Resource recommendation information push method, device, computer equipment and storage medium

Abstract

The present application provides a resource recommendation information push method, device, computer equipment and storage medium, which relates to the field of data processing technology and is used to solve the problem of low recommendation accuracy of existing resource recommendation schemes. The method part includes: collecting the interaction information between the terminal and the target user; extracting the named entity from the interaction information to obtain the resource entity; inputting the resource entity into the vector conversion model to obtain the resource entity vector; matching the target resource from the resource knowledge graph based on the resource entity and the resource entity vector; generating resource recommendation information based on the target resource, and pushing the resource recommendation information to the target user, thereby improving the accuracy of resource recommendation.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a resource recommendation information pushing method, resource recommendation information pushing device, computer equipment and storage medium. Background Art

[0002] The existing recommendation information push solution analyzes the user's historical behavior (such as the user's click, browse, purchase records of in-store dishes, etc.), and then uses a neural network model that can predict user behavior to determine the recommended information to push to customers. However, this push solution is mostly suitable for old customers, and cannot make accurate recommendations for customers with less historical behavior. Summary of the invention

[0003] Based on this, it is necessary to provide a resource recommendation information push method for the above technical problems to solve the problem of low accuracy of existing recommendation information push solutions.

[0004] A first aspect of an embodiment of the present application provides a method for pushing resource recommendation information, including:

[0005] Collecting interaction information between the terminal and the target user; the interaction information includes the target user's intended resource information;

[0006] Performing named entity extraction on the intended resource information to obtain a resource entity;

[0007] Inputting the resource entity into a vector conversion model to obtain a resource entity vector;

[0008] Based on the resource entity and the resource entity vector, the target resource is matched from the resource knowledge graph; wherein the resource knowledge graph is used to characterize the correlation between different resource sample entities;

[0009] Generate resource recommendation information based on the target resource, and push the resource recommendation information to the target user.

[0010] A second aspect of an embodiment of the present application provides a resource recommendation information push device, including:

[0011] Collection module: used to collect interaction information between the terminal and the target user; the interaction information includes the target user's intended resource information;

[0012] Extraction module: used to extract named entities from the intended resource information to obtain resource entities;

[0013] Conversion module: used for inputting the resource entity into a vector conversion model to obtain a resource entity vector;

[0014] Matching module: used to match the target resource from the resource knowledge graph based on the resource entity and the resource entity vector; wherein the resource knowledge graph is used to characterize the correlation between different resource sample entities;

[0015] Recommendation module: used to generate resource recommendation information based on the target resource and push the resource recommendation information to the target user.

[0016] A third aspect of an embodiment of the present application provides a computer device, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor implements the above-mentioned resource recommendation information push method when executing the computer-readable instructions.

[0017] A fourth aspect of an embodiment of the present application provides one or more readable storage media storing computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the above-mentioned resource recommendation information pushing method.

[0018] The embodiment of the present application provides a resource recommendation information push method, device, computer equipment and storage medium, which obtains the interactive information between the terminal and the target user through human-computer interaction, wherein the interactive information includes the intended resource information of the target user, and the resource entity corresponding to the intended resource information of the target user can be obtained by performing named entity extraction on the intended resource information. Then, the resource entity is mapped to the vector space through the vector conversion model to obtain the resource entity vector. Since the pre-constructed resource knowledge graph includes a large number of resource sample entities and can characterize the correlation between different resource sample entities, and the correlation between different resource sample entities can be obtained through vector calculation, the target resource of interest to the target user can be matched from the resource knowledge graph based on the resource entity and resource entity vector corresponding to the intended resource information of the target user, and then the target resource is generated according to the target resource, and the target resource is pushed to the target user to meet the needs of each user and improve the accuracy of resource recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0020] Figure 1 It is a schematic diagram of the implementation process of the resource recommendation information push method in the embodiment of the present application;

[0021] Figure 2It is a schematic diagram of the implementation process of the resource recommendation information push method in another embodiment of the present application;

[0022] Figure 3 It is a schematic diagram of the implementation process of the resource recommendation information push method in another embodiment of the present application;

