Content recommendation method and device, equipment, medium and product

Abstract

The invention provides a content recommendation method and device, equipment, a medium and a product, and the method comprises the steps: obtaining a demand text which indicates the demand information of a first user for a target entity; the demand text is recognized, a query statement is determined, and the query statement is used for querying information related to the target entity; querying from the knowledge base by using the query statement to obtain at least one piece of reference information related to the target entity; and according to the at least one piece of reference information, generating recommended reply content for the demand text. In the method, for the question of the first user, the reference information capable of being used for replying can be accurately screened out from a large amount of information related to the target entity, and the content used for question replying is automatically recommended based on the reference information, so that targeted replying is realized, and the efficiency of communication with the first user for the target entity is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a content recommendation method, device, electronic device, computer-readable storage medium, and computer program product. Background Art

[0002] In product sales, salespeople can identify potential customers and communicate with them in order to market products to them.

[0003] During the communication process between sales staff and potential customers, potential customers may have a lot of questions about the product, and sales staff need to answer the potential customers' questions. However, since product information is usually rich and complex, it is difficult for sales staff to answer them quickly and accurately, and the communication efficiency is low. Summary of the invention

[0004] The present application provides a content recommendation method. The method can automatically recommend content for answering questions and achieve targeted answers. The present application also provides a device, an electronic device, a computer-readable storage medium, and a computer program product corresponding to the above method.

[0005] In a first aspect, the present application provides a content recommendation method, the method comprising:

[0006] Acquire a demand text, where the demand text indicates demand information of the first user for the target entity;

[0007] Identify the requirement text and determine a query statement, where the query statement is used to query information related to the target entity;

[0008] Using the query statement to query from a knowledge base to obtain at least one reference information related to the target entity;

[0009] According to the at least one piece of reference information, a recommended reply content for the demand text is generated.

[0010] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and obtaining the demand text includes:

[0011] Acquire voice information of the first user and the second user;

[0012] The voice information is converted into text to obtain the required text.

[0013] In some possible implementations, the acquiring voice information of the first user and the second user includes:

[0014] Acquire first voice information of the first user and the second user in a current collection period;

[0015] The converting the voice information into text to obtain the required text includes:

[0016] Converting the first voice information into text to obtain a first required text;

[0017] A requirement text is determined according to the first requirement text and a second requirement text obtained in at least one historical collection cycle.

[0018] In some possible implementations, the identifying the requirement text and determining the query statement includes:

[0019] At least based on the requirement text, generate first prompt information; wherein the first prompt information is used to instruct the language model to generate a query statement;

[0020] The first prompt information is sent to a first language model, and a query statement returned by the first language model is received.

[0021] In some possible implementations, the method further includes:

[0022] At least based on the requirement text, generate second prompt information; wherein the second prompt information is used to instruct the language model to perform requirement type recognition;

[0023] Sending the second prompt information to a second language model, and receiving the requirement type returned by the second language model;

[0024] The querying the knowledge base using the query statement to obtain at least one reference information related to the target entity includes:

[0025] The query statement is used to query a knowledge base that matches the requirement type to obtain at least one reference information related to the target entity.

[0026] In some possible implementations, generating, according to the at least one piece of reference information, recommended reply content for the requirement text includes:

[0027] generating third prompt information based at least on the at least one reference information; wherein the third prompt information is used to instruct the language model to perform content recommendation;

[0028] The third prompt information is sent to a third language model, and recommended reply content for the requirement text returned by the third language model is received.

[0029] In some possible implementations, generating the third prompt information at least based on the at least one piece of reference information includes:

[0030] Sorting the at least one piece of reference information according to the similarity between the at least one piece of reference information and the query statement to obtain a sorting result of the at least one piece of reference information;

[0031] The third prompt information is generated based at least on the at least one piece of reference information and a ranking result of the at least one piece of reference information.

[0032] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and after determining the query statement, the method further includes:

[0033] presenting the query statement to the second user;

[0034] A trigger operation of the second user for the query statement is received; wherein the trigger operation indicates whether the query statement matches the requirement text.

[0035] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and the method further includes:

[0036] The recommended reply content for the demand text is presented to the second user in a streaming output manner.

