Method, device and equipment for generating product recommendation words based on artificial intelligence

Abstract

The present invention discloses a method for generating product recommendation speech based on artificial intelligence, which is applied to the field of artificial intelligence technology and is used to solve the technical problems of low production efficiency and poor versatility of product promotion speech. The method provided by the present invention comprises: obtaining the category of the product; according to the category of the product, searching for Internet hot articles related to the corresponding product, extracting the topic description from the Internet hot articles by keyword matching; obtaining the transfer speech of the topic description of the category, the transfer speech is pre-edited and stored; obtaining the description abstract of the product of the category; annotating the highlight description sentence of the product of the category from the description abstract by a pre-trained annotation model; splicing and combining the topic description, the transfer speech and the highlight description sentence to obtain the recommendation speech of the product of the category.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device, computer equipment and storage medium for generating product recommendation words based on artificial intelligence. Background Art

[0002] In the field of product sales, one of the most important links is to describe the relevant information and advantages of the product to customers in order to promote the corresponding products. In this process, the product-related description language is particularly important and has a direct impact on the user's purchase rate. How to obtain product-related sales description language has become a relatively new research direction.

[0003] Currently, when sales personnel promote their products, they generally describe and recommend the product based on their own understanding of the product and industry background. For example, in the insurance field, insurance agents need to prepare some words or scripts before meeting with customers to introduce insurance products. There are also many insurance agents who record short videos using pre-prepared scripts to explain insurance to customers. However, these scripts are also or rely on the insurance agents' own understanding of the product and industry background. Such scripts often affect the subsequent explanation results. Some insurance explanation scripts rely on manual generation by insurance expert agents. The effect and quality of script generation are highly dependent on the experience of experts and agents, and have the defects of weak standardization and low production efficiency.

[0004] Currently, there are also intelligent production methods for product sales talk, but this method can generally only be used for a specific type of product, is not universal, and cannot be directly applied to the generation scenarios of other types of product sales talk. For example, the sales talk of products such as skin care products cannot be directly applied to the sales talk of insurance products.

[0005] Based on this, it is urgent to develop a method for generating product sales talk that is highly versatile and efficient. Summary of the invention

[0006] The embodiments of the present invention provide a method, device, computer equipment and storage medium for generating product recommendation speech based on artificial intelligence to solve the technical problems of low production efficiency and poor versatility of product promotion speech.

[0007] A method for generating product recommendation words based on artificial intelligence, the method comprising:

[0008] Get the product category;

[0009] According to the category of the product, search for Internet hot articles related to the corresponding product, and extract topic descriptions from the Internet hot articles by keyword matching;

[0010] Obtaining the transfer words of the topic description of the category, wherein the transfer words are pre-edited and stored;

[0011] Get description summaries of products in the category;

[0012] Annotating highlight description sentences of the products in the category from the description abstract using a pre-trained annotation model;

[0013] The topic description, the transfer words and the highlight description sentences are combined to obtain the recommendation words for the products of the category.

[0014] A device for generating product recommendation words based on artificial intelligence, the device comprising:

[0015] Category acquisition module, used to obtain product categories;

[0016] A topic extraction module is used to search for Internet hot articles related to the corresponding product according to the category of the product, and extract topic descriptions from the Internet hot articles by keyword matching;

[0017] A transfer speech acquisition module, used to acquire transfer speech of the topic description of the category, wherein the transfer speech is pre-edited and stored;

[0018] An abstract acquisition module, used to acquire description abstracts of products in the category;

[0019] A highlight annotation module, used to annotate the highlight description sentences of the products of the category from the description abstract by using a pre-trained annotation model;

[0020] The splicing module is used to splice and combine the topic description, the transfer words and the highlight description sentences to obtain the recommendation words for the products of the category.

[0021] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for generating product recommendation scripts based on artificial intelligence when executing the computer program.

[0022] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for generating product recommendation scripts based on artificial intelligence.

