Promotion verbal skill generation method and device, electronic equipment and storage medium

Abstract

The invention relates to the technical field of computers, and provides a promotion verbal skill generation method and device, electronic equipment and a storage medium. The method comprises the following steps: firstly, obtaining a consultation session of a user for a corresponding commodity; according to the consultation session, target commodity attribute information concerned by the user is determined; and finally, according to the attribute information of the target commodity, generating a promotion session corresponding to the consultation session. And the promotion session is used to recommend the recommended product to the user. The commodity attribute information concerned by the user is identified through the consultation session so as to select and recommend the recommended product to the user, and the advantages of the recommended product are illustrated to the user through the promotion verbal skill, so that the purchase willingness of the user is enhanced, and the purchase conversion rate is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular, to a method, device, electronic device and storage medium for generating sales talk scripts. Background Art

[0002] Services based on the e-commerce model often require a certain number of customer service representatives to answer customers' relevant questions in real time. However, during major e-commerce shopping festivals, the volume of customer inquiries often shows an explosive growth, far exceeding the reception capacity of customer service representatives. Moreover, due to the particularity of the work content, the turnover rate of customer service employees is relatively high, and enterprises often need to invest a large amount of resources in the training of new employees and the retention of existing employees. To reduce the investment in customer service resources and improve the stability of customer service quality, major e-commerce platforms often introduce intelligent customer service robots based on algorithms.

[0003] Currently, intelligent customer service based on machine learning can already well solve common and simple customer inquiry problems, greatly reducing the proportion of transfers to human agents and significantly reducing the workload of customer service staff. However, in the process of communicating with customers, using traditional knowledge matching or pre-set script templates for automatic replies often results in relatively rigid reply scripts and poor customer experience. Also, during the shopping process, customers often view multiple products simultaneously, compare the parameters between products, and then consult the customer service about the advantages and disadvantages of multiple products. However, fixed script replies based on pre-set script templates often give customers the experience of irrelevant answers, making customers feel not valued, which has a certain negative impact on customers' shopping experience and willingness to purchase products, and at the same time cannot meet the enterprise's personalized recommendation needs for promoted products. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a method, device, electronic device and storage medium for generating sales talk scripts.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0006] In a first aspect, the present invention provides a method for generating sales talk scripts, the method for generating sales talk scripts including:

[0007] Obtain the consultation session of the user for the corresponding product;

[0008] Determine the target product attribute information concerned by the user according to the consultation session;

[0009] Generate a sales session corresponding to the consultation session according to the target product attribute information; wherein, the sales session is used to recommend a recommended product to the user.

[0010] In an alternative embodiment, the step of determining the target product attribute information that the user is interested in according to the consultation session includes:

[0011] Determine the target application scenario that the user is interested in according to the consultation session;

[0012] Determine candidate product attribute information according to the target application scenario and the preset association relationship between product attributes and application scenarios;

[0013] Screen the candidate product attribute information to obtain the target product attribute information that the user is interested in.

[0014] In an alternative embodiment, the association relationship between product attributes and application scenarios is established in the following manner:

[0015] Mark each product attribute to obtain the application scenarios associated with each product attribute;

[0016] Build a knowledge graph based on each product attribute and its associated application scenarios;

[0017] Use a preset speech model to expand the relationships in the knowledge graph to obtain the association relationship between product attributes and application scenarios.

[0018] In an alternative embodiment, the step of generating a sales session corresponding to the consultation session according to the target product attribute information includes:

[0019] Determine the substitutes corresponding to the product, and / or obtain the main promoted products set by the product seller;

[0020] Obtain the first product attribute information of the product, the second product attribute information of the substitute, and / or the third product attribute information of the main promoted product;

[0021] Based on the target product attribute information, the first product attribute information, the second product attribute information, and / or the third product attribute information, determine the recommended product among the product, the substitute, and / or the main promoted product, and generate a sales session for the recommended product.