[0037] In a second aspect, the present application provides a content recommendation device, the device comprising:

[0038] An acquisition module, configured to acquire a demand text, wherein the demand text indicates demand information of the first user for the target entity;

[0039] A determination module, used to identify the demand text and determine a query statement, wherein the query statement is used to query information related to the target entity;

[0040] A query module, used to query the knowledge base using the query statement to obtain at least one reference information related to the target entity;

[0041] A generating module is used to generate recommended reply content for the requirement text according to the at least one reference information.

[0042] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and the acquisition module is specifically configured to:

[0043] Acquire voice information of the first user and the second user;

[0044] The voice information is converted into text to obtain the required text.

[0045] In some possible implementations, the acquisition module is specifically used to:

[0046] Acquire first voice information of the first user and the second user in a current collection period;

[0047] Converting the first voice information into text to obtain a first required text;

[0048] A requirement text is determined according to the first requirement text and a second requirement text obtained in at least one historical collection cycle.

[0049] In some possible implementations, the determining module is specifically configured to:

[0050] At least based on the requirement text, generate first prompt information; wherein the first prompt information is used to instruct the language model to generate a query statement;

[0051] The first prompt information is sent to a first language model, and a query statement returned by the first language model is received.

[0052] In some possible implementations, the determining module is further configured to:

[0053] At least based on the requirement text, generate second prompt information; wherein the second prompt information is used to instruct the language model to perform requirement type recognition;

[0054] Sending the second prompt information to a second language model, and receiving the requirement type returned by the second language model;

[0055] The query module is specifically used for:

[0056] The query statement is used to query a knowledge base that matches the requirement type to obtain at least one reference information related to the target entity.

[0057] In some possible implementations, the generating module is specifically used to:

[0058] generating third prompt information based at least on the at least one reference information; wherein the third prompt information is used to instruct the language model to perform content recommendation;

[0059] The third prompt information is sent to a third language model, and recommended reply content for the requirement text returned by the third language model is received.

[0060] In some possible implementations, the generating module is specifically used to:

[0061] Sorting the at least one piece of reference information according to the similarity between the at least one piece of reference information and the query statement to obtain a sorting result of the at least one piece of reference information;

[0062] The third prompt information is generated based at least on the at least one piece of reference information and a ranking result of the at least one piece of reference information.

[0063] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and the device further includes an interaction module, wherein the interaction module is configured to:

[0064] presenting the query statement to the second user;

[0065] A trigger operation of the second user for the query statement is received; wherein the trigger operation indicates whether the query statement matches the requirement text.

[0066] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and the interaction module is further configured to:

[0067] The recommended reply content for the demand text is presented to the second user in a streaming output manner.

[0068] In a third aspect, the present application provides an electronic device, the electronic device comprising a processor and a memory. The processor and the memory communicate with each other. The processor is used to execute instructions stored in the memory, so that the electronic device performs the content recommendation method in the first aspect or any implementation of the first aspect.

[0069] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, wherein the instructions instruct an electronic device to execute the content recommendation method described in the first aspect or any one of the implementations of the first aspect.

[0070] In a fifth aspect, the present application provides a computer program product comprising instructions, which, when executed on an electronic device, enables the electronic device to execute the content recommendation method described in the first aspect or any one of the implementations of the first aspect.

[0071] Based on the implementations provided in the above aspects, this application can also be further combined to provide more implementations.

[0072] It can be seen from the above technical solutions that this application has the following advantages:

[0073] The present application provides a content recommendation method, which first obtains a demand text, wherein the demand text indicates a first user's demand information for a target entity, then identifies the demand text, determines a query statement, and uses the query statement to query information related to the target entity, uses the query statement to query from a knowledge base, obtains at least one reference information related to the target entity, and generates recommended reply content for the demand text based on the at least one reference information.

[0074] In this method, the demand text is identified to determine the demand of the first user for the target entity, a query statement matching the demand is generated, reference information is screened out in the knowledge base, and recommended reply content is generated in combination with the reference information. In this way, for the first user's question, reference information that can be used for reply can be accurately screened out from a large amount of information related to the target entity, and content for replying to the question can be automatically recommended based on the reference information, so as to achieve targeted replies and improve the efficiency of communication with the first user for the target entity. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the technical method of the embodiments of the present application, the drawings required for use in the embodiments are briefly introduced below.

[0076] Figure 1 A flowchart of a content recommendation method provided in an embodiment of the present application;

[0077] Figure 2 A page diagram of a communication assistant module provided in an embodiment of the present application;

[0078] Figure 3 A schematic diagram of the structure of a content recommendation device provided in an embodiment of the present application;

[0079] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0080] The terms "first" and "second" in the embodiments of the present application are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.

