Tag determination method, information recommendation method, device, equipment and storage medium

Abstract

The embodiment of the present disclosure discloses a label determination method, an information recommendation method, a device, equipment and a storage medium, and relates to the technical field of computers. The label determination method comprises the following steps: determining the similarity of current information to be stored in a warehouse and each piece of information in a preset information warehouse; for each preset similarity threshold in a preset similarity threshold set, determining a target set of information in the warehouse corresponding to the current preset similarity threshold from the preset information warehouse, and determining a sub-label corresponding to the current preset similarity threshold based on the target multi-granularity label of the information in the target set of information in the warehouse, wherein the similarity of the information in the target set of information in the warehouse is greater than the current preset similarity threshold; and determining the multi-granularity label of the current information according to the sub-label corresponding to each preset similarity threshold in the preset similarity threshold set. By using the above technical solution, the multi-granularity label can be set for the information, the representation of the information is more accurate, and the automation degree and efficiency are high.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to tag determination methods, information recommendation methods, apparatus, devices, and storage media. Background Technology

[0002] With the rapid development of information technology and the popularization of the Internet, users have access to more and more information, and the emergence of recommendation systems can effectively improve users' information acquisition efficiency.

[0003] Currently, recommendation systems can filter suitable information from a database to recommend to users. The information in the database is associated with information tags, which are usually set manually. Summary of the Invention

[0004] This disclosure provides a label determination method, apparatus, storage medium, and device, which can optimize existing label determination schemes.

[0005] In a first aspect, embodiments of this disclosure provide a label determination method, including:

[0006] Determine the similarity between the current information to be added to the database and the information in each of the preset databases;

[0007] For each preset similarity threshold in the preset similarity threshold set, a target database information set corresponding to the current preset similarity threshold is determined from the preset information database, and a sub-tag corresponding to the current preset similarity threshold is determined based on the target multi-granularity tags of the database information in the target database information set. The number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the database information in the target database information set is greater than the current preset similarity threshold.

[0008] The multi-granularity label of the current information is determined based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set.

[0009] Secondly, this disclosure also provides an information recommendation method, including:

[0010] Input data is determined based on the multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information database, and the multi-granularity labels of the information in the preset information database are determined based on the label determination method described in any of the embodiments of this disclosure;

[0011] The input data is fed into a preset information recommendation model;

[0012] Based on the output of the preset information recommendation model, the target information to be recommended is determined from the candidate information.

[0013] Thirdly, embodiments of this disclosure also provide a tag determining device, including:

[0014] The similarity determination module is used to determine the similarity between the current information to be added to the database and the information in each database in the preset information database;

[0015] The sub-label determination module is used to determine, for each preset similarity threshold in the preset similarity threshold set, the target database information set corresponding to the current preset similarity threshold from the preset information database, and determine the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity labels of the database information in the target database information set, wherein the number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the database information in the target database information set is greater than the current preset similarity threshold;

[0016] The multi-granularity label determination module is used to determine the multi-granularity label of the current information based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set.

[0017] Fourthly, embodiments of this disclosure also provide an information recommendation device, including:

[0018] An input data determination module is used to determine input data based on multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information library, and the multi-granularity labels of the information in the preset information library are determined based on any of the label determination methods described in the embodiments of this disclosure;

[0019] The data input module is used to input the input data into a preset information recommendation model;

[0020] The information recommendation module is used to determine the target information to be recommended from the candidate information based on the output of the preset information recommendation model.

[0021] Fifthly, embodiments of this disclosure also provide an electronic device, the electronic device comprising:

[0022] One or more processors;

[0023] Storage device for storing one or more programs.

[0024] When the one or more programs are executed by the one or more processors, the one or more processors implement the methods provided in the embodiments of this disclosure.

[0025] In a sixth aspect, embodiments of this disclosure also provide a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform the methods provided in embodiments of this disclosure.

[0026] The tag determination scheme provided in this embodiment determines the similarity between the current information to be added to the database and the information in each database in a preset information database; for each preset similarity threshold in the preset similarity threshold set, it determines the target database information set corresponding to the current preset similarity threshold from the preset information database, and determines the sub-tags corresponding to the current preset similarity threshold based on the target multi-granularity tags of the database information in the target database information set, wherein the similarity of the database information in the target database information set is greater than the current preset similarity threshold; and determines the multi-granularity tags of the current information according to the sub-tags corresponding to each preset similarity threshold in the preset similarity threshold set. By adopting the above technical solution, multi-granularity tags can be set for information, resulting in a more refined representation of the information. For each granularity, a corresponding similarity threshold is pre-set. Based on the similarity threshold, sub-tags for determining the current information are selected from the existing multi-granularity tags in the database. Then, corresponding multi-granularity tags are generated based on the sub-tags. This solution is highly automated and efficient, reducing the bias caused by manual tag determination. New information can be tagged online in real time, and it can be quickly expanded to a large-scale tag system, which is conducive to improving the recommendation effect of information recommendation schemes based on this multi-granularity tag. Attached Figure Description

[0027] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote 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.