[0022] In an alternative embodiment, the step of determining the recommended product among the product, the substitute, and / or the main promoted product based on the target product attribute information, the first product attribute information, the second product attribute information, and / or the third product attribute information, and generating a sales session for the recommended product includes:

[0023] Select a target speech example from a pre-established speech corpus;

[0024] Based on the first product attribute information, the second product attribute information, and / or the third product attribute information, determine the attribute information in which the product, the substitute, and / or the main product have differences, obtain the product differentiation attribute information, and acquire the product attribute evaluation index;

[0025] Construct a prompt based on the target sales talk example, the target product attribute information, the product differentiation attribute information, and the product attribute evaluation index;

[0026] Input the prompt into a preset sales talk large model to determine the recommended product among the product, the substitute, and / or the main product, and generate a sales talk for promoting the recommended product; wherein, the sales talk includes the evaluation result and the evaluation basis of the product, the substitute, and / or the main product, and the sales talk has the characteristics of anthropomorphism and scenario.

[0027] In an alternative embodiment, the product attribute evaluation index includes a product attribute qualitative evaluation criterion and a product attribute quantitative evaluation criterion. The product attribute qualitative evaluation criterion is used to evaluate product attributes with text-type attribute parameters, and the product attribute quantitative evaluation criterion is used to evaluate product attributes with numerical-type attribute parameters.

[0028] In an alternative embodiment, the sales talk corpus is established in the following manner:

[0029] Determine the keywords corresponding to each product attribute to obtain the mapping relationship between the product attribute and the keywords;

[0030] According to the mapping relationship, obtain all the sales talks matched by each keyword from a preset historical sales talk set to obtain a sales talk set for each product attribute; wherein, the sales talk set has the characteristics of anthropomorphism and scenario;

[0031] Use a preset sales talk large model to expand the sales talk set for each product attribute to obtain a sales talk corpus containing a set of sales talk examples for each product attribute.

[0032] In a second aspect, the present invention provides a sales talk generation device, and the sales talk generation device includes:

[0033] An acquisition module, configured to acquire a consultation session of a user for a corresponding product;

[0034] A determination module, configured to determine the target product attribute information concerned by the user according to the consultation session;

[0035] A generation module, configured to generate a sales talk corresponding to the consultation session according to the target product attribute information; wherein, the sales talk is used to recommend a recommended product to the user.

[0036] In a third aspect, the present invention provides an electronic device, including a processor and a memory. The memory stores a computer program. When the processor executes the computer program, the method for generating a sales pitch described in any one of the foregoing embodiments is implemented.

[0037] In a fourth aspect, the present invention provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for generating a sales pitch described in any one of the foregoing embodiments is implemented.

[0038] The method, device, electronic device, and storage medium for generating a sales pitch provided by the present invention include: First, obtain a consultation session of a user for a corresponding product. Then, based on the consultation session, determine the target product attribute information that the user is concerned about. Finally, based on the target product attribute information, generate a sales session corresponding to the consultation session; and the sales session is used to recommend a recommended product to the user. By identifying the product attribute information that the user cares about through the consultation session, a recommended product is selected and recommended to the user, and the advantages of the recommended product are explained to the user through a sales pitch, thereby enhancing the user's willingness to purchase and increasing the purchase conversion rate.

[0039] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, makes a detailed description as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0041] Figure 1 Shows a schematic flowchart of the method for generating a sales pitch provided by an embodiment of the present invention;

[0042] Figure 2 Shows a schematic flowchart of establishing an association relationship provided by an embodiment of the present invention;

[0043] Figure 3 Shows a schematic flowchart of establishing a sales pitch corpus provided by an embodiment of the present invention;

[0044] Figure 4 Shows a functional module diagram of the device for generating a sales pitch provided by an embodiment of the present invention;

[0045] Figure 5 Shows a schematic block diagram of the electronic device provided by an embodiment of the present invention.

[0046] Icons: 100 - Electronic device; 110 - Processor; 120 - Memory; 130 - Communication module; 300 - Sales pitch generation device; 310 - Acquisition module; 330 - Determination module; 350 - Generation module; 370 - Establishment module. Detailed implementation manners

[0047] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Usually, the components of the embodiments of the present invention described and illustrated herein can be arranged and designed in various different configurations.

[0048] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0049] It should be noted that relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the presence of additional identical elements in the process, method, article or device including the element.

[0050] Please refer to Figure 1 , which is a schematic flowchart of a method for generating a sales pitch provided by an embodiment of the present invention.

[0051] Step S202, acquire the consultation session of the user for the corresponding commodity.