[0081] First, some technical terms and application scenarios involved in the embodiments of this application are introduced.

[0082] Product sales can be understood as the process of sales personnel and customers reaching a cooperation agreement on a certain product. Specifically, product sales can include multiple project stages, such as the business opportunity confirmation stage, solution communication stage, business negotiation stage and customer signing stage.

[0083] In the business opportunity confirmation stage, sales staff can conduct preliminary communication with customers to confirm that customers are willing to learn more about the product and have potential intentions. In the solution communication stage, sales staff can demonstrate the solution to key customers and introduce the product and related information in detail. In the business negotiation stage, sales staff can negotiate with customers to confirm specific information such as the purchase quantity and price of the product. In the customer signing stage, sales staff can sign a contract with customers and reach a cooperation agreement.

[0084] During the business opportunity confirmation stage, solution communication stage and business negotiation stage, sales staff and potential customers often communicate a lot. During the communication process between sales staff and potential customers, sales staff can introduce relevant information about the product to potential customers, such as product functions, product features, product operation methods, etc. Potential customers can ask sales staff questions about problems with the product, and sales staff need to answer the potential customers' questions.

[0085] Since sales staff and potential customers often communicate through telephone, video conferencing and other means, sales staff usually do not have sufficient time to search for information in response to potential customers' questions and need to quickly respond to potential customers' questions based on their own understanding of the product.

[0086] However, since product information is usually rich and complex, it is difficult for sales staff to give targeted answers quickly and accurately, and communication efficiency is low.

[0087] In view of this, the present application provides a content recommendation method. The method first obtains a demand text, wherein the demand text indicates the demand information of a first user for a target entity, then identifies the demand text, determines a query statement, and the query statement is used to query information related to the target entity, uses the query statement to query from a knowledge base, obtains at least one reference information related to the target entity, and generates recommended reply content for the demand text based on the at least one reference information.

[0088] In this method, the demand text is identified to determine the demand of the first user for the target entity, a query statement matching the demand is generated, reference information is screened out in the knowledge base, and recommended reply content is generated in combination with the reference information. In this way, for the first user's question, reference information that can be used for reply can be accurately screened out from a large amount of information related to the target entity, and content for replying to the question can be automatically recommended based on the reference information, so as to achieve targeted replies and improve the efficiency of communication with the first user for the target entity.

[0089] To facilitate understanding of the technical solution provided by the embodiments of the present application, it will be described below with reference to the accompanying drawings.

[0090] See also Figure 1 The flowchart of a content recommendation method provided by an embodiment of the present application is shown, and the method specifically includes:

[0091] S101: Obtain the requirement text.

[0092] In the embodiment of the present application, the demand text may indicate the demand information of the first user for the target entity. The first user may be understood as an object that has questions about the target entity and asks questions about the target entity. In other words, the demand text may be question information generated by the first user and related to the target entity.

[0093] In the product sales scenario, the first user can be an enterprise or organization waiting for cooperation (i.e., a potential customer), and the target entity can be a product or a service waiting for sale. If cooperation is reached with the first user, the first user can be associated with the target entity, that is, the first user can purchase the target entity. The demand text can be any question related to the content to be sold, such as "What is the difference between this product and other companies' similar products?", "I still don't quite understand this function", etc.

[0094] Since the demand text needs to be identified later, in the embodiment of the present application, the demand information of the first user for the target entity is presented in the form of text. In some possible implementations, the first user can have a conversation with the second user for the target entity, at which time, voice information of the first user and the second user is obtained, and the voice information is converted into text to obtain the demand text.

[0095] The second user may be understood as a user who has a voice conversation with the first user. In a product sales scenario, the second user may be a salesperson, and the second user and the first user may have a voice conversation with the target entity by telephone, online conference, etc., which is not limited in this embodiment of the present application.

[0096] Text conversion generally refers to the process of converting information in audio format into information in text format. In some possible implementations, text conversion can be achieved using automatic speech recognition (ASR) technology. In ASR technology, by pre-processing speech information such as noise reduction and echo removal, speech features are extracted from the pre-processed speech information, recognized using an acoustic model, and the recognition results are decoded to obtain the required text.

[0097] That is to say, during a conversation between a first user and a second user regarding a target entity, such as when the second user introduces the target entity to the first user and the first user asks questions regarding the second user's introduction, voice information is collected and converted into text, so as to obtain a demand text indicating the first user's demand information regarding the target entity without affecting the conversation between the first user and the second user.

