provisions of relevant laws and regulations and shall not violate public order and good morals.

[0134] In the technical solution disclosed herein, authorization or consent from the data owner is obtained before obtaining or collecting relevant data.

[0135] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0136] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute a method as an embodiment of the present disclosure.

[0137] According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute a method according to an embodiment of the present disclosure.

[0138] According to an embodiment of the present disclosure, a computer program product includes a computer program. When the computer program is executed by a processor, the method according to the embodiment of the present disclosure is implemented.

[0139] Figure 7A schematic block diagram of an example electronic device 700 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0140] like Figure 7 As shown, the device 700 includes a computing unit 701, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 702 or a computer program loaded from a storage unit 708 into a random access memory (RAM) 703. Various programs and data required for the operation of the device 700 can also be stored in the RAM 703. The computing unit 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0141] Various components in device 700 are connected to I / O interface 705, including an input unit 706, such as a keyboard, mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, optical disk, etc.; and a communication unit 709, such as a network card, modem, wireless communication transceiver, etc. The communication unit 709 allows device 700 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0142] The computing unit 701 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the computing unit 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 701 performs the various methods and processes described above, such as the material labeling processing method. For example, in some embodiments, the material labeling processing method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the computing unit 701, one or more steps of the material labeling processing method described above can be performed. Alternatively, in other embodiments, the computing unit 701 can be configured to perform the material labeling processing method in any other appropriate manner (e.g., via firmware).

[0143] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0144] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0145] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can 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 combination of the foregoing.

[0146] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0147] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0148] Computer systems may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The client and server relationship arises through computer programs running on the respective computers and having a client-server relationship to each other.

[0149] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not limited herein.

[0150] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A material label processing method, comprising: Obtaining multiple candidate material labels associated with each of the multiple materials; generating a heterogeneous graph based on the multiple materials and multiple candidate material labels associated with each of the multiple materials, wherein the heterogeneous graph is used to represent associations between the candidate material labels and corresponding materials; Based on the heterogeneous graph, determining a first preset number of candidate material labels associated with each material from a plurality of candidate material labels associated with each of the plurality of materials; For each material, determining first category information corresponding to each of the first preset number of candidate material labels; as well as Based on the first category information, the first preset number of candidate material labels are processed to obtain a first material label set that matches the material, including: in response to determining that the first category information belongs to the second category attribute, identifying a target object in the material to obtain a recognition result and a recognition confidence for the target object; in response to determining that the recognition confidence is less than a preset confidence, matching the recognition result for the target object with the first preset number of candidate material labels respectively; and in response to determining that the recognition result matches at least one candidate material label among the first preset number of candidate material labels, obtaining the first material label set based on the recognition result and the at least one candidate material label; Wherein, the material is video material.

2. The method according to claim 1, wherein The step of processing the first preset number of candidate material labels based on the first category information to obtain a first material label set matching the material includes: In response to determining that the first category information belongs to a first category attribute, determining the matching degree between the material and the first preset number of candidate material labels respectively; and The first preset number of candidate material tags are sorted according to the matching degree, and the first material tag set is obtained based on the sorting result.

3. The method according to claim 1, wherein The step of processing the first preset number of candidate material labels based on the first category information to obtain a first material label set matching the material further includes: In response to determining that the recognition confidence is greater than or equal to the preset confidence, the first material label set is obtained according to the recognition result for the target object and the first preset number of candidate material labels.

4. The method according to claim 3, wherein: The first material tag set includes a plurality of first material tags; and the step of processing the first preset number of candidate material tags based on the first category information to obtain a first material tag set matching the material further includes: Acquire first knowledge data; wherein the first knowledge data is acquired based on description information associated with the material; Based on the first knowledge data, obtaining a first additional material tag corresponding to each of the plurality of first material tags; wherein the first knowledge data is associated with the first material tag; Based on the first knowledge data, obtaining second additional material tags corresponding to each of the first preset number of candidate material tags; and Each of the first additional material tags and each of the second additional material tags are added to the first material tag set.