[0028] Figure 1 This is a schematic flowchart of a label determination method provided in an embodiment of the present disclosure;

[0029] Figure 2 This is a schematic flowchart of a label determination method provided in an embodiment of the present disclosure;

[0030] Figure 3 This is a flowchart illustrating an information recommendation method provided in an embodiment of the present disclosure;

[0031] Figure 4 This is a schematic diagram of the structure of a label determining device provided in an embodiment of the present disclosure;

[0032] Figure 5 This is a schematic diagram of the structure of an information recommendation device provided in an embodiment of the present disclosure;

[0033] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0034] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0035] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0036] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based 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". Definitions of other terms will be given in the description below.

[0037] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0038] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0039] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0040] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0041] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0042] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0043] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0044] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0045] To facilitate understanding of the technical solutions of the embodiments of this disclosure, related technologies are described below. Information features can characterize the properties of information and play an important role in systems such as recommendation systems. In related technologies, the application of information features in the system mainly takes two forms: one is to directly input them into the model in the form of dense features; the other is to input them into the model in the form of information labels. Information labels are usually based on manual settings, such as manual annotation, or manual labeling of sample data. After obtaining the training set, a label model is trained based on the training set, and then the label model is used to predict the labels of the information. However, both of the above methods have certain drawbacks. For the first method, the accuracy of the model is often lower than that of the second method. For the second method, manual annotation is costly and the granularity of the label features is coarse, making it difficult to extend to a large-scale label system, and the accuracy is also limited.

[0046] In this embodiment of the disclosure, multi-granularity labels can be set for information, which makes the representation of information more refined. The label determination process is highly automated and efficient, and it is also beneficial to improve the accuracy of the model.

[0047] Figure 1 This is a flowchart illustrating a label determination method provided in an embodiment of the present disclosure. This embodiment is applicable to situations where labels are determined for information. The method can be executed by a label determination device, which can be implemented in the form of software and / or hardware. Optionally, it can be implemented by an electronic device, such as a mobile terminal like a mobile phone, smartwatch, tablet computer, or personal digital assistant, or a device like a personal computer (PC) or server.

[0048] like Figure 1 As shown, the method includes:

[0049] Step 101: Determine the similarity between the current information to be added to the database and the information in each database in the preset information database.

[0050] In this embodiment, the type of information is not limited; for example, it can be online information (such as advertisements), videos, news, test questions, or products, etc. The preset information database can be understood as a database used to store information with added multi-granularity tags. Information currently existing in the preset information database can be recorded as information within the database. The preset information database can also be used to store associated data of information within the database, which may include related data such as feature data or information identifiers of the information within the database. The current information to be added to the database can be understood as information that is about to be added to the preset information database without added multi-granularity tags.

[0051] For example, before adding the current information to a preset information database, multi-granularity tags corresponding to the current information can be determined first. When determining tags, the similarity between the current information and the information in each database in the preset information database can be determined first. The specific method of determining the similarity is not limited. For example, the similarity between the feature data corresponding to the current information and the feature data corresponding to the information in the database can be calculated, and this similarity can be used as the similarity. The process of determining the similarity can be completed by a preset model, such as the recall model in a recommendation system.

[0052] Optionally, the aforementioned feature data can specifically be dense features. Dense features are features where most values ​​are non-zero, typically represented by floating-point tensors, the opposite of sparse features. Information tags in related technologies can be understood as a type of sparse feature. This step may include: obtaining the current dense feature corresponding to the current information to be added to the database; for each piece of information in the preset database, calculating the similarity between the current dense feature and the corresponding dense feature of the current information in the database, thus obtaining the similarity between the current information and each piece of information in the database. The advantage of this setup is that dense features have strong flexibility and can more comprehensively represent the features of information. Labeling dense features, i.e., generalizing dense features, can more accurately obtain the similarity between the current information and the information in the database.

[0053] For example, the dense features corresponding to information can specifically include the content features of the information. Taking video as an example, the dense features corresponding to video can include video content features, and may also include other features, such as video type features. Dense features can be obtained based on feature extraction using deep neural networks. For example, obtaining the current dense features corresponding to the current information to be added to the database can specifically include: extracting the current dense features corresponding to the current information to be added to the database based on a preset deep neural network. The dense features corresponding to information within the database can be denoted as database-internal dense features. Optionally, for information within the database, the corresponding database-internal dense features can be extracted in advance using a preset deep neural network, and the extracted database-internal dense features can be stored. This extraction process can be completed when the database-internal information determines multi-granularity labels before being added to the database. In this way, the computational efficiency can be improved when calculating the similarity between the current dense features and the current database-internal dense features corresponding to the current database information. When calculating the above similarity, each piece of information within the preset information database can become the current database-internal information in a preset order, and the corresponding similarity calculation can be performed until all database-internal information has been traversed.