[0023] Figure 4 It is a structural diagram of a resource recommendation information push device in an embodiment of the present application;

[0024] Figure 5 It is a schematic diagram of a computer device in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0026] The present application provides a resource recommendation information push method. Taking the application of this implementation scheme to a terminal device as an example, the terminal device communicates with the target user through questions and answers, etc., wherein the intended resource information may include intended music information, intended dish information, intended store information, etc., and records the interaction information at the same time. Then, the resource entities corresponding to the intended resource information are extracted from the interaction information, such as dish entities, etc., and these resource entities are mapped to a high-dimensional space through a vector conversion model to obtain resource entity vectors. Finally, based on the resource entities and resource entity vectors, the target resources are matched from the pre-constructed resource knowledge graph, and these target resources are recommended to the target user, which can be displayed on the user interface of the target user terminal. It should be noted that the resource knowledge graph refers to the knowledge graph corresponding to a single resource, such as the dish knowledge graph corresponding to dish information. Among them, the terminal device includes but is not limited to artificial intelligence terminals, various personal computers, laptops, smart phones, tablet computers, etc.

[0027] In specific implementation, the technical solution of the present application can also be applied to the server. The terminal sends the interaction information between the terminal and the user containing the user's intended resource information to the server, and the server performs named entity extraction, vector conversion, and resource information matching, etc., wherein the server can be an independent server, or a server cluster composed of several servers, or a cloud computing server, etc., which is not limited here.

[0028] See also Figure 1 , Figure 1The flowchart of the resource recommendation information push method in the embodiment of the present application is shown, and the method is applied to a terminal device as an example for explanation, including the following steps:

[0029] S11: Collecting interaction information between the terminal and the target user; the interaction information includes the target user's intended resource information.

[0030] In step S11, the interactive information between the terminal and the target user includes the target user's intended resource information. The interactive information may be in the form of a chat such as text or voice based on a dialog box, or in the form of an intelligent dialogue. The target user refers to a customer who intends to order a meal. The intended resource information may include music information, store information, book information, dish information, etc. Specifically, for example, the dish information may include the name of the dish, the cooking method of the dish, the raw materials of the dish, the taste of the dish, the cuisine of the dish, the taboos of the dish, etc.

[0031] In this embodiment, the terminal communicates with the target user to obtain the target user's intended resource information. Specifically, since the intended resource information is included in the interactive information between the terminal and the target user, the terminal records the interactive information in real time and obtains the target user's intended resource information through analysis. As an implementation method, the user's intended resource information can be stored in a database to facilitate statistics of resource information with higher user intentions, and it can be added to the recommended information. At the same time, the database is updated regularly, and the data information with less selection is deleted in time to save system space. For example, the resource database is used to store the user's intended dish information or frequently visited and purchased dish information.

[0032] S12: Perform named entity extraction on the intended resource information to obtain resource entities.

[0033] In step S12, named entity extraction includes named entity recognition and named entity extraction, wherein named entity recognition is used to identify named entities such as names, place names, and organizational structure names in the corpus. Named entities are the research subject of named entity recognition, and generally include three major categories (entity class, time class, and number class) and seven subcategories (names, place names, organization names, time, date, currency, and percentage) of named entities. The main technical methods of named entity recognition include: rule-based and dictionary-based methods, statistical-based methods, and a hybrid of the two. After successfully performing named entity recognition, in order to extract the named entities in the text, it is necessary to extract the constituent named entity words individually and link phrases and phrases according to the tags, and the named entities can be extracted from the text by using regular or neural network methods. In this embodiment, the resource entity is not limited to the entity of the resource name, but can also be a resource-related information entity. For example, when the resource information is dish information, the corresponding resource entity includes but is not limited to the entity corresponding to the information such as cooking method, raw materials, taste, cuisine, and taboos.