5. The method according to claim 1, wherein The first material tag set includes a plurality of first material tags; and the step of processing the first preset number of candidate material tags based on the first category information to obtain a first material tag set matching the material further includes: Acquire first knowledge data; wherein the first knowledge data is acquired based on description information associated with the material; Based on the first knowledge data, obtaining a first additional material tag corresponding to each of the plurality of first material tags; wherein the first knowledge data is associated with the first material tag; Based on the first knowledge data, obtaining second additional material tags corresponding to each of the first preset number of candidate material tags; and Each of the first additional material tags and each of the second additional material tags are added to the first material tag set.

6. The method according to any one of claims 1 to 5, wherein The determining, based on the heterogeneous graph, a first preset number of candidate material labels associated with each material from the plurality of candidate material labels associated with each of the plurality of materials includes: Performing feature extraction on the node data of each node in the heterogeneous graph to obtain a first feature vector corresponding to each material node and a second feature vector corresponding to each candidate material label node; and For each material, based on the similarity between the first feature vector and each of the second feature vectors, a first preset number of candidate material labels associated with the material are determined from the plurality of candidate material labels associated with each of the plurality of materials.

7. The method according to claim 6, further comprising: In response to acquiring a new material and a new material tag corresponding to the new material, matching the new material tag with a preset tag library and the heterogeneous graph respectively; In response to matching the new material label in the label library and matching the new material label in the heterogeneous graph, generating a relationship graph including the new material based on the new material, the new material label, and the heterogeneous graph based on a neighbor sampling algorithm; Taking the newly added material as a target node, processing the relationship graph using a neighbor aggregation algorithm to generate a feature vector for the newly added material; determining, based on the similarity between the feature vector of the newly added material and each of the second feature vectors, a second preset number of candidate material labels associated with the newly added material from the plurality of candidate material labels associated with each of the plurality of materials; Determining second category information corresponding to each of the second preset number of candidate material tags; as well as Based on the second category information, the second preset number of candidate material labels are processed to obtain a second material label set that matches the newly added material.

8. The method according to any one of claims 1 to 5, wherein The acquiring of a plurality of candidate material tags associated with each of the plurality of materials includes: For each material, determining at least one first candidate material tag associated with the material from a plurality of to-be-processed material tags contained in a material library; and Each first candidate material tag among the at least one first candidate material tag is determined as the candidate material tag.

9. The method according to claim 8, wherein The acquiring of a plurality of candidate material tags associated with each of the plurality of materials further includes: For each material, obtain description information associated with the material; Performing word segmentation processing on the description information to obtain multiple keywords; Match multiple keywords with the preset tag library respectively; In response to matching the keyword in the tag library, determining the keyword as a second candidate material tag associated with the material; and The at least one first candidate material tag and each of the second candidate material tags are determined as the plurality of candidate material tags.

10. A material label processing device, comprising: An acquisition module, configured to acquire a plurality of candidate material tags associated with each of the plurality of materials; A first generating module is configured to generate a heterogeneous graph based on the multiple materials and multiple candidate material labels associated with each of the multiple materials, wherein the heterogeneous graph is used to represent the association relationship between the candidate material labels and the corresponding materials; a first determining module, configured to determine, based on the heterogeneous graph, a first preset number of candidate material labels associated with each material from a plurality of candidate material labels associated with each of the plurality of materials; A second determining module is configured to determine, for each material, first category information corresponding to each of the first preset number of candidate material labels; as well as a first processing module, configured to process the first preset number of candidate material labels based on the first category information to obtain a first material label set matching the material; Wherein, the first processing module includes: an identification unit, configured to, in response to determining that the first category information belongs to a second category attribute, identify a target object in the material, and obtain an identification result and an identification confidence level for the target object; a first matching unit configured to, in response to determining that the recognition confidence is less than a preset confidence, match the recognition result for the target object with the first preset number of candidate material labels respectively; and a second determining unit configured to, in response to determining that the recognition result matches at least one candidate material label among the first preset number of candidate material labels, obtain the first material label set according to the recognition result and the at least one candidate material label; Wherein, the material is video material.