[0052] Step S204, determine the target commodity attribute information concerned by the user according to the consultation session.

[0053] Step S206, generate a sales session corresponding to the consultation session according to the target commodity attribute information; wherein, the sales session is used to recommend the recommended product to the user.

[0054] It can be understood that in the e-commerce customer service scenario, different users often have different concerns. Therefore, in this embodiment, the consultation session of the user for the corresponding product will be obtained first to understand various questions and needs of the user for the product. Then, the product attributes concerned by the user are identified based on the consultation session, and the target product attribute information is obtained. Then, according to the target product attribute information concerned by the user, a corresponding promotion session is generated to recommend the most suitable product to the user as a recommended product. In this way, the product attributes concerned by the user are identified through the consultation session, and the recommended product is selected based on the product attributes and recommended to the user, thereby enhancing the user's purchase intention and increasing the purchase conversion rate.

[0055] It can be seen that based on the above steps, first, the consultation session of the user for the corresponding product is obtained. Then, according to the consultation session, the target product attribute information concerned by the user is determined. Finally, according to the target product attribute information, a promotion session corresponding to the consultation session is generated; and the promotion session is used to recommend the recommended product to the user. The product attribute information concerned by the user is identified through the consultation session to select the recommended product and recommend it to the user, and the advantages of the recommended product are explained to the user through the promotion words, thereby enhancing the user's purchase intention and increasing the purchase conversion rate.

[0056] Optionally, for step S204, an embodiment of the present invention provides a possible implementation manner.

[0057] Step S204-1, according to the consultation session, determine the target application scenario concerned by the user.

[0058] Step S204-3, according to the target application scenario and the preset association relationship between the product attribute and the application scenario, determine the candidate product attribute information.

[0059] Step S204-5, screen the candidate product attribute information to obtain the target product attribute information concerned by the user.

[0060] In this embodiment, the consultation session includes the historical conversation between the user and the intelligent customer service and the question that the user consults this time. Among them, the intelligent customer service refers to a service that uses artificial intelligence technology to automatically provide product purchase consultation to the user. Based on the historical conversation and the question consulted this time, the application scenario concerned by the user can be analyzed, that is, the target application scenario is obtained. Then, according to the preset association relationship between the product attribute and the application scenario, the product attribute information associated with the target application scenario is determined, that is, the candidate product attribute information that the user may be concerned about is obtained. Then, the question that the user consults this time and the candidate product attribute information are used as the input of the conversation model, and the conversation model is used to identify the consultation intention of the user and extract the user's concern points to screen out the product attributes that the user really concerns from the candidate product attribute information, that is, the target product attribute information is obtained.

[0061] For example, assume that the product the user inquires about is a range hood, and based on the historical conversation and the question in this consultation, it is analyzed that the target application scenarios the user is concerned about are small-sized kitchens, frequent stir-frying, and having the elderly and children at home. Then, using the pre-set correlation between product attributes and application scenarios, the product attribute information associated with the target application scenario is determined, that is, the candidate product attribute information the user may be concerned about is obtained, such as the installation size, air volume, and noise of the range hood. Next, the question in this consultation by the user and the candidate product attribute information are input into the conversation model to screen out the product attributes the user is truly concerned about, and then the target product attribute information such as the installation size and air volume of the range hood is obtained.

[0062] It can be understood that in the embodiment of the present invention, the application scenario the user is concerned about is identified according to the user's consultation session, and the product attributes the user is concerned about are inferred based on the application scenario, so as to accurately identify the user's purchase needs, and generate targeted conversation based on the purchase needs, thereby ensuring that the sales conversation in the reply can meet the user's expectations and enhancing the user's purchase intention.

[0063] Optionally, for the correlation between product attributes and application scenarios in step S204-3, the embodiment of the present invention also provides an implementation method for establishing the correlation. Please refer to Figure 2 .

[0064] Step S208, mark each product attribute to obtain the application scenario associated with each product attribute.

[0065] Step S210, establish a knowledge graph based on each product attribute and its associated application scenario.

[0066] Step S212, use the pre-set conversation model to expand the relationships in the knowledge graph to obtain the correlation between product attributes and application scenarios.