[0034] In this embodiment, since the interactive information between the terminal and the target user contains the target user's intended resource information, the user's intended resource information can be extracted from the interactive information through named entity recognition and extraction. As an example, the resource entity is identified through NER (Named Entities Recognition), and each word of the named entity (the word is in the form of: word / named entity annotation) is extracted using a regular expression, and then the single word is connected using a function. For example, the resource information is dish information. Assuming that the text information obtained through human-computer interaction is "white-cut chicken belongs to the Cantonese cuisine series", the regular expression "[Luchuan Yuemin Jiangsu Zhejiang Hunan Anhui] cuisine" can be used to extract the "Cantonese cuisine" cuisine entity from the above text information. In other embodiments, neural networks and other methods can also be used to extract named entities.

[0035] As another example, resource entities are identified through NER, and resource entities are extracted through a neural network. Among them, the neural network method needs to perform sequence annotation in advance, and the sequence annotation includes BIO, BIOSE, IOB, etc., where B (Begin) represents the beginning of the entity segment, I (Internediate) represents the middle of the entity segment, E (End) represents the end of the entity segment, S (Single) represents the entity segment as a single word, and O (Other) represents that the character is not any entity. Taking the user's intended resource information as a dish information as an example, the dish entity can be sequenced using the BIO method, "I O want O to eat O dry B-food_name pot I-food_name cow I-food_name frog I-food_name ah O", and then BiLSTM (Bi-directional Long Short-Term Memory, bidirectional long and short word memory model) is used to extract sequence features, and CRF (Conditional Random Field, conditional random field) is used as the output layer to extract the entity "dry pot bullfrog". It should be noted that in actual application scenarios, other methods such as machine learning, deep learning, unsupervised / supervised learning, etc. can also be used to extract named entities.

[0036] As an embodiment of the present application, the method of performing named entity extraction on the intended resource information to obtain a resource entity includes: if the interactive information is interactive audio information, converting the interactive audio information into interactive text information; the interactive text information includes intended resource text information; and performing named entity extraction on the intended resource text information to obtain a resource entity.

[0037] In this embodiment, since named entity recognition is used to identify entities with specific meanings in text, such as names of people, places, organizations, proper nouns, etc., if the terminal and the target user communicate through human-computer dialogue, and the corresponding interactive information is interactive audio information, the interactive audio information needs to be converted into machine-recognizable interactive text information. Since the interactive information includes the target user's intended resource information, the intended resource text information is correspondingly included in the interactive text information, and then the resource entity is extracted based on the intended resource text information.

[0038] S13: Input the resource entity into a vector conversion model to obtain a resource entity vector.

[0039] In step S13, a vector conversion model is used to map the resource entity to a high-dimensional vector space and convert it into a resource entity vector. The vector conversion model can be a neural network model, such as BERT (Bidirectional Encoder Representations from Transformers, a pre-trained language representation model), word2vec, Fasttext, GloVe, and other models.

[0040] In this embodiment, after obtaining the resource entity corresponding to the target user's intended resource information, the resource entity is mapped into a vector space so that the computer can recognize and analyze the text features, wherein the text features can be represented as characters, words, sentences, or chapters according to different granularity. In this embodiment, a neural network model can be used to obtain the feature vector corresponding to the resource entity.

[0041] As an embodiment of the present application, inputting the resource entity into a vector conversion model to obtain a resource entity vector includes: performing word segmentation on the resource entity to obtain a segmented resource entity; and inputting the segmented resource entity into the vector conversion model to obtain a resource entity vector.

[0042] In this embodiment, in the process of vector conversion of resource entities, the resource entities are first preprocessed, that is, word segmentation. For automatic word segmentation of Chinese text, spaces or other boundary marks are automatically added between words in Chinese text. Word segmentation tools such as jieba can be used to segment resource entities. Then, the resource entities after word segmentation are input into the value vector conversion model to obtain a dish vector or matrix that can represent the hidden features between resource entities.

[0043] S14: Based on the resource entity and the resource entity vector, match the target resource from the resource knowledge graph.

