Related product recommendation method and recommendation device

Abstract

The present application provides a method and device for recommending related products, which are applied to the field of recommendation systems and are conducive to improving the accuracy of recommending related products. The method includes: performing content recognition on target information to obtain target topic labels, target concept labels and target entity labels of the target information; determining multiple candidate entities based on a pre-constructed three-level hierarchical label tree, target topic labels, target concept labels and target entity labels; determining the degree of association between each candidate entity in the multiple candidate entities and the target information based on the three-level hierarchical label tree; and selecting at least one entity from the multiple candidate entities as a related product of the target information based on the degree of association between each candidate entity and the target information.

Description

Technical Field

[0001] The present application relates to the field of recommendation systems, and more specifically, to a method and device for recommending related products. Background Art

[0002] With the rapid development of the Internet and the domestic securities market, Internet information is updated more frequently, and financial information provides data assistance for users to trade and invest and grasp market dynamics. The market usually explodes with a lot of information in a short period of time, and too much information will make users overwhelmed. How to accurately and quickly obtain relevant information on funds, stocks, foreign exchange, etc. from a large amount of information and display it to users in a timely manner becomes very difficult.

[0003] Currently, we can extract keywords from the information content and then perform simple term matching between the keywords and financial products, or determine one or more stocks related to the information content by calculating the similarity between the keyword vector and the stock word vector.

[0004] However, the above method cannot accurately describe the information content, and the accuracy of the financial products related to the information content recommended to users is not high. Summary of the invention

[0005] The present application provides a method and device for recommending related products, which are conducive to improving the accuracy of recommending related products.

[0006] In the first aspect, a method for recommending associated products is provided, the method comprising: performing content recognition on target information, obtaining a target topic tag, a target concept tag and a target entity tag of the target information, wherein the target topic tag is used to indicate the product type corresponding to the target information, the target concept tag is used to indicate the key phrase in the target information, and the target entity tag is used to indicate the product corresponding to the target information. Based on a pre-constructed three-level hierarchical tag tree, a target topic tag, a target concept tag and a target entity tag, multiple candidate entities are determined, the three-level hierarchical tag tree indicates the association relationship between multiple topic tags, multiple concept tags and multiple entity tags, the multiple topic tags include the target topic tag, the multiple concept tags include the target concept tag, and the multiple entity tags include the target entity tag. Based on the three-level hierarchical tag tree, the degree of association between each candidate entity in the multiple candidate entities and the target information is determined. Based on the degree of association between each candidate entity and the target information, at least one entity is selected from the multiple candidate entities as an associated product of the target information.

[0007] In the present application, based on a pre-constructed three-level hierarchical tag tree, the data processing device can utilize the hierarchical relationship of the three-level hierarchical tag tree and the correlation between the upper and lower layers in the three-level hierarchical tag tree, and combine the target topic tag, target concept tag and target entity tag of the target information identified according to the target information to calculate the correlation between the target information and the product, and recommend at least one related product to the user based on the correlation between the target information and the product.

[0008] The three-level hierarchical label tree of the present application is constructed based on a large amount of information content and products. It can deeply mine the semantic features of information content and accurately depict the relationship between information content and business products through the correlation between the upper and lower layers in the three-level hierarchical label tree. In this way, the products related to the information content recommended to users are more explainable and more accurate.

[0009] In combination with the first aspect, in certain implementations of the first aspect, based on a pre-constructed three-level hierarchical tag tree, a target topic tag, a target concept tag, and a target entity tag, multiple candidate entities are determined, including: determining an entity corresponding to an entity tag under a target topic tag in the three-level hierarchical tag tree as a first candidate entity. Determining an entity corresponding to an entity tag under a target concept tag in the three-level hierarchical tag tree as a second candidate entity. Determining an entity corresponding to the target entity tag as a third candidate entity. Determining the first candidate entity, the second candidate entity, and the third candidate entity as multiple candidate entities.

[0010] In combination with the first aspect, in certain implementations of the first aspect, the three-level hierarchical tag tree includes the association degree between multiple topic tags and at least one concept tag under the multiple topic tags, and the association degree between multiple concept tags and at least one entity tag under the multiple concept tags.

[0011] Based on a three-level hierarchical label tree, the association degree between each candidate entity in a plurality of candidate entities and the target information is determined, including: determining the association degree between each candidate entity in the three-level hierarchical label tree and the target topic label as a first association degree, and the association degree between each candidate entity and the target topic label is determined by the association degree between each candidate entity and the concept label under the target topic label and the association degree between the concept label under the target topic label and the target topic label. Determine the association degree between each candidate entity in the three-level hierarchical label tree and the target concept label as a second association degree. Determine the association degree between each candidate entity and the entity corresponding to the target entity label as a third association degree. Determine the sum of the first association degree, the second association degree and the third association degree as the association degree between each candidate entity and the target information.