[0098] In some embodiments, during the conversation between the first user and the second user regarding the target entity, voice information may be acquired at a certain acquisition frequency, for example, the acquisition frequency may be 3 seconds, that is, during the conversation between the first user and the second user regarding the target entity, voice information is acquired once every 3 seconds. In this case, the demand text may include demand texts under multiple acquisition cycles, and the multiple acquisition cycles belong to the same conversation process between the first user and the second user.

[0099] In specific implementation, the first voice information of the first user and the second user in the current collection cycle is obtained, the first voice information is converted into text to obtain the first demand text, and the demand text is determined according to the first demand text and the second demand text obtained in at least one historical collection cycle.

[0100] That is to say, the demand text in each collection cycle includes not only the demand text converted from the voice information collected in the collection cycle, but also the demand text converted from the voice information collected in at least one historical collection cycle before the collection cycle.

[0101] Taking at least one historical collection cycle of 10 collection cycles as an example, for the process of the first user and the second user having a conversation with the target entity, the current collection cycle is the Tth collection cycle, and T is a natural number greater than 0. In the Tth collection cycle, the first voice information is collected, and the first voice information is converted into text to obtain the first demand text. Next, the second demand text obtained in the T-10th collection cycle to the T-1st collection cycle is obtained, that is, the demand text is obtained after the voice information collected from the T-10th collection cycle to the T-1st collection cycle is converted into text, and the demand text obtained from the T-10th collection cycle to the T-1st collection cycle is used as the demand text under the Tth collection cycle.

[0102] In this way, in information-rich scenarios such as a conversation between a first user and a second user regarding a target entity, voice information from multiple collection cycles is considered so that the demand text contains richer demand information, facilitating more accurate recognition of the demand text later.

[0103] S102: Identify the requirement text and determine the query statement.

[0104] Among them, the query statement (query) can be used to query information related to the target entity. That is, by identifying the demand text, the first user's demand information (i.e., question) for the target entity is extracted from the demand text, and then combined with the first user's demand information for the target entity, a query statement that can be used to query information is determined, so as to query reference information to respond to the first user's demand information for the target entity.

[0105] In some possible implementations, a language model can be used to identify the demand text and determine the query statement. The language model has natural language processing capabilities, can identify, analyze, and generate natural language content, and can handle different types of natural language tasks. For example, the language model can be a deep learning model trained using text data.

[0106] In specific implementation, first prompt information is generated based on at least the requirement text, the first prompt information is sent to the first language model, and a query statement returned by the first language model is received. The first prompt information can be used to instruct the language model to generate a query statement.

[0107] That is to say, in the embodiment of the present application, the chart analysis results are generated based on the prompt learning method. Among them, the prompt information can be used to guide the language model to perform specific output in the generative task (such as text generation task, question-answering task, dialogue task), and by generating prompt information, the language model is helped to understand the background and requirements of the task, so that the language model can handle different types of natural language processing tasks without retraining the language model, thereby increasing the scalability and flexibility of the language model.

[0108] In the embodiment of the present application, the first prompt information may include the demand text and the instruction to generate the query statement. Through the prompt capability of the first prompt information, the first language model is instructed to recognize the demand text and generate a query statement for querying information related to the target entity and satisfying the first user's demand information for the target entity. In this way, the query statement is generated quickly and automatically without the need for the second user to manually recognize the demand text.

[0109] Furthermore, in some embodiments, the language model can be used to identify the requirement type of the requirement text. In specific implementation, the second prompt information is generated based on at least the requirement text, the second prompt information is sent to the second language model, and the requirement type returned by the second language model is received.

[0110] The second prompt information can be used to instruct the language model to identify the demand type. The demand type can be understood as the category to which the demand information of the first user for the target entity belongs. For example, in a product sales scenario, the demand type may include demands related to product functions, demands related to solutions, demands related to prices, etc.

[0111] In the embodiment of the present application, the second prompt information may include the demand text and the instruction to determine the demand type. Through the prompt capability of the second prompt information, the second language model is instructed to identify the demand type of the demand text and determine the demand type indicated by the demand text. In this way, the automatic identification of the demand type is achieved, and the intention of the first user to ask questions about the target entity is accurately and quickly determined.