[0067] In this embodiment, in order to establish the correlation between product attributes and application scenarios, each product attribute will be marked first to determine the application scenario related to each product attribute. For example, according to information such as the product manual and user manual, the application scenario associated with each product attribute can be marked. For example, for the product attribute of the capacity of the dishwasher, it can be associated with application scenarios such as the number of family members and the number of dishes washed each time. For the product attribute of the installation size of the range hood, it can be associated with application scenarios such as the kitchen area, small-sized houses, and large-sized houses. For the product attribute of the air volume of the range hood, it can be associated with application scenarios such as cooking habits such as stir-frying and frying, and cooking cuisines such as Sichuan cuisine and Hunan cuisine.

[0068] Then, based on the application scenarios marked for each product attribute, a knowledge graph is constructed to connect the product attributes with the application scenarios through this knowledge graph. Then, using the knowledge built into the preset conversation model, the expression ways of the application scenarios are increased, and relevant application scenarios are added to the product attributes to expand and optimize the relationships in the knowledge graph, thus obtaining the association relationship between the product attributes and the application scenarios. In this way, by using the association relationship between the product attributes and the application scenarios, the user's needs can be understood more accurately, so as to generate more appropriate recommended conversation words.

[0069] Optionally, for step S206, an embodiment of the present invention provides a possible implementation manner.

[0070] Step S206-1, determine the substitutes corresponding to the product, and / or obtain the main promoted products set by the product seller.

[0071] Step S206-3, obtain the first product attribute information of the product, the second product attribute information of the substitute, and / or the third product attribute information of the main promoted product.

[0072] Step S206-5, based on the target product attribute information, the first product attribute information, the second product attribute information, and / or the third product attribute information, determine the recommended product among the product, the substitute, and / or the main promoted product, and generate a sales conversation for the recommended product.

[0073] In this embodiment, the substitutes corresponding to the product can be determined first, then the attribute information of the product, that is, the first product attribute information, and the attribute information of the substitute, that is, the second product attribute information, are obtained. Then, based on the target product attribute information, the first product attribute information, and the second product attribute information, the product and the substitute are compared to determine the recommended product that best matches the user's needs, and a sales conversation for the recommended product is generated to explain the advantages of the recommended product to the user.

[0074] Optionally, in some implementation manners, the product seller may preset a main promoted product in advance, then the attribute information of the main promoted product, that is, the third product attribute information, can be obtained, and then based on the target product attribute information, the first product attribute information, and the third product attribute information, the product and the main promoted product are compared to determine the recommended product that best matches the user's needs, and a sales conversation for the recommended product is generated to explain the advantages of the recommended product to the user. And, in some other implementation manners, the product, the substitute, and the main promoted product can also be compared based on the target product attribute information, the first product attribute information, the second product attribute information, and the third product attribute information to determine the recommended product that best matches the user's needs, and a sales conversation for the recommended product is generated to explain the advantages of the recommended product to the user.

[0075] Optionally, for step S206-5, an embodiment of the present invention provides a possible implementation manner.

[0076] Step S206-5-1, select a target speech example from a pre-established speech corpus.

[0077] Step S206-5-3, based on the first product attribute information, the second product attribute information, and / or the third product attribute information, determine the attribute information in which the product, substitute, and / or main product have differences, obtain the product differential attribute information, and obtain the product attribute evaluation index.

[0078] Step S206-5-5, construct a prompt word based on the target speech example, the target product attribute information, the product differential attribute information, and the product attribute evaluation index.

[0079] Step S206-5-7, input the prompt word into a preset speech model to determine a recommended product among the product, substitute, and / or main product, and generate a sales conversation for the recommended product; wherein, the sales conversation includes the evaluation result and the evaluation basis of the product, substitute, and / or main product, and the sales conversation has the characteristics of anthropomorphism and scenario.

[0080] In this embodiment, for the process of determining the recommended product and generating the sales conversation, a suitable target speech example can be first selected from a pre-established speech corpus. Then, based on the first product attribute information of the product, the second product attribute information of the substitute, and / or the third product attribute information of the main product, the attribute parameters of the product, substitute, and / or main product are compared in full, and the product attributes with the same attribute parameters are excluded, and the product attributes with different attribute parameters are retained, that is, the attribute information in which the product, substitute, and / or main product have differences is obtained, then the product differential attribute information is obtained, and the corresponding product attribute evaluation index is obtained.