[0012] In combination with the first aspect, in some implementations of the first aspect, if a first candidate entity among multiple candidate entities belongs to an entity corresponding to a target entity label, the degree of association between the first candidate entity and the entity corresponding to the target entity label is 1. If the first candidate entity does not belong to an entity corresponding to the target entity label, the degree of association between the first candidate entity and the entity corresponding to the target entity label is 0.

[0013] In combination with the first aspect, in certain implementations of the first aspect, based on the massive historical information, multiple topic tags, multiple concept tags, and multiple entity tags are determined. The associations between the multiple topic tags and the multiple concept tags, and the associations between the multiple concept tags and the multiple entity tags are calculated. Based on the associations between the multiple topic tags and the multiple concept tags, and the associations between the multiple concept tags and the multiple entity tags, a three-level hierarchical tag tree is constructed.

[0014] In combination with the first aspect, in certain implementations of the first aspect, based on the correlation between each candidate entity and the target information, at least one entity is selected from multiple candidate entities as an associated product of the target information, including: selecting at least one entity with the greatest correlation with the target information from multiple candidate entities as an associated product of the target information.

[0015] In a second aspect, a device for recommending related products is provided, which is used to execute the method in any possible implementation of the first aspect. Specifically, the device includes a module for executing the method in any possible implementation of the first aspect.

[0016] In a third aspect, a device for recommending related products is provided, comprising a processor, the processor being coupled to a memory and being configured to execute instructions in the memory to implement the method in any possible implementation of the first aspect. Optionally, the device further comprises a memory. Optionally, the device further comprises a communication interface, the processor being coupled to the communication interface.

[0017] In a fourth aspect, a processor is provided, comprising: an input circuit, an output circuit and a processing circuit. The processing circuit is used to receive a signal through the input circuit and transmit a signal through the output circuit, so that the processor executes the method in any possible implementation of the first aspect.

[0018] In the specific implementation process, the above-mentioned processor can be a chip, the input circuit can be an input pin, the output circuit can be an output pin, and the processing circuit can be a transistor, a gate circuit, a trigger, and various logic circuits. The input signal received by the input circuit can be, for example, but not limited to, received and input by a receiver, and the signal output by the output circuit can be, for example, but not limited to, output to a transmitter and transmitted by the transmitter, and the input circuit and the output circuit can be the same circuit, which is used as an input circuit and an output circuit at different times. This application does not limit the specific implementation of the processor and various circuits.

[0019] In a fifth aspect, a processing device is provided, comprising a processor and a memory. The processor is used to read instructions stored in the memory, and can receive signals through a receiver and transmit signals through a transmitter to execute the method in any possible implementation of the first aspect.

[0020] Optionally, there are one or more processors and one or more memories.

[0021] Optionally, the memory may be integrated with the processor, or the memory may be provided separately from the processor.

[0022] In the specific implementation process, the memory can be a non-transitory memory, such as a read-only memory (ROM), which can be integrated with the processor on the same chip or can be set on different chips. This application does not limit the type of memory and the setting method of the memory and the processor.

[0023] It should be understood that the relevant data interaction process, such as sending indication information, can be a process of outputting indication information from a processor, and receiving capability information can be a process of receiving input capability information from a processor. Specifically, the processed output data can be output to a transmitter, and the input data received by the processor can come from a receiver. Among them, the transmitter and the receiver can be collectively referred to as a transceiver.

[0024] The processing device in the fifth aspect mentioned above can be a chip. The processor can be implemented by hardware or by software. When implemented by hardware, the processor can be a logic circuit, an integrated circuit, etc.; when implemented by software, the processor can be a general-purpose processor, which is implemented by reading the software code stored in the memory. The memory can be integrated in the processor or can be located outside the processor and exist independently.

[0025] In a sixth aspect, a computer program product is provided, the computer program product comprising: a computer program (also referred to as code, or instruction), which, when executed, enables a computer to execute a method in any possible implementation of the first aspect.

[0026] In a seventh aspect, a computer-readable storage medium is provided, which stores a computer program (also referred to as code, or instructions) which, when executed on a computer, enables the computer to execute a method in any possible implementation of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a schematic flow chart of a method for recommending associated products provided in an embodiment of the present application;

[0028] Figure 2 is a schematic flow chart of a method for constructing a three-level hierarchical label tree provided in an embodiment of the present application;

[0029] Figure 3 is a schematic diagram of a three-level hierarchical label tree provided in an embodiment of the present application;

[0030] Figure 4 is a schematic flow chart of another method for recommending associated products provided in an embodiment of the present application;

[0031] Figure 5 is a schematic block diagram of a device for recommending related products provided in an embodiment of the present application;

[0032] Figure 6 It is a schematic block diagram of another device for recommending related products provided in an embodiment of the present application. DETAILED DESCRIPTION

[0033] Before introducing the method and device for recommending related products provided in the embodiments of the present application, the following points are explained.