[0023] The above-mentioned method, device, computer equipment and storage medium for generating product recommendation scripts based on artificial intelligence first obtain the category of the product, then collect Internet hot articles related to the products of the category, extract the topic description from the Internet hot articles, and then obtain the transfer script of the topic description of the product category that is pre-edited and stored, and then obtain the descriptive abstract of the products of the category, and annotate the highlight description sentences of the products of the category from the descriptive abstract through a pre-trained annotation model, and finally splice and combine the topic description, the transfer script and the highlight description sentence to obtain the recommendation script of the products of the category. The method for generating product recommendation scripts based on artificial intelligence proposed in this application can be applied to product promotion scripts in all popular fields, has the characteristics of wide versatility, and the entire generation process of the product recommendation script is completed by artificial intelligence, and also has the effect of high production efficiency of product recommendation scripts. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0025] Figure 1 This is a schematic diagram of an application environment of a method for generating product recommendation scripts based on artificial intelligence in one embodiment of the present invention;

[0026] Figure 2 It is a flow chart of a method for generating product recommendation words based on artificial intelligence in one embodiment of the present invention;

[0027] Figure 3 Embodiment of the present invention Figure 2 Implementation flow chart of step S105;

[0028] Figure 4 This is a schematic diagram of the network structure of the annotation model of the first embodiment of the present invention;

[0029] Figure 5 It is a structural schematic diagram of a device for generating product recommendation words based on artificial intelligence in one embodiment of the present invention;

[0030] Figure 6 is a schematic diagram of a computer device in one embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] The method for generating product recommendation words based on artificial intelligence provided in this application can be applied in Figure 1 In an application environment, the computer device can communicate with the server through a network. The computer device can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices. The server can be implemented as an independent server or a server cluster consisting of multiple servers.

[0033] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0034] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0035] In one embodiment, if Figure 2 As shown, a method for generating product recommendation words based on artificial intelligence is provided, and the method is applied in Figure 1 The computer device in is taken as an example to illustrate, including the following steps S101 to S106.

[0036] S101. Obtain product categories.

[0037] The product category may be obtained by manual selection, for example, the user may input or select the category of the corresponding product in the operation interface, and the computer obtains the product category according to the product category selected by the user.

[0038] Among them, the product categories include cosmetics, insurance, medical devices and / or real estate.

[0039] In other embodiments, the product category may also be classified using a pre-trained product classification model. Specifically, the step of obtaining the product category further includes:

[0040] Inputting the electronic manual of the product into a pre-trained product classification model;

[0041] The category of the product is outputted by the classifier of the product classification model.

[0042] Among them, the product classification model needs to be trained in advance. The method for training the product classification model is to input an electronic instruction manual sample that has been pre-labeled with categories into the product classification model, and output the predicted category of the product corresponding to the electronic instruction manual sample through the product classification model to be trained. When the predicted category of the product is different from the category labeled in advance, adjust the parameters of the product classification model, and cyclically input new electronic instruction manual samples, perform category prediction and parameter adjustment on the product classification model until the predicted category of the product corresponding to the electronic instruction manual sample output by the product classification model is the same as the category labeled in advance, thereby obtaining a trained product classification model.

[0043] It is understandable that when there are fewer possible categories to which a product belongs, the category of the product can be obtained by manual selection, saving time and manpower for training the product classification model. When there are more possible categories to which a product belongs, manual selection of product categories may result in the wrong category being selected. In this case, it is preferred to intelligently classify the product categories through a product classification model, which can improve the accuracy of product classification.

[0044] S102: searching for Internet hot articles related to the corresponding product according to the category of the product, and extracting topic descriptions from the Internet hot articles by keyword matching.

[0045] The extracted topic description is used to provide background information for the recommended products in this category, so as to help consumers understand the usage background of the product and improve users' awareness of the product.

[0046] In one embodiment, the step of searching for Internet hot articles related to the corresponding product according to the category of the product and extracting topic descriptions from the Internet hot articles by keyword matching further includes:

[0047] Acquire a web article whose pageview volume is greater than a preset value, and determine the acquired web article as the Internet hot article;

[0048] Classify the Internet hot articles by using a pre-trained article classification model to obtain the category to which each Internet hot article belongs;

[0049] The topic description is extracted from the hot Internet articles in the same category as the product by keyword matching.

[0050] In one of the embodiments, when the preset value of the page views is, for example, 100,000, it means that the network articles with page views greater than 100,000 can be determined as the Internet hot articles.

[0051] There are many ways to extract the topic description from the hot Internet articles, for example, the topic description can be extracted by keyword matching. Specifically, the step of extracting the topic description from the hot Internet articles of the same category as the product by keyword matching further includes:

[0052] Obtaining at least one pre-configured keyword corresponding to the category of the product;

[0053] Identifying at least one topic segment containing at least one of the keywords in the Internet hot article;

[0054] Obtaining the number of data values ​​contained in each of the topic fragments;

[0055] The topic segment containing the largest number of data values ​​is determined as the topic description.

[0056] In other embodiments, semantic analysis may be performed on the topic segment, and the topic segments may be sorted in descending order according to the analyzed semantic feature vectors, and the top topic segment may be determined as the topic description.