[0081] Next, based on the obtained target speech example, the target product attribute information, the product differential attribute information, and the product attribute evaluation index, a prompt word is constructed. That is, the prompt word includes a scenario-based and anthropomorphic speech example at the product attribute granularity, the product attributes concerned by the user, as well as the differential attributes of the product and their evaluation indexes. Optionally, in order to guide the speech model to recommend the main product set by the product seller from the perspective of the product attributes concerned by the user, the prompt word can also include the selling point speech of the main product.

[0082] Finally, input the constructed prompt into the conversation model so that the conversation model can utilize its learning and reasoning capabilities to analyze the information in the prompt, determine the most suitable recommended product among the products, substitutes, and / or main products, and generate a sales conversation for the recommended product. Moreover, the finally generated sales conversation not only includes the evaluation results and bases for the products, substitutes, and / or main products, but also features anthropomorphism and scenarioization. Among them, the evaluation basis can be determined based on information such as product manuals, industry standards, third-party evaluations, and product introductions on official websites.

[0083] It should be noted that the training samples of the conversation model in this embodiment are obtained by heuristically constructing some conversation corpora using open-source large language models such as LLM (Large Language Model) and modifying these conversation corpora to improve the quality of the conversation. Then, the large language model is fine-tuned using these high-quality conversation corpora to obtain the conversation model. In this way, when the conversation model is actually used, it can better generate anthropomorphic, scenarioized, and personalized conversations, thereby increasing the user's willingness to purchase.

[0084] Optionally, the product attribute evaluation criteria in step S206-5-5 include product attribute qualitative evaluation criteria and product attribute quantitative evaluation criteria. The product attribute qualitative evaluation criteria are used to evaluate product attributes with text-type attribute parameters; the product attribute quantitative evaluation criteria are used to evaluate product attributes with numerical-type attribute parameters.

[0085] In this embodiment, product attribute evaluation criteria can be constructed based on information such as product manuals, industry standards, third-party evaluations, and product introductions on official websites. The product attribute evaluation criteria are divided into two types: The first is the product attribute qualitative evaluation criteria, which are used to evaluate product attributes with text-type attribute parameters. That is, the product attribute qualitative evaluation criteria evaluate the attribute parameters from a qualitative perspective. For example, for the product attribute of the water pipe material of a water purifier, its product attribute qualitative evaluation criteria can be that pure copper material is better than stainless steel material.

[0086] The second is the product attribute quantitative evaluation criteria, which are used to evaluate product attributes with numerical-type attribute parameters. That is, the product attribute quantitative evaluation criteria evaluate the attribute parameters from a quantitative perspective. For example, for the product attribute of the air volume of a range hood, its product attribute quantitative evaluation criteria can be that the larger the numerical value of the air volume, the better the performance of the range hood. For the product attribute of the noise of a range hood, its product attribute quantitative evaluation criteria can be that the smaller the numerical value of the noise, the better the performance of the range hood.

[0087] It can be understood that, in the embodiments of the present invention, through these two evaluation criteria of qualitative and quantitative, the attribute differences between different products can be comprehensively compared, so as to facilitate providing more accurate and personalized recommendations for users.

[0088] Optionally, for the dialogue term corpus in step S206-5-1, the embodiments of the present invention also provide an implementation manner for establishing a dialogue term corpus. Please refer to Figure 3 .

[0089] Step S214, determine the keywords corresponding to each product attribute, and obtain the mapping relationship between the product attribute and the keyword.

[0090] Step S216, according to the mapping relationship, obtain all the sales scripts matched by each keyword from the preset historical sales script set, and obtain the sales script set for each product attribute; wherein, the sales script set has the characteristics of anthropomorphism and scenario.

[0091] Step S218, use the preset dialogue model to expand the sales script set for each product attribute, and obtain a dialogue term corpus containing the dialogue example set for each product attribute.