[0034] First, in the embodiments shown below, each term and English abbreviation, such as topic tags, concept tags, entity tags, etc., are illustrative examples given for the convenience of description and should not constitute any limitation to this application. This application does not exclude the possibility of defining other terms that can achieve the same or similar functions in existing or future protocols.

[0035] Second, the first, second and various numbers in the embodiments shown below are only used for the convenience of description and are not used to limit the scope of the embodiments of the present application. For example, different candidate entities are distinguished, different associations are distinguished, and the like.

[0036] Third, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b and c can mean: a, or b, or c, or a and b, or a and c, or b and c, or a, b and c, where a, b, c can be single or multiple.

[0037] In the implementation of calculating the association between information content and financial products, keywords can be extracted from the information content, and then simple term matching can be performed by combining keywords with financial products, or cosine similarity can be calculated between keyword vectors and stock word vectors. In addition, the association between information content and financial products can be calculated by building a label system based on financial business scenarios.

[0038] However, the above methods do not sufficiently explore the semantic features of information content and lack semantic understanding of information content, so the accuracy of financial products related to information content recommended to users is not high.

[0039] In view of this, an embodiment of the present application provides a method and device for recommending related products, which can calculate the correlation between real-time information content and financial products based on a three-layer label tree constructed by information content and product business and the correlation between upper nodes and lower nodes in the label tree, thereby recommending products associated with real-time information content to users.

[0040] Exemplarily, the present application may be applied to information content-related financial products, but the embodiments of the present application are not limited to this.

[0041] Figure 1 1 is a schematic flow chart of a method 100 for recommending related products provided in an embodiment of the present application. The method 100 may be executed by a data processing device. For example, the data processing device may be a server or other device with data processing function, which is not limited in the embodiment of the present application. The method 100 includes the following steps:

[0042] S101, performing content recognition on target information to obtain a target topic tag, a target concept tag, and a target entity tag of the target information.

[0043] Among them, the target topic label is used to indicate the product type corresponding to the target information, the target concept label is used to indicate the key phrases in the target information, and the target entity label is used to indicate the product corresponding to the target information.

[0044] S102, determining a plurality of candidate entities based on a pre-constructed three-level hierarchical tag tree, a target topic tag, a target concept tag, and a target entity tag.

[0045] The three-level hierarchical tag tree represents the association relationship between multiple topic tags, multiple concept tags and multiple entity tags. The multiple topic tags include the target topic tag, the multiple concept tags include the target concept tag, and the multiple entity tags include the target entity tag.

[0046] S103, based on the three-level hierarchical label tree, determine the correlation between each candidate entity in the plurality of candidate entities and the target information.

[0047] S104, based on the relevance between each candidate entity and the target information, selecting at least one entity from the plurality of candidate entities as a related product of the target information.

[0048] In an embodiment of the present application, based on a pre-constructed three-level hierarchical tag tree, the data processing device can utilize the hierarchical relationship of the three-level hierarchical tag tree and the correlation between the upper and lower layers in the three-level hierarchical tag tree, and combine the target topic tag, target concept tag and target entity tag of the target information identified according to the target information to calculate the correlation between the target information and the product, and recommend at least one related product to the user based on the correlation between the target information and the product.

[0049] The three-level hierarchical label tree in the embodiment of the present application is constructed based on a large amount of information content and business products. It can deeply mine the semantic features of the information content and accurately depict the relationship between the information content and the business products through the correlation between the upper and lower layers in the three-level hierarchical label tree. In this way, the products associated with the information content recommended to users are more interpretable and more accurate.

[0050] Figure 2 2 is a schematic flow chart of a method 200 for constructing a three-level hierarchical label tree provided in an embodiment of the present application. It should be understood that the steps of method 200 can be performed before the steps of method 100. Method 200 includes the following steps:

[0051] S201, based on massive historical information, determine multiple topic tags, multiple concept tags and multiple entity tags.

[0052] Exemplarily, the data processing device may cluster the information content according to the user's click behavior, and then manually annotate the clustering results to construct topic tags.

[0053] Optionally, the subject tag may include at least one of funds, stocks, bonds, insurance, financial management, and futures.

[0054] Exemplarily, the data processing device can extract key phrases from the massive information, semantically normalize and merge the key phrases, and use the effective key phrases as concept tags. In the embodiment of the present application, a class of entities with the same attributes is collectively referred to as concepts.

[0055] Exemplarily, the data processing device may form an entity tag from a product name, and other attributes of the product may be used as attributes of the entity tag to assist in linking entities.