[0057] This embodiment extracts topic descriptions from hot Internet articles and uses the topic descriptions as part of product recommendation words, so that consumers can understand the background of product application and understand the recommended products more quickly.

[0058] S103: Obtain transfer words for the topic description of the category, wherein the transfer words are pre-edited and stored.

[0059] In one of the embodiments, the source of the transfer words is pre-edited and stored by the user, and is used to pre-edit the transfer words corresponding to each product category. For example, in the insurance field, the transfer words for the product category of critical illness insurance can be "Major diseases come unexpectedly, and only by being prepared at all times and guarding against risks can you live in peace!", and the transfer words for the product category of accident insurance can be "Accidents come unexpectedly, and only by being prepared at all times and guarding against risks can you live in peace!".

[0060] When obtaining the transfer speech of a certain topic description, the transfer speech of the topic description of the corresponding product category is obtained and the transfer speech is used as part of the product recommendation speech, so that the semantics of the final generated product recommendation speech is more natural, and the product recommendation speech generated based on artificial intelligence is more fluent.

[0061] S104: Obtain description abstracts of products in the category.

[0062] In one embodiment, the explanatory abstract includes but is not limited to product posters, advertising sales pitches, etc. When the product is an insurance product, the explanatory abstract is specifically an insurance poster.

[0063] S105: annotate highlight description sentences of the products of the category from the description abstract using a pre-trained annotation model.

[0064] The annotation model is specifically used to mark the beginning, middle and end of the highlight description sentence in the explanatory abstract.

[0065] In one embodiment, the step of annotating highlight description sentences of the products of the category from the description abstract using a pre-trained annotation model further includes S301 to S305.

[0066] S301, splitting the explanatory abstract into a plurality of consecutive sentences in sequence;

[0067] S302, converting the plurality of continuous sentences into corresponding continuous text feature vectors in sequence;

[0068] S303, inputting the plurality of continuous text feature vectors into the Transformer module of the trained annotation model in batches, and outputting the category of the sentence corresponding to each of the text feature vectors through the classifier of the trained annotation model, wherein the category of the sentence includes beginning, middle and end;

[0069] S304, extracting sentences with output categories of beginning and middle;

[0070] S305: sequentially concatenate the extracted sentences to obtain the highlight description sentences.

[0071] According to an application scenario of this embodiment, for example: the extracted sentences are concatenated and combined in sequence, and the resulting highlight description sentence is "Ping An Fu 20, comprehensive coverage for mild, moderate and severe illnesses, with increased coverage and high benefits."

[0072] In one embodiment, the network structure diagram of the annotation model is as follows: Figure 4As shown, sentencei1, sentencei2, sentencei3 represent the consecutive sentences included in the explanatory abstract. Any three consecutive sentences in the explanatory abstract can also be represented by sentencei i-1 、sentence i 、sentence i+1 , EMB is emb feature vector, represents the text feature vector obtained after converting each of the sentences, transformer represents the Transformer module, LSTM (Long Short-Term Memory) represents the long short-term memory artificial neural network, Linear represents the classifier, B represents the classification result is the beginning, I represents the classification result is the middle, and O represents the classification result is the end.

[0073] In one embodiment, the step of training the annotation model includes: inputting a sentence sample carrying a highlight label into the annotation model, outputting a predicted label of the sentence sample through the classifier of the annotation model to be trained, when the predicted label is different from the highlight label of the sentence sample that has been annotated in advance, calculating the loss of the annotation model according to the loss function of the annotation model, and adjusting the parameters of the annotation model, cyclically inputting new sentence samples, predicting labels and calculating losses for the annotation model, until the loss function of the annotation model converges according to the result of the loss calculation, thereby obtaining a trained annotation model.

[0074] In one embodiment, the loss function of the labeling model may be a cross entropy loss function.

[0075] It can be understood that the number of classifiers of the annotation model is the same as the number of sentences input into the annotation model in each batch. Figure 3 In the given embodiment of the annotation model, the number of classifiers is three, and in other embodiments, the number of classifiers may also be four or five. The more classifiers there are, the more sentences the annotation model can annotate in a single batch, and the higher the annotation efficiency. However, the data processing volume of the intermediate coding layer of the annotation model is also larger, and the data processing capability of the computer device is also higher. When actually constructing the annotation model, the user can design the structure of the annotation model based on actual needs and in combination with the data processing capability of the computer used for sentence annotation.