[0092] It can be understood that, in the related art, during the process of a user's conversation with an intelligent customer service, the intelligent customer service usually directly replies with the attribute parameters of the product. This will result in the user lacking a perceptual understanding of the specific product performance, so it is difficult to achieve the purpose of consultation, and the user's purchase intention cannot be converted into actual purchase behavior. That is, the intelligent customer service in the related art can only mechanically reply with the attribute parameters of the product; while the intelligent customer service in the embodiments of the present invention can, according to the usage scenario of the product and in combination with the attribute parameters of the product, explain the characteristics and advantages of the product to the user. Therefore, the embodiments of the present invention will pre-establish a dialogue term corpus with anthropomorphism and scenario, and based on these scripts, guide the dialogue model to generate recommended scripts, so that the user can truly feel the characteristics of the product and enhance the user's perception of the product performance.

[0093] In this embodiment, based on historical sales information, common words used when introducing each product attribute can be collected to obtain keywords corresponding to each product attribute, and a mapping relationship between the product attribute and the keyword is established. Then, in the historical sales pitch set, based on the mapping relationship, the sales pitches matching each keyword are obtained to obtain the sales pitch set for each product attribute. These sales pitch sets have the characteristics of anthropomorphism and scenario. Next, the sales pitch set for each product attribute can be optimized by screening, modifying, etc., to obtain the optimized sales pitch set for each product attribute. Finally, based on the few-shot learning ability and generalization generation ability of the pitch large model, some optimized sales pitches can be randomly selected and input into the pitch large model, so that the pitch large model learns the expression of sales pitches related to product attributes, and the sales pitch set for each product attribute is expanded to obtain the pitch example set for each product attribute, that is, the pitch corpus is obtained.

[0094] It can be understood that the pitch examples in the pitch corpus have the characteristics of anthropomorphism and scenario. For example, for the product attribute of the air volume of the range hood, its attribute parameter is 30m 3 / min. If directly reply to the user "The air volume of the range hood is 30m 3 / min", this cannot let the user perceive the performance of the product. Therefore, for this product attribute, the scenario-based pitch provided by the embodiment of the present invention can be: "The air volume of the range hood is 30m 3 / min. Even if there is a large amount of oil fume generated by stir-frying Sichuan cuisine in an open kitchen, this air volume can suck it up instantly. If it is a small kitchen or other cuisines, it goes without saying." Also, for the product attribute of the power of the water heater, its attribute parameter is first-level energy efficiency. If directly reply to the user "The power of the water heater is first-level energy efficiency and it is very energy-saving", this cannot let the user perceive the performance of the product. Therefore, for this product attribute, the anthropomorphic pitch provided by the embodiment of the present invention can be: "The power of the water heater is first-level energy efficiency, and the power consumption is very low. The electricity bill saved in a year can be used to have a hot pot meal."

[0095] It can be understood that in the embodiment of the present invention, by establishing a pitch corpus with the characteristics of anthropomorphism and scenario, and based on the pitches in the pitch corpus, the pitch large model is guided to generate anthropomorphic and scenario-based recommended pitches, so that the user can truly feel the characteristics of the product and improve the user's perception of the product performance, thereby improving the user's purchase intention and purchase conversion rate.

[0096] To execute the corresponding steps in the above embodiments and each possible way, an implementation manner of a sales pitch generation device is given below. Please refer to Figure 4, which is a functional module diagram of the sales pitch generation device provided by an embodiment of the present invention. It should be noted that the basic principle and the technical effects produced by the sales pitch generation device 300 provided in this embodiment are the same as those in the above embodiment. For the sake of brief description, for the parts not mentioned in this embodiment, reference can be made to the corresponding content in the above embodiment. The sales pitch generation device 300 includes:

[0097] An acquisition module 310, configured to acquire the consultation session of the user for the corresponding commodity.

[0098] A determination module 330, configured to determine the target commodity attribute information concerned by the user according to the consultation session.

[0099] A generation module 350, configured to generate a sales session corresponding to the consultation session according to the target commodity attribute information; wherein, the sales session is used to recommend the recommended product to the user.

[0100] Optionally, the determination module 330 is further configured to: determine the target application scenario concerned by the user according to the consultation session; determine the candidate commodity attribute information according to the target application scenario and the preset association relationship between the commodity attribute and the application scenario; screen the candidate commodity attribute information to obtain the target commodity attribute information concerned by the user.