[0044] In step S14, knowledge graph technology is an important component of artificial intelligence, which describes the relationship between concepts, entities and their keys in the objective world in a structured way. In this embodiment, the resource knowledge graph can be understood as a more structured resource knowledge base based on a graph, which can be a recommended music knowledge base, a dish knowledge base, etc. It not only includes a large number of resource sample entities, but also can be used to characterize the correlation between different resource sample entities. The correlation between different resource sample entities can be calculated based on the resource sample entity vectors corresponding to each resource sample entity. It should be noted that in this embodiment, the resource knowledge graph refers to the knowledge graph of a single resource, which is used to characterize the correlation between resource sample entities of a single resource. For example, the dish knowledge graph contains entities of dishes and dish-related information, which are used to characterize the correlation between dish entities.

[0045] In this embodiment, the correlation between different resource entities can be determined by the distance between the resource entities. The closer the distance, the stronger the correlation. In actual calculation, the distance between resource entities can be calculated based on the resource entity vector corresponding to the resource entity, for example, by calculating the distance values ​​of the resource entity vector Cosine, Jaccard, Pearson, etc., to obtain the correlation between different resource entities. The smaller the distance value, the stronger the correlation between the resource entities. Since the pre-constructed resource knowledge graph can be used to characterize the correlation between different resource samples, the resource sample entity that is the same or similar to the resource entity corresponding to the user's intended resource information can be matched from the resource knowledge graph based on the resource entity and the resource entity vector as the target resource entity, thereby obtaining a target resource that is more in line with the user's selection.

[0046] As an embodiment of the present application, the target resource is matched from the resource knowledge graph based on the resource entity and the resource entity vector, including: matching a first target resource from the resource knowledge graph; the resource sample entity corresponding to the first target resource is consistent with the resource entity; if the first target resource is not matched, matching a second target resource from the resource knowledge graph based on the correlation between the different resource sample entities and the resource entity vector; and using the first target resource and / or the second target resource as the target resource.

[0047] In this embodiment, the resource knowledge graph contains resource sample entities corresponding to several resource samples, and the resource samples are used to construct the resource knowledge graph. When acquiring the target resource, first search the resource knowledge graph for an entity with the same name as the resource entity corresponding to the user's intended resource information, such as "dry pot bullfrog", or an entity with the same raw materials, and use the dish corresponding to the entity as the first target resource. If the first target resource is not found, then based on the correlation between different resource samples in the resource knowledge graph, calculate an entity vector that is close to the resource entity vector, that is, an entity vector similar to the raw materials, cuisine, etc. of the resource entity corresponding to a box of dish information of the target user, and use the dish corresponding to the entity vector as the second target resource. The first target resource and / or the second target resource are used as the target resources that are ultimately pushed to the target user.

[0048] S15: Generate resource recommendation information based on the target resource, and push the resource recommendation information to the target user.

[0049] In step S15, the target resource can be the resource name of the user's intended resource, or it can be the relevant attributes of the resource, etc. For example, for the user's intended dish, the target dish obtained through the dish knowledge graph can be the name, raw materials, taste, etc. of the target dish.

[0050] In this embodiment, after acquiring the target resource based on the resource knowledge graph, the terminal generates corresponding resource recommendation information based on the target resource, which can be displayed on the user interface, and the target user can view it on the terminal or select the intended dish by clicking or other operations. As another implementation method, the terminal can also recommend resources with high sales or good reputation to the user for the user to choose.

[0051] The embodiment of the present application provides a resource recommendation information push method, which obtains the intended resource information of the target user through human-computer interaction, then collects the interactive information containing the above-mentioned intended resource information, and performs named entity extraction on the interactive information to obtain the resource entity corresponding to the intended resource information of the target user. Then, the resource entity is mapped to the vector space through the vector conversion model to obtain the resource entity vector. Since the pre-constructed resource knowledge graph can characterize the correlation between different resource sample entities, and the correlation between different resource sample entities can be obtained through vector calculation, the target resource of interest to the target user can be matched from the resource knowledge graph based on the resource entity and resource entity vector corresponding to the intended resource information of the target user. After that, the target resource is generated according to the target resource, and the target resource is pushed to the target user to meet the needs of each user and improve the accuracy of resource recommendation.