[0076] This embodiment uses a pre-trained annotation model to intelligently extract highlight description sentences from the product description abstract, and uses the highlight description sentences as part of the product recommendation language, so that the finally generated product recommendation language can concisely point out the advantages of the recommended product, allowing users to understand the recommended product in a shorter time, thereby increasing the purchase rate.

[0077] S106: Combine the topic description, the transfer words and the highlight description sentences to obtain the recommendation words for the products in the category.

[0078] In one embodiment, the step of combining the topic description, the transfer words and the highlight description sentence to obtain the recommendation words for the products in the category further includes:

[0079] The topic description is used as the beginning of the recommendation words, the highlight description sentence is used as the end of the recommendation words, and the transfer words are used as the middle part, and they are combined and spliced ​​into the recommendation words for the product.

[0080] In this embodiment, using the topic description as the beginning of the recommendation language allows users to quickly understand the application scenarios and application requirements of the recommended product, using the highlight description sentence as the end of the recommendation language allows users to concisely understand the characteristics of the recommended product, and splicing the transfer language between the topic description and the highlight description sentence makes the final product recommendation language more fluent and the semantic transition more natural.

[0081] Taking the critical illness insurance category as an example, the recommended words obtained according to an application scenario of this embodiment are as follows:

[0082] {February 4 is World Cancer Day. Data from the World Health Organization shows that in 2020, the number of cancer patients diagnosed worldwide reached 19.3 million, and the number of people who died from cancer increased to 10 million.}{Major diseases come unexpectedly. Only by being prepared at all times and preventing risks can you live in peace!}{Ping An Fu 20, comprehensive coverage for mild, moderate and severe diseases, and high insurance benefits}.

[0083] Among them, the topic description is "February 4 is World Cancer Day. Data from the World Health Organization show that in 2020, the number of diagnosed cancer patients worldwide reached 19.3 million, and the number of people who died from cancer increased to 10 million." The transfer script is "Major diseases come unexpectedly. Only by being prepared at all times and guarding against risks can you live in peace!", and the highlight description sentence is "Ping An Fu 20, comprehensive coverage of light, moderate and severe diseases, and high insurance growth and payment."

[0084] The method for generating product recommendation scripts based on artificial intelligence proposed in the present embodiment first obtains the category of the product, then collects Internet hot articles related to the products of the category, extracts the topic description from the Internet hot articles, and then obtains the transfer script of the topic description of the product category that is pre-edited and stored, and then obtains the descriptive abstract of the products of the category, and annotates the highlight description sentences of the products of the category from the descriptive abstract through a pre-trained annotation model, and finally splices and combines the topic description, the transfer script and the highlight description sentence to obtain the recommendation script of the products of the category. The method for generating product recommendation scripts based on artificial intelligence proposed in the present application can be applied to product promotion scripts in all popular fields, has the characteristics of wide versatility, and the entire generation process of the product recommendation script is completed by artificial intelligence, and also has the effect of high production efficiency of the product recommendation script.

[0085] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0086] In one embodiment, a device for generating product recommendation words based on artificial intelligence is provided, and the device for generating product recommendation words based on artificial intelligence corresponds to the method for generating product recommendation words based on artificial intelligence in the above embodiment. Figure 5 As shown, the device 100 for generating product recommendation speech based on artificial intelligence includes a category acquisition module 11, a topic extraction module 12, a transfer speech acquisition module 13, an abstract acquisition module 14, a highlight marking module 15 and a splicing module 16. The functional modules are described in detail as follows:

[0087] Category acquisition module 11, used to acquire the category of the product;

[0088] The topic extraction module 12 is used to search for Internet hot articles related to the corresponding product according to the category of the product, and extract topic descriptions from the Internet hot articles by keyword matching;

[0089] A transfer speech acquisition module 13 is used to acquire transfer speech of the topic description of the category, wherein the transfer speech is pre-edited and stored;

[0090] Abstract acquisition module 14, used to acquire description abstracts of products in the category;

[0091] A highlight annotation module 15 is used to annotate the highlight description sentences of the products of the category from the description abstract by using a pre-trained annotation model;

[0092] The splicing module 16 is used to splice and combine the topic description, the transfer words and the highlight description sentences to obtain the recommendation words for the products of the category.

[0093] In one embodiment, the category acquisition module 11 includes:

[0094] A manual input unit, used to input the electronic manual of the product into a pre-trained product classification model;

[0095] A category output unit is used to output the category of the product through the classifier of the product classification model.