[0056] It should be understood that the entity tag can be a specific fund, stock, bond, insurance, financial management or futures. For example, the entity tag e is a stock with a stock code of 12345x.

[0057] S202, calculating the association degree between the multiple topic tags and the multiple concept tags, and the association degree between the multiple concept tags and the multiple entity tags.

[0058] After constructing the topic tags, concept tags, and entity tags, the data processing device may calculate the association degree between the topic tags and the concept tags and the association degree between the concept tags and the entity tags.

[0059] Exemplarily, the steps of calculating the association between topic tags and concept tags are as follows:

[0060] Step 1: Count the number of information articles that include both topic label p and concept label c in the topics of massive information articles, denoted as

[0061] Step 2: Count the number of news articles that include the topic label p in the topics of the massive news articles, denoted as n p .

[0062] Step 3: The correlation between topic label p and concept label c is

[0063] Step 4: If p(p|c) < δ1, then the hierarchical relationship between the topic label p and the concept label c is not established. If p(p|c) ≥ δ1, then the hierarchical relationship between the topic label p and the concept label c is established, and the degree of association is p(p|c).

[0064] Exemplarily, the steps of calculating the association between concept labels and entity labels are as follows:

[0065] Step 1: Segment massive information articles and remove stop words.

[0066] Step 2: Use the trained entity recognition model to identify entities mentioned in massive information articles.

[0067] Step 3: Match the entity mentioned in the news article with the name, abbreviation, alias, code and other terms of the entity tag in the entity tag library. If the match is successful, link the entity mentioned in the news article to the entity tag. Assume that the matched entity tag is e.

[0068] Step 4: Based on the massive amount of information articles after word segmentation and the identified entity tag e, count the context words of the entity tag e and the words of the entity corresponding to the entity tag e, and calculate the correlation between the entity tag e and the concept tag c.

[0069] For example, the correlation between the entity tag e and the concept tag c can be expressed by the following formula:

[0070]

[0071] Among them, X represents the window length of the context word of entity tag e, x i The context word representing the entity tag e. If the key phrase corresponding to the concept tag c does not include the name, abbreviation, alias or code of the entity corresponding to the entity tag e, then execute the above formula (a). If the key phrase corresponding to the concept tag c includes the name, abbreviation, alias or code of the entity corresponding to the entity tag e, then execute the above formula (b). p(x i |e) is equal to the context word x containing the entity tag e i The ratio of the number of occurrences to the number of occurrences of the entity tag e.

[0072] In the above formula (a), p(c|x i ) can be expressed by the following formula:

[0073]

[0074] in, The key phrases corresponding to the concept label c include x i If the key phrase corresponding to the concept tag c includes x i , then execute the above formula (c), which represents the context word x of the entity corresponding to the entity tag e i The inverse of the length of the key phrase corresponding to the concept tag c is equivalent to normalizing the word length. If the key phrase corresponding to the concept tag c does not include x i substring, then execute the above formula (d).

[0075] Step 5: If p(c|e) < δ2, then the hierarchical relationship between the concept tag c and the entity tag e is not established. If p(c|e) ≥ δ2, then the association relationship between the topic tag p and the concept tag c is established, and its association degree is p(c|e).

[0076] S203, constructing a three-level hierarchical tag tree based on the associations between the multiple topic tags and the multiple concept tags, and the associations between the multiple concept tags and the multiple entity tags.

[0077] Figure 3 Schematic diagram of a three-level hierarchical label tree provided in an embodiment of the present application. Figure 3 As shown, the three-level hierarchical label tree is composed of topic label p, concept label c and entity label e from top to bottom. Among them, topic label p includes topic label p1, topic label p2 and topic label p3. Concept label c includes concept label c1, concept label c2 and concept label c3, and entity label e includes entity label e1, entity label e2 and entity label e3.

[0078] like Figure 3 As shown, topic label p1 has an association relationship with concept label c1 and concept label c3, topic label p2 has an association relationship with concept label c2 and concept label c3, and topic label p3 has an association relationship with concept label c3. Concept label c1 has an association relationship with entity label e1, concept label c2 has an association relationship with entity label e3, and concept label c3 has an association relationship with entity label e2.

[0079] Figure 4 4 is a schematic flow chart of another method 400 for recommending related products provided in an embodiment of the present application. The method 400 may be executed by a data processing device, but the embodiment of the present application does not limit this. The method 400 includes the following steps:

[0080] S401, identifying the target information through the trained topic classification model to obtain the target topic label of the target information.

[0081] Exemplarily, a data processing device can manually annotate massive information related to a topic, train a deep learning network with the annotated massive information to obtain a trained topic classification model, perform topic identification on target information that is updated daily in real time through the trained topic classification model, and obtain a target topic label for the target information.