[0054] Step 102: For each preset similarity threshold in the preset similarity threshold set, determine the target database information set corresponding to the current preset similarity threshold from the preset information database, and determine the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity label of the database information in the target database information set. The number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the database information in the target database information set is greater than the current preset similarity threshold.

[0055] In this embodiment of the disclosure, multi-granularity tags are added to the information. The multi-granularity includes at least two granularities. A corresponding preset similarity threshold can be pre-set for each granularity. All preset similarity thresholds corresponding to all granularities constitute a preset similarity threshold set. Each preset similarity threshold is different, and the specific value is not limited. For example, if there are three granularities and the corresponding preset similarity thresholds are K1, K2, and K3, then the preset similarity threshold set can be denoted as [K1, K2, K3], where K1 ≠ K2 ≠ K3.

[0056] For example, each preset similarity threshold in the preset similarity threshold set can become the current preset similarity threshold in a preset order, and corresponding sub-labels can be determined, until all preset similarity thresholds have been traversed. As in the example above, assuming the current preset similarity threshold is K1, the database information with a similarity greater than K1 is added to the target database information set corresponding to K1, and based on the target multi-granularity labels of the database information in the target database information set, the sub-label id1 corresponding to K1 is determined, and so on, the sub-label id2 corresponding to K2 and the sub-label id3 corresponding to K3 are then determined.

[0057] For example, in the target database information set corresponding to the current preset similarity threshold, each target database information has a corresponding multi-granularity label. The multi-granularity labels corresponding to different target database information may be the same or different. The target multi-granularity labels can be determined from them according to a preset determination method. Generally, the number of target multi-granularity labels corresponding to the current preset similarity threshold is 1. Optionally, the target multi-granularity label can be determined as a sub-label corresponding to the current preset similarity threshold.

[0058] For example, the specific determination method is not limited. For instance, the most frequently occurring multi-granularity tag in the information set of the target database can be selected as the target multi-granularity tag, or the similarity corresponding to each multi-granularity tag can be combined to determine it.

[0059] Step 103: Determine the multi-granularity label of the current information based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set.

[0060] For example, sub-tags can be sorted according to a preset sorting rule to obtain multi-granularity tags for the current information. The preset sorting rule can, for example, be related to the magnitude of a preset similarity threshold corresponding to the sub-tag. For instance, sub-tags can be sorted in descending order of their corresponding preset similarity thresholds. Assuming there are n preset similarity thresholds, sorting the sub-tags in descending order of their corresponding preset similarity thresholds will yield id1, id2, ..., id n The multi-granularity tag ID of the current information can be represented as ID = (id1, id2, ..., id... n As illustrated above, assuming K1>K2>K3, the multi-granularity label ID of the current information can be represented as ID=(id1, id2, id3).

[0061] The tag determination method provided in this embodiment determines the similarity between current information to be added to the database and information in each database in a preset information database; for each preset similarity threshold in a preset similarity threshold set, a target database information set corresponding to the current preset similarity threshold is determined from the preset information database, and a sub-tag corresponding to the current preset similarity threshold is determined based on the target multi-granularity tags of the database information in the target database information set, wherein the similarity of the database information in the target database information set is greater than the current preset similarity threshold; and the multi-granularity tags of the current information are determined according to the sub-tags corresponding to each preset similarity threshold in the preset similarity threshold set. By adopting the above technical solution, multi-granularity tags can be set for information, resulting in a more refined representation of the information. A corresponding similarity threshold is set for each granularity. Based on the similarity threshold, sub-tags for determining the current information are selected from the existing multi-granularity tags in the database. Then, corresponding multi-granularity tags are generated based on the sub-tags. This solution is highly automated and efficient. New information can be labeled online in real time, which can reduce the bias caused by manual label determination. It can also be quickly expanded to a large-scale tag system, which is conducive to improving the recommendation effect of the information recommendation solution based on this multi-granularity tag.

[0062] In some embodiments, before determining sub-tags for each preset similarity threshold, the information in the database can be pre-filtered. For example, before determining the target set of information in the database corresponding to each preset similarity threshold in the preset similarity threshold set, the method further includes: filtering a preset number of database entries with the highest similarity from the preset information database based on the similarity, to obtain an initial set of database entries. Specifically, determining the target set of information in the database corresponding to each preset similarity threshold in the preset similarity threshold set includes: determining the target set of information in the database corresponding to each preset similarity threshold in the preset similarity threshold set from the initial set of database entries. The advantage of this approach is that it avoids an excessive number of database entries exceeding the preset similarity threshold, which could increase the risk of system crash. Furthermore, when using a recall model to determine similarity, the output of the recall model is usually a certain number of results with high similarity; therefore, it can better match the recall model and reduce the cost of determining similarity.

[0063] In some embodiments, the target database information set corresponding to the smallest preset similarity threshold (denoted as set A) may be determined first, and the target database information sets corresponding to other preset similarity thresholds in the preset similarity threshold set may be subsets of set A.