[0112] It should be noted that the embodiments of the present application do not limit the first language model and the second language model. In some embodiments, the first language model and the second language model may be the same language model. In other embodiments, the first language model and the second language model may be different language models.

[0113] The embodiment of the present application can also confirm with the second user whether the query statement is correct. Specifically, after determining the query statement, the query statement is presented to the second user, and a trigger operation of the second user for the query statement is received. The trigger operation indicates whether the query statement matches the requirement text.

[0114] That is, after identifying the demand text and determining the query statement, the query statement is presented to the second user, allowing the second user to determine whether the automatically generated query statement matches the demand text, that is, whether the query statement satisfies the first user's demand information for the target entity, and after receiving the second user's trigger operation on the query statement, the subsequent processing steps are performed. In this way, by interacting with the second user, the problem of low content recommendation accuracy caused by the mismatch between the query statement and the demand text is avoided, and at the same time, computing resources are saved.

[0115] S103: Query the knowledge base using a query statement to obtain at least one piece of reference information related to the target entity.

[0116] The knowledge base can be understood as a system for storing, organizing and providing information resources. In the embodiment of the present application, the knowledge base can store a large amount of information related to the entity. In the product sales scenario, the knowledge base can store information related to each product, such as product introductions, user manuals, demonstration videos, etc. In addition, the knowledge base can also store information related to product sales cases, such as historical sales cases.

[0117] Reference information can be understood as information related to the target entity and can be used as a reference when replying to content recommendations. In some embodiments, the reference information can be presented in the form of a document.

[0118] In the embodiment of the present application, after determining the query statement, the query statement is used to search in the knowledge base to obtain at least one reference information related to the target entity. For example, the query statement is used to search in the knowledge base and the top-K query results are used as reference information.

[0119] In some possible implementations, in addition to identifying the requirement text and determining the query statement, the requirement text is also identified as the requirement type to determine the requirement type. In this case, the query statement is used to query the knowledge base matching the requirement type to obtain at least one reference information related to the target entity.

[0120] In other words, when identifying the requirement type for the requirement text, the knowledge base can be first filtered based on the requirement type, and then the search can be performed in the knowledge base that matches the requirement type. For example, when the requirement type is a requirement related to product functions, the knowledge base that matches the requirement type can be a knowledge base that stores the function introductions of various products.

[0121] In this way, the scope of the knowledge base can be narrowed by utilizing the demand type, thereby improving the accuracy of queries and, in turn, the accuracy of subsequent content recommendations.

[0122] S104: Generate recommended reply content for the demand text according to at least one piece of reference information.

[0123] The recommended reply content can be understood as content used to assist in replying to the demand text. In the product sales scenario, the second user can use the recommended reply content to reply to the first user's demand information for the target entity, thereby improving the reply speed and accuracy, and at the same time improving the richness of the reply content.

[0124] In some possible implementations, a language model may be used to generate recommended reply content. In a specific implementation, based on at least one reference information, a third prompt information is generated, the third prompt information is sent to a third language model, and the recommended reply content for the required text returned by the third language model is received. The third prompt information may be used to instruct the language model to recommend content.

[0125] In the embodiment of the present application, the third prompt information may include at least one reference information and an instruction for generating recommended reply content. Through the prompt capability of the third prompt information, the third language model is instructed to analyze and integrate at least one reference information to generate recommended reply content for the required text. In this way, the reference information is combined to realize the automatic generation of recommended reply content, and the recommended reply content is quickly provided without manual query, thereby improving communication efficiency.

[0126] It should be noted that the embodiments of the present application do not limit the third language model. In some embodiments, the first language model, the second language model, and the third language model may be the same language model. In other embodiments, the first language model, the second language model, and the third language model may all be different language models. In still other embodiments, the third language model may be the same language model as the first language model and a different language model from the second language model, or the third language model may be the same language model as the second language model and a different language model from the first language model.

[0127] In some embodiments, the reference information may be sorted first, and then the recommended reply content may be generated. Specifically, according to the similarity between the at least one reference information and the query statement, the at least one reference information is sorted to obtain the sorting result of the at least one reference information, and then, based on at least the at least one reference information and the sorting result of the at least one reference information, a third prompt information is generated, and then the third prompt information is sent to the third language model, and the recommended reply content for the demand text returned by the third language model is received.