[0052] See also Figure 2 , Figure 2 FIG. 1 is a schematic diagram showing the implementation flow of a method for pushing resource recommendation information in another embodiment of the present application. Figure 1 The implementation flow diagram corresponding to the embodiment shown in the figure, in this embodiment, before step S13, also includes S21-S22, the specific steps are as follows:

[0053] S21: Acquire training corpora for different application scenarios; the training corpora include different resource data.

[0054] In step S21, the training is expected to include at least one resource data. The resource data includes the name of the resource and other attribute information related to the resource. For example, the collected resource data is resource data, which includes but is not limited to entities corresponding to information such as the cooking method, raw materials, taste, cuisine, and taboos of dishes. Different application scenarios can be public data, related websites, and so on.

[0055] In this embodiment, data samples need to be collected in advance to train the vector model, and a large amount of resource data can be captured from the Internet, such as the OSCAR data set, related websites, etc. Then named entity extraction is performed on these resource data to obtain different resource data, which are used as training data to pre-train the vector model.

[0056] S22: Based on the training corpus, the initial vector conversion model is trained until convergence to obtain a vector conversion model.

[0057] In this embodiment, the initial vector model is trained and tested based on the training corpus until the model converges when the corresponding vector is stably output, and the training is completed to obtain a trained vector conversion model.

[0058] This embodiment collects a large amount of resource data in different application scenarios as training samples under the vector conversion model to improve the accuracy of model training.

[0059] See also Figure 3 , Figure 3 FIG. 1 is a schematic diagram showing the implementation flow of a method for pushing resource recommendation information in another embodiment of the present application. Figure 1 The implementation flow diagram corresponding to the embodiment shown in the figure, in this embodiment, before step S14, also includes S31-S35, and the specific steps are as follows:

[0060] S31: Collect resource sample sets for different application scenarios.

[0061] In this embodiment, a large amount of resource data can be purchased through public data sets, relevant websites, purchase channels, etc. as resource samples to enrich the content of the resource knowledge graph as much as possible. It should be noted that different types of resources correspond to different resource knowledge graphs, such as the dish knowledge graph corresponding to the dish resource and the music knowledge graph corresponding to the music resource.

[0062] S32: performing named entity extraction on the resource sample set to obtain a resource sample entity set.

[0063] In this embodiment, the resource sample entity set includes at least one resource sample entity. The resource sample set includes several resource samples. For example, taking dishes as an example, the resource sample can be the name of the dish, the raw materials of the dish, the cuisine of the dish, the taste of the dish, the suitable population of the dish, etc. By performing named entity recognition and named entity extraction on each resource data sample in the resource sample set, the resource sample entity corresponding to each resource data sample is obtained, and these resource sample entities form a resource sample entity set.

[0064] S33: Perform vector conversion on each resource sample entity to obtain a resource sample vector set.

[0065] In this embodiment, each resource sample entity can be vector converted through an entity vector conversion model, such as a neural network model BERT word2vec, Fasttext, GloVe or other models, and the resource sample entity can be converted into a resource sample vector or matrix to obtain resource sample entity features.

[0066] S34: Calculate the distance values ​​between different resource sample vectors in the resource sample vector set.

[0067] In this embodiment, the distance values ​​between resource sample vectors corresponding to different resource sample entities can be calculated by common distance calculation methods such as Cosine, Jaccard, and Pearson. The smaller the distance value, the stronger the correlation between the resource sample entities involved in the calculation. The correlation distribution of resource sample entities in the resource sample entity set is finally obtained through distance calculation.

[0068] S35: Construct a resource knowledge graph based on the resource sample entities and the distance values.

[0069] In this embodiment, since the distance value between different resource sample vectors can reflect the correlation between different resource sample entities, the correlation distribution of resource sample entities in the resource sample entity set is finally obtained through distance calculation. By combining each resource sample entity, a resource knowledge graph that can characterize the correlation between resource entities can be constructed.

[0070] This embodiment collects massive amounts of resource data and constructs a resource knowledge graph through vector transformation and vector correlation calculation. The constructed resource knowledge graph is not only rich in resource data but also able to characterize the correlation between different resource sample entities, effectively improving the accuracy of recommending dishes to users, enhancing user experience, and attracting customers to the store.

[0071] It should be understood that the size of the serial numbers of the steps in the above embodiments does 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.