[0064] In some embodiments, determining the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity labels of the information within the target database includes: for each multi-granularity label corresponding to the information within the database in the target database, determining the similarity to the information within the database to which the current multi-granularity label belongs, and accumulating the determined similarities to obtain the label score of the current multi-granularity label; determining the multi-granularity label with the highest label score as the target multi-granularity label, and determining the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity label. The advantage of this setup is that it allows for cumulative voting based on similarity, enabling the determined target multi-granularity labels to more accurately represent the current information, more reasonably and accurately determine the target multi-granularity labels, and thus more reasonably and accurately determine the sub-labels.

[0065] In some embodiments, the method further includes: when the information set in the target database is empty, determining the sub-tag corresponding to the current preset similarity threshold based on the information identifier of the current information. The advantage of this setting is that it ensures the integrity of multi-granularity tags.

[0066] Optionally, the information identifier of the current information can be directly determined as the sub-tag corresponding to the current preset similarity threshold. The information identifiers for different pieces of information can be the same or different.

[0067] Optionally, the information identifier is a unique identifier for the current information. This approach eliminates the need for pre-classification of information, increases the automation of label determination, and facilitates expansion to large-scale labeling systems. The unique identifier can be a string, such as a numeric code or a combination of numbers and letters; its specific form is not limited.

[0068] In some embodiments, after determining the multi-granularity tags of the current information, the method may further include: adding the determined multi-granularity tags to the current information and adding the current information to the preset information database.

[0069] Figure 2 This is a flowchart illustrating a label determination method provided in an embodiment of the present disclosure. The embodiments of the present disclosure are optimized based on the various optional solutions in the above embodiments. Specifically, the method includes the following steps:

[0070] Step 201: Obtain the current dense features corresponding to the current information to be added to the database.

[0071] For example, the current dense features corresponding to the current information to be added to the database can be extracted based on a preset deep neural network.

[0072] Step 202: For each piece of information in the preset information database, calculate the similarity between the current dense feature and the corresponding dense feature in the current database, and obtain the similarity between the current information and each piece of information in the database.

[0073] For example, a recall model can be used to calculate the similarity between the current dense feature and the dense features in each of the preset information databases, so as to obtain the similarity between the current information and the information in each database.

[0074] Step 203: Based on similarity, select a preset number of information items with the highest similarity from the preset information database to obtain an initial set of information items in the database.

[0075] For example, a recall model can be used to output X pieces of information in the database that are most similar to the current information, where X is a preset number, and these X pieces of information in the database constitute the initial set of information in the database.

[0076] Step 204: For the current preset similarity threshold in the preset similarity threshold set, determine the target database information set corresponding to the current preset similarity threshold from the initial database information set.

[0077] For example, the preset similarity thresholds in the preset similarity threshold set are 0.95, 0.60, 0.85, 0.80, 0.75, and 0.70, respectively. The current preset similarity threshold can be determined sequentially according to the preset similarity thresholds in ascending or descending order. Assuming the current preset similarity threshold is 0.95, in the initial set of information within the database, that is, among the aforementioned X pieces of information within the database, there are 8 pieces of information within the database with a similarity greater than 0.95, namely 0.99, 0.985, 0.98, 0.97, 0.97, 0.965, 0.962, and 0.961. Then, the target set of information within the database corresponding to 0.95 includes these 8 pieces of information within the database.

[0078] If there is no information in the initial database with a similarity greater than 0.95, then the target database with a similarity of 0.95 is empty.

[0079] Step 205: For each multi-granularity tag corresponding to the information in the target database, determine the similarity to the information in the database to which the current multi-granularity tag belongs, and accumulate the determined similarities to obtain the tag score of the current multi-granularity tag.

[0080] For example, assuming the multi-granularity tags corresponding to the 8 pieces of information in the database are ID1, ID2, ID3, ID1, ID2, ID3, ID4, and ID3, then there are a total of 4 types of multi-granularity tags. The tag score corresponding to ID1 is 0.99 + 0.97 = 1.96, the tag score corresponding to ID2 is 0.985 + 0.97 = 1.955, the tag score corresponding to ID3 is 0.98 + 0.965 + 0.961 = 2.606, and the tag score corresponding to ID4 is 0.962.

[0081] Step 206: Determine the multi-granularity label with the highest label score as the target multi-granularity label, and determine the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity label.

[0082] As in the example above, if the tag score corresponding to ID3 is the highest, then the sub-tag id1 corresponding to the current preset similarity threshold of 0.95 is ID3.

[0083] If the target database information set corresponding to 0.95 is empty, and assuming the current information identifier is denoted as id, then the sub-label id1 corresponding to the current preset similarity threshold of 0.95 is id1 = id.

[0084] Step 207: Determine whether there are any untraversed preset similarity thresholds in the preset similarity threshold set. If yes, proceed to step 208; otherwise, proceed to step 209.

[0085] For example, if there are still preset similarity thresholds in the preset similarity threshold set for which no corresponding sub-tag has been determined, then step 208 can be executed to determine the next preset similarity threshold as the current preset similarity threshold, such as determining 0.60 as the current preset similarity threshold, and then returning to execute step 204 to determine the sub-tag corresponding to 0.60. If the last preset similarity threshold of 0.70 has a corresponding sub-tag already determined, then step 209 can be executed to determine the multi-granularity tag of the current information based on each sub-tag.