[0128] That is, after obtaining reference information related to the target entity, the reference information is sorted, for example, the similarity between the reference information and the query statement is determined, and the reference information with high similarity has a higher sorting result, and the reference information with low similarity has a lower sorting result. In this way, when the sorted reference information is analyzed and integrated using the language model, the language model can use the sorting results to learn the similarity between the reference information and the query statement, and then combine the sorting results to generate recommended reply content for the demand text, enrich the input of the language model, and improve the accuracy of the language model output.

[0129] After generating the recommended reply content for the demand text, the recommended reply content for the demand text can also be presented to the second user in a streaming output manner. For example, by repeatedly calling the streaming iterator, the content in the streaming iterator is presented to the second user in real time, achieving the effect of "streaming output", avoiding the second user from waiting too long, and presenting the generated recommended reply content in time.

[0130] In this method, the demand text is identified to determine the demand of the first user for the target entity, a query statement matching the demand is generated, reference information is screened out in the knowledge base, and recommended reply content is generated in combination with the reference information. In this way, for the first user's question, reference information that can be used for reply can be accurately screened out from a large amount of information related to the target entity, and content for replying to the question can be automatically recommended based on the reference information, so as to achieve targeted replies and improve the efficiency of communication with the first user for the target entity.

[0131] The above article introduces the content recommendation method provided by this application, and the following will explain it in conjunction with specific scenarios.

[0132] The content recommendation method provided in this application can be applied in an office automation (OA) system. Among them, the OA system, which can also be called a business platform or a business system, can be used for office assistance in an enterprise. The OA system usually includes multiple system modules that provide different functions, such as a calendar module, an instant messaging module, a form module, a task management module, a meeting module, etc.

[0133] The OA system may provide a communication assistant module, which may assist the second user (i.e., salesperson) in communicating with the first user (i.e., potential customer). The communication assistant module is usually deployed with a voice collection component, which may collect voice information during the conversation between the first user and the second user.

[0134] See also Figure 2A page diagram of a communication assistant module is shown, in which a page 20 of the communication assistant module includes an information display area 21, in which relevant information of the first user is displayed, for example, information related to the first user recorded at various stages of product sales. In a conversation between a second user and the first user, the second user can view relevant information of the first user in the information display area 21, and obtain historical communication information with the first user, so as to conduct this conversation more targetedly.

[0135] In addition, conversation control controls, such as an answer control and a hang-up control, may also be provided in the information display area 21. The second user may answer the conversation with the first user by triggering the answer control, or may hang up the conversation with the first user by triggering the hang-up control.

[0136] In an embodiment of the present application, page 20 of the communication assistant module also includes an artificial intelligence (AI) assistant area 22. The AI ​​assistant area 22 can display recommended reply content to the second user during the conversation between the second user and the first user.

[0137] To illustrate with a specific example, during the conversation between the first user and the second user, the voice collection component deployed by the communication assistant module can collect voice information and convert the voice information into text to obtain a demand text indicating the first user's demand information for the target entity. Then, the demand text is identified, the query statement is determined, and the query statement 221 is presented to the second user in the AI ​​assistant area 22, for example, "Introduction to the research and development solution of new catering products of xx". The second user views the query statement 221 in the AI ​​assistant area 22 to determine whether the query statement matches the demand information of the first user for the target entity. If so, the query statement 221 displayed in the AI ​​assistant area 22 is clicked. The query statement is used to query from the knowledge base to obtain reference information related to the target entity, and further generate recommended reply content for the demand text, and the recommended reply content 222 for the demand text is presented to the second user in the AI ​​assistant area 22 in a streaming output manner.

[0138] During the conversation between the first user and the second user regarding the target entity, the recommended reply content is presented to the second user in real time based on the content of the conversation. The second user can obtain a large amount of reference information related to the first user's demand information (such as questions) without manual search, and can view it intuitively. In this way, the second user can respond to the first user's questions accurately, quickly and comprehensively based on the recommended reply content, thereby improving the communication efficiency and quality between the first user and the second user.

[0139] Combination of the above Figure 1 and Figure 2The content recommendation method provided in the embodiment of the present application is introduced in detail. The following will introduce the device and equipment provided in the embodiment of the present application in conjunction with the accompanying drawings.

[0140] See also Figure 3 The structure diagram of the content recommendation device shown in FIG. 30 includes:

[0141] An acquisition module 301 is used to acquire a demand text, where the demand text indicates demand information of a first user for a target entity;

[0142] A determination module 302 is used to identify the demand text and determine a query statement, where the query statement is used to query information related to the target entity;

[0143] A query module 303 is used to query the knowledge base using the query statement to obtain at least one reference information related to the target entity;

[0144] The generating module 304 is configured to generate recommended reply content for the requirement text according to the at least one piece of reference information.