[0082] Optionally, the data processing device may match the identified topic tag with the topic tags in the topic tag library. If the match is successful, the matched topic tag in the topic tag library is assigned to the target information to obtain the target topic tag of the target information. If the match fails, the identified topic tag is added to the topic tag library as a new topic tag, and the new topic tag is assigned to the target information to obtain the target topic tag of the target information.

[0083] S402, identifying the target information through a key phrase extraction model to obtain a target concept label of the target information.

[0084] Exemplarily, the data processing device can extract key phrases from the target information that is updated daily in real time through a key phrase extraction model, and calculate the maximum similarity between the extracted key phrases and the semantics of the concept tag library. If the maximum similarity is less than a preset threshold, the extracted key phrase is added to the concept tag library as a new concept tag, and the entity tag associated with the new concept tag is determined based on the method of calculating the association described above, as well as the subject tag associated with the new concept tag. If the maximum similarity is greater than or equal to the preset threshold, the concept tag corresponding to the extracted key phrase is merged with the concept tag in the concept tag library, and the concept tag in the concept tag library is assigned to the target information to obtain the target concept tag of the target information.

[0085] S403, identifying the target information through the trained entity recognition module to obtain the target entity label of the target information.

[0086] Exemplarily, the data processing device can identify the entities mentioned in the target information through a trained entity recognition module, and then match the entities mentioned in the target information with the names, aliases, abbreviations, full names, codes and other terms of the entity tags in the entity tag library. If the match is successful, the entity mentioned in the target information is assigned the corresponding entity tag in the entity tag library to obtain the target entity tag of the target information.

[0087] Exemplarily, the entity recognition module is a sequence labeling model. The above-mentioned topic classification model, key phrase extraction model and entity recognition module may have a deep learning network structure, and the use of deep learning can improve the accuracy of information tag recognition.

[0088] It should be understood that the topic tag library may include multiple topic tags in a three-level hierarchical tag tree constructed based on massive historical information. The concept tag library may include multiple concept tags in a three-level hierarchical tag tree constructed based on massive historical information. The entity tag library may include multiple entity tags in a three-level hierarchical tag tree constructed based on massive historical information.

[0089] It should also be understood that the number of the above-mentioned target topic tags, target concept tags, and target entity tags can be at least one, but the embodiments of the present application are not limited to this.

[0090] For example, the target information d has k target topic labels, namely p1, p2, ..., p k , there are m target concept labels, namely c1, c2, ..., c m , there are n target entity labels, namely e1, e2, ..., e n .

[0091] S404: Determine the entity corresponding to the entity tag under the target topic tag in the three-level hierarchical tag tree as the first candidate entity.

[0092] In this step, the data processing device can determine that the entity corresponding to all associated entity tags under the target topic tag of the target information d is the first candidate entity based on the constructed three-level hierarchical tag tree.

[0093] by Figure 3 Taking the three-level hierarchical tag tree shown as an example, it is assumed that the target topic tags of the target information d include p1 and p3, wherein the entity tags under the target topic tag p1 include e1 and e2, and the entity tags under the target topic tag p3 include e2. In this way, all the associated entity tags under the target topic tag of the target information d include e1 and e2, and the determined first candidate entity includes the entities corresponding to the entity tags e1 and e2.

[0094] S405: Determine the entity corresponding to the entity tag under the target concept tag in the three-level hierarchical tag tree as the second candidate entity.

[0095] In this step, the data processing device can determine that the entities corresponding to all associated entity tags under the target concept tag of the target information d are the second candidate entities based on the constructed three-level hierarchical tag tree.

[0096] by Figure 3 Taking the three-level hierarchical label tree shown as an example, it is assumed that the target concept labels of the target information d include c1 and c3, wherein the entity labels under the target concept label c1 include e1, and the entity labels under the target concept label c3 include e2. In this way, all the associated entity labels under the target concept label of the target information d include e1 and e2, and the determined second candidate entities include the entities corresponding to the entity labels e1 and e2.

[0097] S406: Determine the entity corresponding to the target entity tag as the third candidate entity.

[0098] In this step, the data processing device can determine that the entity corresponding to the target entity tag of the target information d is the third candidate entity based on the constructed three-level hierarchical tag tree.

[0099] by Figure 3 Taking the three-level hierarchical label tree shown as an example, assuming that the target entity label of the target information d includes e2, the determined third candidate entity includes the entity corresponding to the entity label e2.

[0100] S407: Determine the first candidate entity, the second candidate entity, and the third candidate entity as the multiple candidate entities.

[0101] In this step, the data processing device may merge the first candidate entity, the second candidate entity and the third candidate entity to obtain the above-mentioned multiple candidate entities. Taking the first candidate entity, the second candidate entity and the third candidate entity determined in step S404, step S405 and step S406 as an example, the above-mentioned multiple candidate entities may include entities corresponding to entity labels e1 and e2.