[0086] Step 208: Determine the next preset similarity threshold in the preset similarity threshold set as the current preset similarity threshold, and return to step 204.

[0087] Step 209: Determine the multi-granularity label of the current information based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set.

[0088] For example, the multi-granularity tag ID of the current information can be recorded as (id1, id2, ..., id6).

[0089] Step 210: Add the determined multi-granularity tags to the current information and add the current information to the preset information database.

[0090] For example, the current information is added to a preset information database, becoming new information in the database.

[0091] Optionally, to prevent the preset information database from overflowing, preset deletion conditions can be set. These conditions can be configured according to actual needs, such as timed deletion. For example, each piece of information added to the database can be automatically deleted from the preset information database after one month.

[0092] The tag determination method provided in this embodiment can cluster dense features of each piece of information to be added to the database into generalized tags according to multiple pre-set similarity thresholds. Due to the diversity of similarity thresholds, the final clustering result is also multi-granular. When determining the sub-tags corresponding to each similarity threshold, a cumulative voting method based on similarity is adopted, so that the determined target multi-granular tags can more accurately represent the current information and describe the information features more finely. This clustering method has low complexity, high automation and high efficiency. New information can be added online in real time to determine tags, and can be quickly expanded to a large-scale tag system. Introducing the determined multi-granular tags into the recommendation model can fill the modeling gap of multi-granular tags and help improve the recommendation effect of the information recommendation scheme based on the multi-granular tags.

[0093] For example, to facilitate understanding of the multi-granularity labels in this disclosure embodiment, a simple explanation is given below using a preset similarity threshold of 2, assuming they are 0.9 and 0.8 respectively. Assuming the initial preset information database contains 0 items, and the unique identifier of the current information a to be added is 1, then the multi-granularity label of information a can be represented as ID(a) = (1, 1). The unique identifier of the next information b to be added is 2. Assuming the similarity between information a and information b is less than 0.8, then the multi-granularity label of information b can be represented as ID(b) = (2, 2). The unique identifier of the next information c to be added is 3. Assuming the similarity between information c and information a is greater than 0.8 and less than 0.9, and the similarity between information c and information b is less than 0.8, then the multi-granularity label of information c can be represented as ID(c) = (3, (1, 1)). The unique identifier of the next piece of information d to be added to the database is 4. Assuming that the similarity between information d and information a, information b and information c are 0.81, 0.85 and 0.89 respectively, all of which are greater than 0.8 and less than 0.9, information c has the highest similarity. Therefore, the sub-label corresponding to the threshold of 0.9 is the unique identifier 4 of d, and the sub-label corresponding to the threshold of 0.8 is the multi-granularity label of information c. Then the multi-granularity label of information d can be represented as ID(d) = (4, (3, (1, 1))). By analogy, the multi-granularity labels of a large number of pieces of information can be automatically determined, and a large-scale multi-granularity label system can be formed.

[0094] Figure 3This is a flowchart illustrating an information recommendation method provided in an embodiment of the present disclosure. The embodiments of the present disclosure are applicable to information recommendation scenarios. The method can be executed by an information recommendation device, which can be implemented in the form of software and / or hardware. Optionally, it can be implemented by an electronic device, such as a mobile terminal like a mobile phone, smartwatch, tablet computer, or personal digital assistant, or a device like a personal computer or server.

[0095] Step 301: Determine the input data based on the multi-granularity labels of the candidate information.

[0096] The candidate information includes information in a preset information database, and the multi-granularity labels of the information in the preset information database are determined based on the label determination method described in any of the embodiments of this disclosure.

[0097] For example, a pre-defined information recommendation model can be trained using training samples carrying multi-granularity labels to obtain the model, which is then used for information recommendation. When recommending information, the input data to the pre-defined model may include multi-granularity labels of candidate information, as well as other relevant data such as dense features of the candidate information; the specific details are not limited. There can be multiple candidate information items. The input data for each candidate information item can be determined individually, or the input data for at least two candidate information items can be determined in batches, depending on the input requirements of the pre-defined information recommendation model.

[0098] Step 302: Input the input data into the preset information recommendation model.

[0099] Step 303: Based on the output of the preset information recommendation model, determine the target information to be recommended from the candidate information.

[0100] For example, based on the output of a preset information recommendation model, one or more target information to be recommended are determined from multiple candidate information.

[0101] The information recommendation method provided in this disclosure uses information in a preset information database as candidate information and determines input data based on the multi-granularity labels of the candidate information. The multi-granularity labels provide a more detailed characterization of information features, thereby making the input data more accurate. After inputting the input data into the preset information recommendation model, the appropriate target information to be recommended can be accurately determined based on the output of the preset information recommendation model, thereby improving the recommendation effect.