[0101] Optionally, the generation module 350 is further configured to: determine the substitute corresponding to the commodity, and / or, acquire the main promoted product set by the commodity seller; acquire the first commodity attribute information of the commodity, the second commodity attribute information of the substitute, and / or the third commodity attribute information of the main promoted product; based on the target commodity attribute information, the first commodity attribute information, the second commodity attribute information, and / or the third commodity attribute information, determine the recommended product among the commodity, the substitute, and / or the main promoted product, and generate the sales session of the recommended product.

[0102] Optionally, the generation module 350 is further configured to: select the target sales pitch example from the pre-established sales pitch corpus; based on the first commodity attribute information, the second commodity attribute information, and / or the third commodity attribute information, determine the attribute information with differences among the commodity, the substitute, and / or the main promoted product, obtain the commodity differential attribute information and acquire the commodity attribute evaluation index; construct a prompt word based on the target sales pitch example, the target commodity attribute information, the commodity differential attribute information, and the commodity attribute evaluation index; input the prompt word into the preset sales pitch large model to determine the recommended product among the commodity, the substitute, and / or the main promoted product, and generate the sales session of the recommended product; wherein, the sales session includes the evaluation result and the evaluation basis of the commodity, the substitute, and / or the main promoted product, and the sales session has the characteristics of anthropomorphism and scenario.

[0103] Optionally, the product attribute evaluation indicators include a qualitative evaluation criterion for product attributes and a quantitative evaluation criterion for product attributes. The qualitative evaluation criterion for product attributes is used to evaluate product attributes whose attribute parameters are of the text type, and the quantitative evaluation criterion for product attributes is used to evaluate product attributes whose attribute parameters are of the numerical type.

[0104] Optionally, the sales pitch generation device 300 further includes a building module 370 for: marking each product attribute to obtain the application scenarios associated with each product attribute; establishing a knowledge graph based on each product attribute and its associated application scenarios; and using a preset sales pitch large model to expand the relationships in the knowledge graph to obtain the association relationships between product attributes and application scenarios.

[0105] Optionally, the building module 370 is further used for: determining the keywords corresponding to each product attribute to obtain the mapping relationship between the product attributes and the keywords; according to the mapping relationship, obtaining all the sales pitches matched by each keyword from the preset historical sales pitch set to obtain the sales pitch set for each product attribute; where the sales pitch set has the characteristics of anthropomorphism and scenarioization; and using a preset sales pitch large model to expand the sales pitch set for each product attribute to obtain a speech corpus containing the speech example sets for each product attribute.

[0106] Please refer to Figure 5 , which is a block diagram of the electronic device provided by the embodiment of the present invention. The electronic device 100 includes a processor 110, a memory 120, and a communication module 130. Each component is electrically connected directly or indirectly to each other to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.

[0107] The processor 110 is used to read / write the data or programs stored in the memory 120 and execute corresponding functions. It can be a general-purpose processor, including a CPU (Central Processing Unit), an NP (Network Processor), etc.; it can also be a DSP digital signal processor, an ASIC application-specific integrated circuit, an FPGA field-programmable gate array, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0108] The memory 120 is used to store programs or data. The memory 120 can be a RAM (Random Access Memory), a ROM (Read Only Memory), a PROM (Programmable Read-Only Memory), an EPROM (Erasable Programmable Read-Only Memory), an EEPROM (Electric Erasable Programmable Read-Only Memory), etc.

[0109] The communication module 130 is used to communicate signaling or data with other devices.

[0110] It can be understood that Figure 5 the structure shown is only a schematic diagram of the structure of the electronic device 100, and the electronic device 100 may also include more or fewer components than those shown Figure 5 in the figure, or have a different configuration from that shown Figure 5 in the figure. Figure 5 Each component shown in the figure can be implemented by hardware, software, or a combination thereof.

[0111] The memory in the electronic device provided in the embodiments of the present invention stores a computer program. When the processor executes the computer program, the method for generating sales talk disclosed in the embodiments of the present invention is implemented.

[0112] The embodiments of the present invention also provide a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for generating sales talk disclosed in the embodiments of the present invention is implemented.