[0102] S408, calculating the correlation between each candidate entity among the plurality of candidate entities and the target information.

[0103] For example, the candidate entity e z The correlation between (z∈(1,2,……,n)) and the target information d can be expressed by the following formula:

[0104]

[0105] in, Represents the sum of the associations of entity tags associated with all target topic tags of target information d. represents the sum of the associations of entity labels under all target concept labels of target information d. 1 means that if entity e z is the entity corresponding to the target entity label identified by the target information d, that is, e z ∈(1,2,……,n), then entity e z The correlation degree between entity e and target information d is 1. z If entity e is not the entity corresponding to the target entity label identified by target information d, then entity e z The correlation with the target information d is 0.

[0106] In the above formula, p(e z |p i )=∑ j p(e z |c j )p(c j |p i ), p(e z|c j ) represents entity e z With concept label c j The correlation between them, p(c j |p i ) represents the concept label c j With the hashtag p i The degree of correlation between them.

[0107] S409, selecting at least one entity from the plurality of candidate entities as a related product of the target information according to the degree of association between each candidate entity and the target information.

[0108] In the embodiment of the present application, the data processing device can obtain multiple candidate entities associated with the target information based on the constructed three-level hierarchical label tree. In this way, the semantic content of the target information can be identified based on the three granularities of topic label, concept label and entity label, which can more accurately characterize the target information.

[0109] Optionally, S409 includes: selecting at least one entity with the greatest correlation with the target information from the multiple candidate entities as a related product of the target information.

[0110] In an embodiment of the present application, the data processing device can obtain the degree of association between each candidate entity and the target information based on the formula of the degree of association between each candidate entity and the target information, and then sort the degrees of association between multiple candidate entities and the target information from large to small, and select the top N entities with the largest association as related products of the target information to recommend to the user.

[0111] Exemplarily, N can be set to 3, but the embodiment of the present application does not limit the value of N.

[0112] It should be understood that the sequence numbers of the above processes do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0113] Combined with the above Figures 1 to 4 , describes in detail the method for recommending related products according to the embodiment of the present application, and will now be combined with Figure 5 and Figure 6 A device for recommending related products according to an embodiment of the present application is described in detail.

[0114] Figure 5 A schematic block diagram of a device 500 for recommending related products provided in an embodiment of the present application is shown. The device 500 includes an acquisition module 510 and a processing module 520 .

[0115] Among them, the acquisition module 510 is used to: perform content recognition on the target information, obtain the target topic label, target concept label and target entity label of the target information, the target topic label is used to indicate the product type corresponding to the target information, the target concept label is used to indicate the key phrase in the target information, and the target entity label is used to indicate the product corresponding to the target information. The processing module 520 is used to: determine multiple candidate entities based on the pre-constructed three-level hierarchical label tree, target topic label, target concept label and target entity label, the three-level hierarchical label tree indicates the association relationship between multiple topic labels, multiple concept labels and multiple entity labels, multiple topic labels include the target topic label, multiple concept labels include the target concept label, and multiple entity labels include the target entity label; based on the three-level hierarchical label tree, determine the association degree between each candidate entity in the multiple candidate entities and the target information; and, based on the association degree between each candidate entity and the target information, select at least one entity from the multiple candidate entities as the associated product of the target information.

[0116] Optionally, the processing module 520 is used to: determine the entity corresponding to the entity tag under the target topic tag in the three-level hierarchical tag tree as the first candidate entity; determine the entity corresponding to the entity tag under the target concept tag in the three-level hierarchical tag tree as the second candidate entity; determine the entity corresponding to the target entity tag as the third candidate entity; and determine the first candidate entity, the second candidate entity and the third candidate entity as multiple candidate entities.

[0117] Optionally, the three-level hierarchical label tree includes the association degree between multiple topic labels and at least one concept label under the multiple topic labels, and the association degree between multiple concept labels and at least one entity label under the multiple concept labels. The processing module 520 is used to: determine the association degree between each candidate entity and the target topic label in the three-level hierarchical label tree as the first association degree, and the association degree between each candidate entity and the target topic label is determined by the association degree between each candidate entity and the concept label under the target topic label and the association degree between the concept label under the target topic label and the target topic label; determine the association degree between each candidate entity and the target concept label in the three-level hierarchical label tree as the second association degree; determine the association degree between each candidate entity and the entity corresponding to the target entity label as the third association degree; and determine the sum of the first association degree, the second association degree and the third association degree as the association degree between each candidate entity and the target information.

[0118] Optionally, if the first candidate entity among the multiple candidate entities belongs to the entity corresponding to the target entity label, the association degree between the first candidate entity and the entity corresponding to the target entity label is 1. If the first candidate entity does not belong to the entity corresponding to the target entity label, the association degree between the first candidate entity and the entity corresponding to the target entity label is 0.