[0102] Figure 4 This is a schematic diagram of the structure of a tag determining device provided in an embodiment of this disclosure, as shown below. Figure 4 As shown, the device includes:

[0103] The similarity determination module 401 is used to determine the similarity between the current information to be added to the database and the information in each database in the preset information database;

[0104] The sub-label determination module 402 is used to determine, for each preset similarity threshold in the preset similarity threshold set, the target library information set corresponding to the current preset similarity threshold from the preset information database, and determine the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity labels of the library information in the target library information set, wherein the number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the library information in the target library information set is greater than the current preset similarity threshold;

[0105] The multi-granularity label determination module 403 is used to determine the multi-granularity label of the current information based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set.

[0106] The tag determination device provided in this embodiment can set multi-granularity tags for information, resulting in a more refined representation of the information. For each granularity, a corresponding similarity threshold is pre-set. Based on the similarity threshold, sub-tags for determining the current information are selected from the existing multi-granularity tags in the database. Then, corresponding multi-granularity tags are generated based on the sub-tags. This scheme has a high degree of automation and efficiency, which can reduce the deviation caused by manual tag determination. New information can be tagged online in real time, and it can be quickly expanded to a large-scale tag system, which is conducive to improving the recommendation effect of the information recommendation scheme based on the multi-granularity tags.

[0107] Optionally, the device may also include:

[0108] The initial database information set determination module is used to, before determining the target database information set corresponding to the current preset similarity threshold from the preset information database for each preset similarity threshold in the preset similarity threshold set, filter a preset number of database information with the highest similarity from the preset information database based on the similarity to obtain the initial database information set;

[0109] Specifically, determining the target database information set corresponding to the current preset similarity threshold from the preset information database for each preset similarity threshold in the preset similarity threshold set includes: determining the target database information set corresponding to the current preset similarity threshold from the initial database information set for each preset similarity threshold in the preset similarity threshold set.

[0110] Optionally, determining the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity label of the information in the target database set includes: for each multi-granularity label corresponding to the information in the database in the target database set, determining the similarity to the information in the database to which the current multi-granularity label belongs, and accumulating the determined similarities to obtain the label score of the current multi-granularity label; determining the multi-granularity label with the highest label score as the target multi-granularity label, and determining the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity label.

[0111] Optionally, the device further includes a tag determination module, used to determine the sub-tag corresponding to the current preset similarity threshold based on the information identifier of the current information when the information set in the target database is empty.

[0112] Optionally, the information identifier is a unique identifier for the current information.

[0113] Optional, the similarity determination module includes:

[0114] The dense feature acquisition unit is used to acquire the current dense features corresponding to the current information to be entered into the database.

[0115] The similarity determination unit is used to calculate the similarity between the current dense feature and the corresponding current dense feature in the current database for each piece of information in the preset information database, so as to obtain the similarity between the current information and each piece of information in the database.

[0116] The label determination device provided in this disclosure can execute the label determination method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of the method execution.

[0117] Figure 5 This is a schematic diagram of the structure of an information recommendation device provided in an embodiment of this disclosure, as shown below. Figure 5 As shown, the device includes:

[0118] The input data determination module 501 is used to determine input data based on the multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information library, and the multi-granularity labels of the information in the preset information library are determined based on any of the label determination methods described in the embodiments of this disclosure;

[0119] Data input module 502 is used to input the input data into a preset information recommendation model;

[0120] The information recommendation module 503 is used to determine the target information to be recommended from the candidate information based on the output of the preset information recommendation model.

[0121] The information recommendation device provided in this embodiment uses information in a preset information database as candidate information and determines input data based on the multi-granularity labels of the candidate information. The multi-granularity labels describe the information features more finely, thereby making the input data more accurate. After the input data is input into the preset information recommendation model, the appropriate target information to be recommended can be accurately determined based on the output of the preset information recommendation model, thereby improving the recommendation effect.

[0122] The information recommendation device provided in this disclosure can execute the information recommendation method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of executing the method.

[0123] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.

[0124] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Reference is made below. Figure 6 It illustrates an electronic device suitable for implementing embodiments of the present disclosure (e.g., Figure 6 The diagram below shows the structure of the terminal device or server 600. The terminal device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and vehicle terminals (e.g., vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0125] like Figure 6 As shown, electronic device 600 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 602 or a program loaded from storage device 608 into random access memory (RAM) 603. The RAM 603 also stores various programs and data required for the operation of electronic device 600. The processing unit 601, ROM 602, and RAM 603 are interconnected via bus 604. An edit / output (I / O) interface 605 is also connected to bus 604.

[0126] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic device 600 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 An electronic device 600 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively.

[0127] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a storage device 608, or installed from a ROM 602. When the computer program is executed by the processing device 601, it performs the functions defined in the methods of embodiments of this disclosure.

[0128] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0129] The electronic device provided in this embodiment belongs to the same inventive concept as the tag determination method or information recommendation method provided in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0130] This disclosure provides a computer storage medium storing a computer program that, when executed by a processor, implements the tag determination method and / or information recommendation method provided in the above embodiments.