[0145] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and the acquisition module 301 is specifically configured to:

[0146] Acquire voice information of the first user and the second user;

[0147] The voice information is converted into text to obtain the required text.

[0148] In some possible implementations, the acquisition module 301 is specifically used to:

[0149] Acquire first voice information of the first user and the second user in a current collection period;

[0150] Converting the first voice information into text to obtain a first required text;

[0151] A requirement text is determined according to the first requirement text and a second requirement text obtained in at least one historical collection cycle.

[0152] In some possible implementations, the determining module 302 is specifically configured to:

[0153] At least based on the requirement text, generate first prompt information; wherein the first prompt information is used to instruct the language model to generate a query statement;

[0154] The first prompt information is sent to a first language model, and a query statement returned by the first language model is received.

[0155] In some possible implementations, the determining module 302 is further configured to:

[0156] At least based on the requirement text, generate second prompt information; wherein the second prompt information is used to instruct the language model to perform requirement type recognition;

[0157] Sending the second prompt information to a second language model, and receiving the requirement type returned by the second language model;

[0158] The query module 303 is specifically used for:

[0159] The query statement is used to query a knowledge base that matches the requirement type to obtain at least one reference information related to the target entity.

[0160] In some possible implementations, the generating module 304 is specifically configured to:

[0161] generating third prompt information based at least on the at least one reference information; wherein the third prompt information is used to instruct the language model to perform content recommendation;

[0162] The third prompt information is sent to a third language model, and recommended reply content for the requirement text returned by the third language model is received.

[0163] In some possible implementations, the generating module 304 is specifically configured to:

[0164] Sorting the at least one piece of reference information according to the similarity between the at least one piece of reference information and the query statement to obtain a sorting result of the at least one piece of reference information;

[0165] The third prompt information is generated based at least on the at least one piece of reference information and a ranking result of the at least one piece of reference information.

[0166] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and the device further includes an interaction module, wherein the interaction module is configured to:

[0167] presenting the query statement to the second user;

[0168] A trigger operation of the second user for the query statement is received; wherein the trigger operation indicates whether the query statement matches the requirement text.

[0169] In some possible implementations, the first user and the second user have a conversation regarding the target entity, and the interaction module is further configured to:

[0170] The recommended reply content for the demand text is presented to the second user in a streaming output manner.

[0171] The content recommendation device 30 according to the embodiment of the present application may correspond to executing the method described in the embodiment of the present application, and the above and other operations and / or functions of each module / unit of the content recommendation device 30 are respectively to implement Figure 1 For the sake of brevity, the corresponding processes of each method in the illustrated embodiment are not described in detail here.

[0172] The present application also provides an electronic device. The electronic device is specifically used to implement Figure 4 The functions of the content recommendation device 30 in the illustrated embodiment.

[0173] Figure 4 A schematic diagram of the structure of an electronic device 400 is provided. Figure 4 As shown, the electronic device 400 includes a bus 401, a processor 402, a communication interface 403 and a memory 404. The processor 402, the memory 404 and the communication interface 403 communicate with each other via the bus 401.

[0174] The bus 401 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0175] The processor 402 may be any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

[0176] The communication interface 403 is used for communicating with the outside. For example, the communication interface 403 can be used for communicating with a terminal.

[0177] The memory 404 may include a volatile memory, such as a random access memory (RAM). The memory 404 may also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD), or a solid state drive (SSD).

[0178] The memory 404 stores executable codes, and the processor 402 executes the executable codes to perform the aforementioned content recommendation method.

[0179] Specifically, in implementing Figure 3 In the case of the embodiment shown, and Figure 3 When each module or unit of the content recommendation device 30 described in the embodiment is implemented by software, Figure 3 The software or program code required for the functions of each module / unit in the content recommendation system may be partially or completely stored in the memory 404. The processor 402 executes the program code corresponding to each unit stored in the memory 404 to perform the aforementioned content recommendation method.

[0180] The embodiment of the present application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that can be stored by a computing device or a data storage device such as a data center that contains one or more available media. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state hard disk). The computer-readable storage medium includes instructions that instruct the computing device to execute the content recommendation method applied to the content recommendation device 30.