[0072] In one embodiment, a resource recommendation information push device 400 is provided, and the resource recommendation information push device corresponds to the resource recommendation information push method in the above embodiment. Figure 4 As shown, the resource recommendation information push device includes a collection module 401, an extraction module 402, a conversion module 403, a matching module 404 and a recommendation module 405. The detailed description of each functional module is as follows:

[0073] Collection module 401: used to collect interaction information between the terminal and the target user; the interaction information includes the user's intended resource information;

[0074] Extraction module 402: used to extract named entities from the intended resource information to obtain resource entities;

[0075] Conversion module 403: used for inputting the resource entity into a vector conversion model to obtain a resource entity vector;

[0076] Matching module 404: used to match the target resource from the resource knowledge graph based on the resource entity and the resource entity vector; wherein the resource knowledge graph is used to characterize the correlation between different resource sample entities;

[0077] Recommendation module 405: used to generate resource recommendation information according to the target resource, and push the resource recommendation information to the target user.

[0078] The extraction module includes:

[0079] A conversion submodule: used for converting the interactive audio information into interactive text information if the interactive information is interactive audio information; the interactive text information includes intended resource text information;

[0080] Extraction submodule: used to extract named entities from the intended resource text information to obtain resource entities.

[0081] For the specific definition of the resource recommendation information push device, please refer to the definition of the resource recommendation information push method above, which will not be repeated here. Each module in the above-mentioned resource recommendation information push device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0082] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 5 As shown. The computer device includes a processor, a memory, a network interface, a display device (display screen) and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a readable storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer-readable instructions. The internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. The network interface of the computer device is used to communicate with an external server through a network connection. When the computer-readable instructions are executed by the processor, a method for pushing resource recommendation information is implemented. The readable storage medium provided in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0083] In one embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, wherein the processor executes the computer-readable instructions to implement the following steps:

[0084] Collecting interaction information between the terminal and the target user; the interaction information includes the target user's intended resource information;

[0085] Performing named entity extraction on the intended resource information to obtain a resource entity;

[0086] Inputting the resource entity into a vector conversion model to obtain a resource entity vector;

[0087] Based on the resource entity and the resource entity vector, the target resource is matched from the resource knowledge graph; wherein the resource knowledge graph is used to characterize the correlation between different resource sample entities;

[0088] Generate resource recommendation information based on the target resource, and push the resource recommendation information to the target user.

[0089] In one embodiment, one or more computer-readable storage media storing computer-readable instructions are provided. The readable storage media provided in this embodiment include non-volatile readable storage media and volatile readable storage media. The readable storage media store computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the following steps are implemented:

[0090] Collecting interaction information between the terminal and the target user; the interaction information includes the target user's intended resource information;

[0091] Performing named entity extraction on the intended resource information to obtain a resource entity;

[0092] Inputting the resource entity into a vector conversion model to obtain a resource entity vector;

[0093] Based on the resource entity and the resource entity vector, the target resource is matched from the resource knowledge graph; wherein the resource knowledge graph is used to characterize the correlation between different resource sample entities;

[0094] Generate resource recommendation information based on the target resource, and push the resource recommendation information to the target user.

[0095] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through computer-readable instructions, and the computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When the computer-readable instructions are executed, they may include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0096] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0097] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A resource recommendation information push method, characterized in that: The resource recommendation information pushing method comprises: Collecting interaction information between the terminal and the target user; the interaction information includes the target user's intended resource information; Performing named entity extraction on the intended resource information to obtain resource entities; the named entity extraction includes named entity recognition and named entity extraction; Input the resource entity into a vector conversion model to obtain a resource entity vector; the vector conversion model is used to map the resource entity to a high-dimensional vector space and convert it into the resource entity vector; Based on the resource entity and the resource entity vector, the target resource is matched from the resource knowledge graph; wherein the resource knowledge graph is used to characterize the correlation between different resource sample entities; Generate resource recommendation information based on the target resource, and push the resource recommendation information to the target user; Before the step of matching the target resource from the resource knowledge graph based on the resource entity and the resource entity vector, the method further includes: Collect resource sample sets for different application scenarios; Performing named entity extraction on the resource sample set to obtain a resource sample entity set; the resource sample entity set includes at least one resource sample entity; Performing vector conversion on each of the resource sample entities to obtain a resource sample vector set; Calculating distance values ​​between different resource sample vectors in the resource sample vector set; Constructing a resource knowledge graph based on the resource sample entity and the distance value; The matching of the target resource from the resource knowledge graph based on the resource entity and the resource entity vector includes: A first target resource is matched from the resource knowledge graph; the resource sample entity corresponding to the first target resource is consistent with the resource entity; If the first target resource is not matched, matching a second target resource from the resource knowledge graph according to the correlation between the different resource sample entities and the resource entity vector; The first target resource and / or the second target resource is used as the target resource.