[0131] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0132] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0133] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0134] The aforementioned computer-readable medium carries one or more programs. When the aforementioned one or more programs are executed by the electronic device, the electronic device causes the following: to determine the similarity between current information to be added to a database and information within each database in a preset information database; for each preset similarity threshold in a preset similarity threshold set, to determine a target database information set corresponding to the current preset similarity threshold from the preset information database, and to determine a sub-tag corresponding to the current preset similarity threshold based on the target multi-granularity tags of the information within the target database information set, wherein the number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the information within the target database information set is greater than the current preset similarity threshold; and to determine a multi-granularity tag for the current information based on the sub-tags corresponding to each preset similarity threshold in the preset similarity threshold set.

[0135] Alternatively, the aforementioned computer-readable medium carries one or more programs that, when executed by the electronic device, cause the electronic device to: determine input data based on multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information database, and the multi-granularity labels of the information in the preset information database are determined based on any of the label determination methods described in the embodiments of this disclosure; input the input data into a preset information recommendation model; and determine target information to be recommended from the candidate information based on the output of the preset information recommendation model.

[0136] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including but not limited to object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone 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 remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0137] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0138] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The names of modules do not necessarily limit the module itself; for example, a similarity determination module can also be described as "a module for determining the similarity between current information to be added to the database and information in each database of a preset database".

[0139] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0140] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0141] According to one or more embodiments of this disclosure, a label determination method is provided, comprising:

[0142] Determine the similarity between the current information to be added to the database and the information in each of the preset databases;

[0143] For each preset similarity threshold in the preset similarity threshold set, a target database information set corresponding to the current preset similarity threshold is determined from the preset information database, and a sub-tag corresponding to the current preset similarity threshold is determined based on the target multi-granularity tags of the database information in the target database information set. The number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the database information in the target database information set is greater than the current preset similarity threshold.

[0144] The multi-granularity label of the current information is determined based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set.

[0145] According to one or more embodiments of this disclosure, before determining the target database information set corresponding to the current preset similarity threshold from the preset information database for each preset similarity threshold in the preset similarity threshold set, the method further includes:

[0146] Based on the similarity, a preset number of information with the highest similarity are selected from the preset information database to obtain an initial set of information in the database.

[0147] Wherein, determining the target database information set corresponding to the current preset similarity threshold from the preset information database for each preset similarity threshold in the preset similarity threshold set includes:

[0148] For each preset similarity threshold in the preset similarity threshold set, the target database information set corresponding to the current preset similarity threshold is determined from the initial database information set.

[0149] According to one or more embodiments of this disclosure, determining the sub-tag corresponding to the current preset similarity threshold based on the target multi-granularity tags of the target information set in the target database includes:

[0150] For each multi-granularity tag corresponding to the information in the target database, determine the similarity to the information in the database to which the current multi-granularity tag belongs, and accumulate the determined similarities to obtain the tag score of the current multi-granularity tag;

[0151] The multi-granularity label with the highest label score is determined as the target multi-granularity label, and the sub-label corresponding to the current preset similarity threshold is determined based on the target multi-granularity label.

[0152] According to one or more embodiments of this disclosure, it further includes:

[0153] If the information set in the target database is empty, the sub-tag corresponding to the current preset similarity threshold is determined based on the information identifier of the current information.

[0154] According to one or more embodiments of this disclosure, the information identifier is a unique identifier for the current information.

[0155] According to one or more embodiments of this disclosure, determining the similarity between the current information to be added to the database and the information in each database of the preset information database includes:

[0156] Obtain the current dense features corresponding to the current information to be added to the database;

[0157] For each piece of information in the preset information database, calculate the similarity between the current dense feature and the corresponding dense feature in the current database, and obtain the similarity between the current information and each piece of information in the database.

[0158] According to one or more embodiments of this disclosure, an information recommendation method is provided, comprising:

[0159] Input data is determined based on the multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information database, and the multi-granularity labels of the information in the preset information database are determined based on the label determination method described in any of the embodiments of this disclosure;

[0160] The input data is fed into a preset information recommendation model;

[0161] Based on the output of the preset information recommendation model, the target information to be recommended is determined from the candidate information.

[0162] According to one or more embodiments of the present disclosure, a tag determining apparatus is provided, comprising:

[0163] The similarity determination module is used to determine the similarity between the current information to be added to the database and the information in each database in the preset information database;

[0164] The sub-label determination module is used to determine, for each preset similarity threshold in the preset similarity threshold set, the target database information set corresponding to the current preset similarity threshold from the preset information database, and determine the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity labels of the database information in the target database information set, wherein the number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the database information in the target database information set is greater than the current preset similarity threshold;

[0165] The multi-granularity label determination module is used to determine the multi-granularity label of the current information based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set.