[0119] Optionally, the processing module 520 is used to: determine multiple topic tags, multiple concept tags and multiple entity tags based on massive historical information; calculate the association between multiple topic tags and multiple concept tags, and the association between multiple concept tags and multiple entity tags; and, based on the association between multiple topic tags and multiple concept tags, and the association between multiple concept tags and multiple entity tags, construct a three-level hierarchical tag tree.

[0120] Optionally, the processing module 520 is used to select at least one entity with the greatest correlation with the target information from multiple candidate entities as a related product of the target information.

[0121] In an optional example, those skilled in the art can understand that the device 500 can be specifically a data processing device in the above embodiment, or the functions of the data processing device in the above embodiment can be integrated in the device 500. The above functions can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The device 500 can be used to execute the various processes and / or steps corresponding to the data processing device in the above method embodiment.

[0122] It should be understood that the apparatus 500 herein is embodied in the form of functional modules. The term "module" herein may refer to an application specific integrated circuit (ASIC), an electronic circuit, a processor (e.g., a shared processor, a dedicated processor, or a group processor, etc.) and a memory for executing one or more software or firmware programs, a combined logic circuit, and / or other suitable components that support the described functions. In an embodiment of the present application, Figure 5 The device 500 may also be a chip or a chip system, such as a system on chip (SoC).

[0123] Figure 6 6 is a schematic block diagram of another device 600 for recommending related products provided in an embodiment of the present application. The device 600 includes a processor 610, a transceiver 620, and a memory 630. The processor 610, the transceiver 620, and the memory 630 communicate with each other through an internal connection path, the memory 630 is used to store instructions, and the processor 610 is used to execute the instructions stored in the memory 630 to control the transceiver 620 to send and / or receive signals.

[0124] It should be understood that the device 600 can be specifically a data processing device in the above-mentioned embodiment, or the function of the data processing device in the above-mentioned embodiment can be integrated in the device 600, and the device 600 can be used to execute the various steps and / or processes corresponding to the data processing device in the above-mentioned method embodiment. Optionally, the memory 630 may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type. The processor 610 may be used to execute the instructions stored in the memory, and when the processor executes the instructions, the processor may execute the various steps and / or processes corresponding to the data processing device in the above-mentioned method embodiment.

[0125] It should be understood that in the embodiment of the present application, the processor 610 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0126] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software. The steps of the method disclosed in conjunction with the embodiment of the present application can be directly embodied as a hardware processor for execution, or a combination of hardware and software modules in a processor for execution. The software module can be located in a storage medium mature in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in a memory, and the processor executes the instructions in the memory, and completes the steps of the above method in conjunction with its hardware. To avoid repetition, it is not described in detail here.

[0127] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0128] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0129] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0130] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0131] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0132] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0133] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for recommending related products, characterized in that: include: Performing content recognition on the target information to obtain a target topic tag, a target concept tag, and a target entity tag of the target information, wherein the target topic tag is used to indicate the product type corresponding to the target information, the target concept tag is used to indicate a key phrase in the target information, and the target entity tag is used to indicate the product corresponding to the target information; Determine multiple candidate entities based on a pre-constructed three-level hierarchical label tree, the target topic label, the target concept label, and the target entity label, wherein the three-level hierarchical label tree represents associations among multiple topic labels, multiple concept labels, and multiple entity labels, wherein the multiple topic labels include the target topic label, the multiple concept labels include the target concept label, and the multiple entity labels include the target entity label; Based on the three-level hierarchical label tree, determining the degree of association between each candidate entity in the plurality of candidate entities and the target information; Based on the relevance between each candidate entity and the target information, selecting at least one entity from the plurality of candidate entities as a related product of the target information; The step of determining a plurality of candidate entities based on the pre-constructed three-level hierarchical label tree, the target topic label, the target concept label and the target entity label comprises: Determine the entity corresponding to the entity tag under the target topic tag in the three-level hierarchical tag tree as the first candidate entity; Determine the entity corresponding to the entity tag under the target concept tag in the three-level hierarchical tag tree as the second candidate entity; Determine the entity corresponding to the target entity label as a third candidate entity; determining the first candidate entity, the second candidate entity, and the third candidate entity as the plurality of candidate entities; The three-level hierarchical tag tree includes the association degree between the multiple topic tags and at least one concept tag under the multiple topic tags, and the association degree between the multiple concept tags and at least one entity tag under the multiple concept tags; The determining, based on the three-level hierarchical label tree, the degree of association between each candidate entity in the plurality of candidate entities and the target information comprises: Determining the degree of association between each candidate entity and the target topic label in the three-level hierarchical label tree as a first degree of association, wherein the degree of association between each candidate entity and the target topic label is determined by the degree of association between each candidate entity and a concept label under the target topic label and the degree of association between the concept label under the target topic label and the target topic label; Determine the degree of association between each candidate entity in the three-level hierarchical label tree and the target concept label as a second degree of association; Determine the degree of association between each candidate entity and the entity corresponding to the target entity label as a third degree of association; The sum of the first degree of association, the second degree of association and the third degree of association is determined as the degree of association between each candidate entity and the target information.