2. The resource recommendation information push method according to claim 1, characterized in that: The performing named entity extraction on the intended resource information to obtain a resource entity includes: If the interactive information is interactive audio information, converting the interactive audio information into interactive text information; the interactive text information includes intended resource text information; Perform named entity extraction on the intended resource text information to obtain the resource entity.

3. The resource recommendation information push method according to claim 1, characterized in that: Before the step of inputting the resource entity into a vector conversion model to obtain a resource entity vector, the step further includes: Acquire training corpora for different application scenarios; the training corpora include at least one resource data; Based on the training corpus, the initial vector conversion model is trained until convergence to obtain a vector conversion model.

4. The resource recommendation information push method according to claim 1, characterized in that: The step of inputting the resource entity into a vector conversion model to obtain a resource entity vector comprises: Performing word segmentation processing on the resource entity to obtain a resource entity after word segmentation; The resource entity after word segmentation is input into the vector conversion model to obtain a resource entity vector.

5. A resource recommendation information push device, characterized in that: The resource recommendation information pushing device comprises: Collection module: used to collect interaction information between the terminal and the target user; the interaction information includes the target user's intended resource information; Extraction module: used to perform named entity extraction on the intended resource information to obtain resource entities; the named entity extraction includes named entity recognition and named entity extraction; Conversion module: used for inputting the resource entity into a vector conversion model to obtain a resource entity vector; the vector conversion model is used for mapping the resource entity into a high-dimensional vector space and converting it into the resource entity vector; Matching module: used to match the target resource from the resource knowledge graph based on the resource entity and the resource entity vector; wherein the resource knowledge graph is used to characterize the correlation between different resource sample entities; Recommendation module: used to generate resource recommendation information based on the target resource, and push the resource recommendation information to the target user; Before the step of matching the target resource from the resource knowledge graph based on the resource entity and the resource entity vector, the method further includes: Collect resource sample sets for different application scenarios; Performing named entity extraction on the resource sample set to obtain a resource sample entity set; the resource sample entity set includes at least one resource sample entity; Performing vector conversion on each of the resource sample entities to obtain a resource sample vector set; Calculating distance values ​​between different resource sample vectors in the resource sample vector set; Constructing a resource knowledge graph based on the resource sample entity and the distance value; The matching of the target resource from the resource knowledge graph based on the resource entity and the resource entity vector includes: A first target resource is matched from the resource knowledge graph; the resource sample entity corresponding to the first target resource is consistent with the resource entity; If the first target resource is not matched, matching a second target resource from the resource knowledge graph according to the correlation between the different resource sample entities and the resource entity vector; The first target resource and / or the second target resource is used as the target resource.

6. The resource recommendation information push device according to claim 5, characterized in that: The extraction module comprises: A conversion submodule: used for converting the interactive audio information into interactive text information if the interactive information is interactive audio information; the interactive text information includes intended resource text information; Extraction submodule: used to extract named entities from the intended resource text information to obtain resource entities.

7. A computer device comprising a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor, characterized in that: When the computer-readable instructions are executed by a processor, the resource recommendation information pushing method according to any one of claims 1 to 4 is implemented.

8. One or more readable storage media, wherein the readable storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by a processor, the resource recommendation information push method according to any one of claims 1 to 4 is implemented.

Images