[0166] According to one or more embodiments of this disclosure, an information recommendation device is provided, comprising:

[0167] An input data determination module is used to determine input data based on multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information library, and the multi-granularity labels of the information in the preset information library are determined based on any of the label determination methods described in the embodiments of this disclosure;

[0168] The data input module is used to input the input data into a preset information recommendation model;

[0169] The information recommendation module is used to determine the target information to be recommended from the candidate information based on the output of the preset information recommendation model.

[0170] According to one or more embodiments of this disclosure, an electronic device is provided, the electronic device comprising:

[0171] One or more processors;

[0172] Storage device for storing one or more programs.

[0173] When the one or more programs are executed by the one or more processors, the one or more processors implement the tag determination method and / or information recommendation method provided in the embodiments of this disclosure.

[0174] According to one or more embodiments of the present disclosure, a storage medium comprising computer-executable instructions is provided, which, when executed by a computer processor, are used to perform the tag determination method and / or information recommendation method provided in the embodiments of the present disclosure.

[0175] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0176] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0177] Although the subject matter has been described using language specific to structural features and / or methodological logic, 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. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A label determination method, characterized in that, include: Determine the similarity between the current information to be added to the database and the information in each of the preset databases; For each preset similarity threshold in the preset similarity threshold set, a target database information set corresponding to the current preset similarity threshold is determined from the preset information database, and a sub-tag corresponding to the current preset similarity threshold is determined based on the target multi-granularity tags of the database information in the target database information set. The number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the database information in the target database information set is greater than the current preset similarity threshold. Based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set, the multi-granularity label of the current information is determined. The multi-granularity is at least two granularities. For each granularity, a corresponding preset similarity threshold is preset. The preset similarity thresholds corresponding to all granularities constitute the preset similarity threshold set, wherein each preset similarity threshold is different.

2. The method according to claim 1, characterized in that, Before determining the target database information set corresponding to the current preset similarity threshold from the preset information database for each preset similarity threshold in the preset similarity threshold set, the method further includes: Based on the similarity, a preset number of information with the highest similarity are selected from the preset information database to obtain an initial set of information in the database. Wherein, determining the target database information set corresponding to the current preset similarity threshold from the preset information database for each preset similarity threshold in the preset similarity threshold set includes: For each preset similarity threshold in the preset similarity threshold set, the target database information set corresponding to the current preset similarity threshold is determined from the initial database information set.

3. The method according to claim 1, characterized in that, The determination of the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity label of the information in the target database set includes: For each multi-granularity tag corresponding to the information in the target database, determine the similarity to the information in the database to which the current multi-granularity tag belongs, and accumulate the determined similarities to obtain the tag score of the current multi-granularity tag; The multi-granularity label with the highest label score is determined as the target multi-granularity label, and the sub-label corresponding to the current preset similarity threshold is determined based on the target multi-granularity label.

4. The method according to claim 1, characterized in that, Also includes: If the information set in the target database is empty, the sub-tag corresponding to the current preset similarity threshold is determined based on the information identifier of the current information.

5. The method according to claim 4, characterized in that, The information identifier is a unique identifier for the current information.

6. The method according to claim 1, characterized in that, Determining the similarity between the current information to be added to the database and the information in each database of the preset database includes: Obtain the current dense features corresponding to the current information to be added to the database; For each piece of information in the preset information database, calculate the similarity between the current dense feature and the corresponding dense feature in the current database, and obtain the similarity between the current information and each piece of information in the database.

7. An information recommendation method, characterized in that, include: Input data is determined based on the multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information database, and the multi-granularity labels of the information in the preset information database are determined based on the label determination method as described in any one of claims 1-6; The input data is fed into a preset information recommendation model; Based on the output of the preset information recommendation model, the target information to be recommended is determined from the candidate information.

8. A label determining device, characterized in that, include: The similarity determination module is used to determine the similarity between the current information to be added to the database and the information in each database in the preset information database; The sub-label determination module is used to determine, for each preset similarity threshold in the preset similarity threshold set, the target database information set corresponding to the current preset similarity threshold from the preset information database, and determine the sub-label corresponding to the current preset similarity threshold based on the target multi-granularity labels of the database information in the target database information set, wherein the number of preset similarity thresholds in the preset similarity threshold set is at least two, and the similarity of the database information in the target database information set is greater than the current preset similarity threshold; The multi-granularity label determination module is used to determine the multi-granularity label of the current information based on the sub-labels corresponding to each preset similarity threshold in the preset similarity threshold set. The multi-granularity is at least two granularities. A corresponding preset similarity threshold is preset for each granularity. The preset similarity thresholds corresponding to all granularities constitute a preset similarity threshold set, wherein each preset similarity threshold is different.

9. An information recommendation device, characterized in that, include: An input data determination module is used to determine input data based on multi-granularity labels of candidate information, wherein the candidate information includes information in a preset information library, and the multi-granularity labels of the information in the preset information library are determined based on the label determination method as described in any one of claims 1-6; The data input module is used to input the input data into a preset information recommendation model; The information recommendation module is used to determine the target information to be recommended from the candidate information based on the output of the preset information recommendation model.

10. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

11. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform the method as described in any one of claims 1-7.

Images