2. The method according to claim 1, characterized in that If a first candidate entity among the multiple candidate entities belongs to the entity corresponding to the target entity label, the association degree between the first candidate entity and the entity corresponding to the target entity label is 1; If the first candidate entity does not belong to the entity corresponding to the target entity label, the association degree between the first candidate entity and the entity corresponding to the target entity label is 0.

3. The method according to claim 1, characterized in that The method further comprises: Based on the massive historical information, determining the plurality of topic tags, the plurality of concept tags, and the plurality of entity tags; Calculating associations between the multiple topic tags and the multiple concept tags, and associations between the multiple concept tags and the multiple entity tags; The three-level hierarchical tag tree is constructed based on the associations between the multiple topic tags and the multiple concept tags, and the associations between the multiple concept tags and the multiple entity tags.

4. The method according to any one of claims 1 to 3, characterized in that The selecting at least one entity from the plurality of candidate entities as a related product of the target information includes: At least one entity having the greatest relevance to the target information is selected from the plurality of candidate entities as a related product of the target information.

5. A device for recommending related products, characterized in that: include: an acquisition module, configured to perform content recognition on target information, and obtain a target topic tag, a target concept tag, and a target entity tag of the target information, wherein the target topic tag is used to indicate a product type corresponding to the target information, the target concept tag is used to indicate a key phrase in the target information, and the target entity tag is used to indicate a product corresponding to the target information; A processing module, configured to determine an entity corresponding to an entity tag under the target topic tag in a pre-constructed three-level hierarchical tag tree as a first candidate entity; and determine an entity corresponding to an entity tag under the target concept tag in the three-level hierarchical tag tree as a second candidate entity; Determine the entity corresponding to the target entity tag as a third candidate entity; and determine the first candidate entity, the second candidate entity and the third candidate entity as a plurality of candidate entities, wherein the three-level hierarchical tag tree represents an association relationship between a plurality of topic tags, a plurality of concept tags and a plurality of entity tags, the plurality of topic tags include the target topic tag, the plurality of concept tags include the target concept tag, and the plurality of entity tags include the target entity tag; The processing module is also used to: determine the degree of association between each of the multiple candidate entities in the three-level hierarchical label tree and the target topic label as a first degree of association, and the degree of association between each candidate entity and the target topic label is determined by the degree of association between each candidate entity and the concept label under the target topic label and the degree of association between the concept label under the target topic label and the target topic label; determine the degree of association between each candidate entity in the three-level hierarchical label tree and the target concept label as a second degree of association; determine the degree of association between each candidate entity and the entity corresponding to the target entity label as a third degree of association; and determine the sum of the first degree of association, the second degree of association and the third degree of association as the degree of association between each candidate entity and the target information, wherein the three-level hierarchical label tree includes the degree of association between the multiple topic labels and at least one concept label under the multiple topic labels, and the degree of association between the multiple concept labels and at least one entity label under the multiple concept labels; The processing module is further configured to: based on the degree of association between each candidate entity and the target information, select at least one entity from the plurality of candidate entities as a related product of the target information.

6. The recommendation device according to claim 5, characterized in that: If a first candidate entity among the multiple candidate entities belongs to the entity corresponding to the target entity label, the association degree between the first candidate entity and the entity corresponding to the target entity label is 1; If the first candidate entity does not belong to the entity corresponding to the target entity label, the association degree between the first candidate entity and the entity corresponding to the target entity label is 0.

7. The recommendation device according to claim 5, characterized in that: The processing module is used for: Based on the massive historical information, determining the plurality of topic tags, the plurality of concept tags, and the plurality of entity tags; Calculating associations between the multiple topic tags and the multiple concept tags, and associations between the multiple concept tags and the multiple entity tags; The three-level hierarchical tag tree is constructed based on the associations between the multiple topic tags and the multiple concept tags, and the associations between the multiple concept tags and the multiple entity tags.

8. The recommendation device according to any one of claims 5 to 7, characterized in that The processing module is used for: At least one entity having the greatest relevance to the target information is selected from the plurality of candidate entities as a related product of the target information.

9. A device for recommending related products, characterized in that: include: A processor, wherein the processor is coupled to a memory, wherein the memory is used to store a computer program, and when the processor calls the computer program, the device executes the method according to any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, the computer program comprising instructions for implementing the method according to any one of claims 1 to 4.

11. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 4 when being executed by a processor.

Images