11. The device according to claim 10, wherein The first processing module includes: a first determining unit configured to, in response to determining that the first category information belongs to a first category attribute, respectively determine a matching degree between the material and the first preset number of candidate material labels; and A sorting unit is configured to sort the first preset number of candidate material labels according to the matching degree, and obtain the first material label set based on the sorting result.

12. The device according to claim 10, wherein The first processing module further includes: A third determining unit is configured to obtain the first material label set according to the recognition result for the target object and the first preset number of candidate material labels in response to determining that the recognition confidence is greater than or equal to the preset confidence.

13. The device according to claim 12, wherein The first material tag set includes a plurality of first material tags; the first processing module further includes: A first acquisition unit, configured to acquire first knowledge data; wherein the first knowledge data is acquired based on description information associated with the material; a second acquiring unit, configured to acquire, based on the first knowledge data, a first additional material tag corresponding to each of the plurality of first material tags; wherein the first knowledge data is associated with the first material tag; a third acquiring unit, configured to acquire, based on the first knowledge data, second additional material tags corresponding to each of the first preset number of candidate material tags; and An adding unit is configured to add each of the first additional material tags and each of the second additional material tags to the first material tag set.

14. The device according to claim 10, wherein The first material tag set includes a plurality of first material tags; the first processing module further includes: A first acquisition unit, configured to acquire first knowledge data; wherein the first knowledge data is acquired based on description information associated with the material; a second acquiring unit, configured to acquire, based on the first knowledge data, a first additional material tag corresponding to each of the plurality of first material tags; wherein the first knowledge data is associated with the first material tag; a third acquiring unit, configured to acquire, based on the first knowledge data, second additional material tags corresponding to each of the first preset number of candidate material tags; and An adding unit is configured to add each of the first additional material tags and each of the second additional material tags to the first material tag set.

15. The device according to any one of claims 10 to 14, wherein The first determining module includes: an extraction unit, configured to perform feature extraction on the node data of each node in the heterogeneous graph to obtain a first feature vector corresponding to each material node and a second feature vector corresponding to each candidate material label node; and The fourth determining unit is configured to determine, for each material, a first preset number of candidate material labels associated with the material from the plurality of candidate material labels associated with each of the plurality of materials based on similarities between the first feature vector and each of the second feature vectors.

16. The apparatus according to claim 15, further comprising: a matching module, configured to, in response to acquiring a new material and a new material tag corresponding to the new material, match the new material tag with a preset tag library and the heterogeneous graph respectively; a second generating module configured to generate, in response to matching the new material label in the label library and matching the new material label in the heterogeneous graph, a relationship graph including the new material based on the new material, the new material label, and the heterogeneous graph based on a neighbor sampling algorithm; an aggregation module, configured to process the relationship graph using a neighbor aggregation algorithm with the newly added material as a target node, and generate a feature vector for the newly added material; a third determining module, configured to determine, based on similarities between the feature vector of the new material and each of the second feature vectors, a second preset number of candidate material labels associated with the new material from the plurality of candidate material labels associated with each of the plurality of materials; A fourth determining module, configured to determine second category information corresponding to each of the second preset number of candidate material tags; as well as The second processing module is configured to process the second preset number of candidate material labels based on the second category information to obtain a second material label set that matches the newly added material.

17. The device according to any one of claims 10 to 14, wherein The acquisition module includes: a fifth determining unit, configured to determine, for each material, at least one first candidate material tag associated with the material from a plurality of to-be-processed material tags contained in the material library; and The sixth determining unit is configured to determine each first candidate material tag in the at least one first candidate material tag as the candidate material tag.

18. The device according to claim 17, wherein The acquisition module also includes: a fourth acquiring unit, configured to acquire, for each material, description information associated with the material; A word segmentation unit, configured to perform word segmentation processing on the description information to obtain a plurality of keywords; A second matching unit is used to match the multiple keywords with the preset tag library respectively; a seventh determining unit, configured to, in response to matching the keyword in the tag library, determine the keyword as a second candidate material tag associated with the material; and An eighth determining unit is configured to determine the at least one first candidate material label and each of the second candidate material labels as the plurality of candidate material labels.

19. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 9.

20. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 9.

21. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 9.

Images