[0113] In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions, and operations of devices, methods, and computer program products according to multiple embodiments of the present invention. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system for performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0114] In addition, each functional module in various embodiments of the present invention can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0115] If the above-mentioned functions are implemented in the form of software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0116] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for generating sales pitches, characterized in that: The sales pitch generation method comprises: Obtain the user's consultation session on the corresponding product; Determining target commodity attribute information of interest to the user according to the consultation session; A sales promotion session corresponding to the consultation session is generated according to the target product attribute information; wherein the sales promotion session is used to recommend the recommended product to the user.

2. The method for generating sales pitches according to claim 1, characterized in that: The step of determining the target commodity attribute information that the user is interested in based on the consultation session includes: Determining, based on the consultation session, a target application scenario of interest to the user; Determine candidate product attribute information according to the target application scenario and the association relationship between the preset product attributes and the application scenario; The candidate product attribute information is screened to obtain the target product attribute information that the user is interested in.

3. The method for generating sales pitches according to claim 2, characterized in that: The association between the product attributes and the application scenarios is established in the following manner: Mark each product attribute and obtain the application scenario associated with each product attribute; Build a knowledge graph based on each product attribute and its associated application scenarios; By using the preset big model of speech, the relationship in the knowledge graph is expanded to obtain the correlation between the product attributes and the application scenarios.

4. The method for generating sales pitches according to claim 1, characterized in that: The step of generating a sales promotion session corresponding to the consultation session according to the target product attribute information includes: Determine the substitutes for the product, and / or obtain the main recommended product set by the product seller; Acquire the first commodity attribute information of the commodity, the second commodity attribute information of the substitute, and / or the third commodity attribute information of the main product; Based on the target product attribute information, the first product attribute information, the second product attribute information and / or the third product attribute information, the recommended product is determined from among the products, the substitutes and / or the main recommended product, and a promotion session for the recommended product is generated.

5. The method for generating sales pitches according to claim 4, characterized in that: The step of determining the recommended product from among the products, substitute products and / or main recommended products based on the target product attribute information, the first product attribute information, the second product attribute information and / or the third product attribute information, and generating a promotion session for the recommended product comprises: Select target speech examples from the pre-established speech corpus; Based on the first commodity attribute information, the second commodity attribute information and / or the third commodity attribute information, determining attribute information in which the commodity, the substitute and / or the main recommended product have differences, obtaining differentiated commodity attribute information and obtaining commodity attribute evaluation indicators; Constructing prompt words based on the target speech example, the target product attribute information, the product differentiation attribute information, and the product attribute evaluation index; The prompt words are input into a preset large model of speech to determine the recommended product among the product, the substitutes and / or the main product, and generate a sales promotion conversation for the recommended product; wherein the sales promotion conversation includes the evaluation results and evaluation basis of the product, the substitutes and / or the main product, and the sales promotion conversation has the characteristics of personification and scenario.

6. The method for generating sales pitches according to claim 5, characterized in that: The commodity attribute evaluation index includes a commodity attribute qualitative evaluation standard and a commodity attribute quantitative evaluation standard. The commodity attribute qualitative evaluation standard is used to evaluate commodity attributes whose attribute parameters are text type, and the commodity attribute quantitative evaluation standard is used to evaluate commodity attributes whose attribute parameters are numerical type.

7. The method for generating sales pitches according to claim 5, characterized in that: The discourse corpus is established in the following manner: Determine the keywords corresponding to each product attribute and obtain the mapping relationship between product attributes and keywords; According to the mapping relationship, all sales words matched by each keyword are obtained from the preset historical sales words set, and a sales words set for each commodity attribute is obtained; wherein the sales words set has the characteristics of personification and scenario; By using the preset sales talk model, the sales talk set for each product attribute is expanded to obtain a sales talk corpus containing a talk example set for each product attribute.

8. A sales pitch generation device, characterized in that: The sales talk generating device comprises: An acquisition module, used to acquire the user's consultation session on the corresponding product; A determination module, configured to determine the target commodity attribute information that the user is interested in based on the consultation session; A generating module is used to generate a sales promotion session corresponding to the consulting session according to the target product attribute information; wherein the sales promotion session is used to recommend the recommended product to the user.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and when the processor executes the computer program, the method for generating sales talk according to any one of claims 1 to 7 is implemented.

10. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, implements the sales pitch generation method according to any one of claims 1 to 7.

Images