[0181] The embodiment of the present application further provides a computer program product, which includes one or more computer instructions. When the computer instructions are loaded and executed on a computing device, the process or function described in the embodiment of the present application is generated in whole or in part.

[0182] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer or data center to another website, computer or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0183] When the computer program product is executed by a computer, the computer executes any of the aforementioned content recommendation methods. The computer program product may be a software installation package, and when any of the aforementioned content recommendation methods is needed, the computer program product may be downloaded and executed on a computer.

[0184] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to each embodiment of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0185] The units involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the unit / module does not, in some cases, constitute a limitation on the unit itself.

[0186] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0187] In the context of the present application embodiment, machine-readable medium can be a tangible medium that can contain or store a program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable medium can include but is not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0188] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the system or device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0189] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0190] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0191] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0192] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A content recommendation method, characterized in that: The method comprises: Acquire a demand text, where the demand text indicates demand information of the first user for the target entity; Identify the requirement text and determine a query statement, where the query statement is used to query information related to the target entity; Using the query statement to query from a knowledge base to obtain at least one reference information related to the target entity; According to the at least one piece of reference information, a recommended reply content for the demand text is generated.

2. The method according to claim 1, characterized in that: The first user and the second user have a conversation regarding the target entity, and obtaining the demand text includes: Acquire voice information of the first user and the second user; The voice information is converted into text to obtain the required text.

3. The method according to claim 2, characterized in that The acquiring the voice information of the first user and the second user includes: Acquire first voice information of the first user and the second user in a current collection period; The converting the voice information into text to obtain the required text includes: Converting the first voice information into text to obtain a first required text; A requirement text is determined according to the first requirement text and a second requirement text obtained in at least one historical collection cycle.

4. The method according to claim 1, characterized in that: The step of identifying the requirement text and determining the query statement includes: At least based on the requirement text, generate first prompt information; wherein the first prompt information is used to instruct the language model to generate a query statement; The first prompt information is sent to a first language model, and a query statement returned by the first language model is received.

5. The method according to claim 1, characterized in that The method further comprises: At least based on the requirement text, generate second prompt information; wherein the second prompt information is used to instruct the language model to perform requirement type recognition; Sending the second prompt information to a second language model, and receiving the requirement type returned by the second language model; The querying the knowledge base using the query statement to obtain at least one reference information related to the target entity includes: The query statement is used to query a knowledge base that matches the requirement type to obtain at least one reference information related to the target entity.

6. The method according to claim 1, characterized in that The step of generating, according to the at least one piece of reference information, recommended reply content for the requirement text includes: generating third prompt information based at least on the at least one reference information; wherein the third prompt information is used to instruct the language model to perform content recommendation; The third prompt information is sent to a third language model, and recommended reply content for the requirement text returned by the third language model is received.

7. The method according to claim 6, characterized in that The generating the third prompt information at least based on the at least one piece of reference information includes: Sorting the at least one piece of reference information according to the similarity between the at least one piece of reference information and the query statement to obtain a sorting result of the at least one piece of reference information; The third prompt information is generated based at least on the at least one piece of reference information and a ranking result of the at least one piece of reference information.

8. The method according to claim 1, characterized in that The first user and the second user have a conversation with respect to the target entity. After determining the query statement, the method further includes: presenting the query statement to the second user; A trigger operation of the second user for the query statement is received; wherein the trigger operation indicates whether the query statement matches the requirement text.

9. The method according to any one of claims 1 to 8, characterized in that: The first user and the second user conduct a dialogue with respect to the target entity, and the method further comprises: The recommended reply content for the demand text is presented to the second user in a streaming output manner.

10. A content recommendation device, characterized in that: The device comprises: An acquisition module, configured to acquire a demand text, wherein the demand text indicates demand information of the first user for the target entity; A determination module, used to identify the demand text and determine a query statement, wherein the query statement is used to query information related to the target entity; A query module, used to query the knowledge base using the query statement to obtain at least one reference information related to the target entity; A generating module is used to generate recommended reply content for the requirement text according to the at least one reference information.

11. An electronic device, characterized in that: The electronic device comprises a processor and a memory; The processor is configured to execute instructions stored in the memory, so that the electronic device performs the method according to any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that: The method comprises instructions, wherein the instructions instruct an electronic device to execute the method as claimed in any one of claims 1 to 9.

13. A computer program product, characterized in that The computer program product comprises computer readable instructions for implementing the method according to any one of claims 1 to 9.

Images