[0096] Among them, the product classification model needs to be trained in advance. The method for training the product classification model is to input an electronic instruction manual sample that has been pre-labeled with categories into the product classification model, and output the predicted category of the product corresponding to the electronic instruction manual sample through the product classification model to be trained. When the predicted category of the product is different from the category labeled in advance, adjust the parameters of the product classification model, and cyclically input new electronic instruction manual samples, perform category prediction and parameter adjustment on the product classification model until the predicted category of the product corresponding to the electronic instruction manual sample output by the product classification model is the same as the category labeled in advance, thereby obtaining a trained product classification model.

[0097] It is understandable that when there are fewer possible categories to which a product belongs, the category of the product can be obtained by manual selection, saving time and manpower for training the product classification model. When there are more possible categories to which a product belongs, manual selection of product categories may result in the wrong category being selected. In this case, it is preferred to intelligently classify the product categories through a product classification model, which can improve the accuracy of product classification.

[0098] In one of the embodiments, the source of the transfer words is pre-edited and stored by the user, and is used to pre-edit the transfer words corresponding to each product category. For example, in the insurance field, the transfer words for the product category of critical illness insurance can be "Major diseases come unexpectedly, and only by being prepared at all times and guarding against risks can you live in peace!", and the transfer words for the product category of accident insurance can be "Accidents come unexpectedly, and only by being prepared at all times and guarding against risks can you live in peace!".

[0099] When obtaining the transfer speech of a certain topic description, the transfer speech of the topic description of the corresponding product category is obtained and the transfer speech is used as part of the product recommendation speech, so that the semantics of the final generated product recommendation speech is more natural, and the product recommendation speech generated based on artificial intelligence is more fluent.

[0100] In one embodiment, the topic extraction module 12 includes:

[0101] An article acquisition unit, used for acquiring a network article whose pageview volume is greater than a preset value, and determining the acquired network article as the Internet hot article;

[0102] A classification unit, used to classify the Internet hot articles by using a pre-trained article classification model to obtain the category to which each Internet hot article belongs;

[0103] The topic description extraction unit is used to extract the topic description from the Internet hot articles of the same category as the product by keyword matching.

[0104] In one of the embodiments, when the preset value of the page views is, for example, 100,000, it means that the network articles with page views greater than 100,000 can be determined as the Internet hot articles.

[0105] In one embodiment, the topic description extraction unit further includes:

[0106] A keyword acquisition subunit, used to acquire at least one pre-configured keyword corresponding to the category of the product;

[0107] A topic segment identification subunit, used for identifying at least one topic segment containing at least one of the keywords in the Internet hot article;

[0108] A number acquisition subunit, used to acquire the number of data values ​​contained in each of the topic segments;

[0109] The topic description determination subunit is used to determine the topic segment containing the largest number of data values ​​as the topic description.

[0110] In other embodiments, semantic analysis may be performed on the topic segment, and the topic segments may be sorted in descending order according to the analyzed semantic feature vectors, and the top topic segment may be determined as the topic description.

[0111] This embodiment extracts topic descriptions from hot Internet articles and uses the topic descriptions as part of product recommendation words, so that consumers can understand the background of product application and understand the recommended products more quickly.

[0112] In one embodiment, the bright spot marking module includes:

[0113] A splitting unit, used for splitting the explanatory abstract into a plurality of consecutive sentences in sequence;

[0114] A vector conversion unit, used for converting the plurality of continuous sentences into corresponding continuous text feature vectors in sequence;

[0115] An input unit, used to input the plurality of continuous text feature vectors into the Transformer module of the trained annotation model in batches, and output the category of the sentence corresponding to each of the text feature vectors through the classifier of the trained annotation model, wherein the category of the sentence includes beginning, middle and end;

[0116] An extraction unit, used to extract sentences with output categories of beginning and middle;

[0117] The combining unit is used to sequentially combine the extracted sentences to obtain the highlight description sentence.

[0118] According to an application scenario of this embodiment, for example: the extracted sentences are concatenated and combined in sequence, and the resulting highlight description sentence is "Ping An Fu 20, comprehensive coverage for mild, moderate and severe illnesses, with increased coverage and high benefits."

[0119] In one embodiment, the step of training the annotation model includes: inputting a sentence sample carrying a highlight label into the annotation model, outputting a predicted label of the sentence sample through the classifier of the annotation model to be trained, when the predicted label is different from the highlight label of the sentence sample that has been annotated in advance, calculating the loss of the annotation model according to the loss function of the annotation model, and adjusting the parameters of the annotation model, cyclically inputting new sentence samples, predicting labels and calculating losses for the annotation model, until the loss function of the annotation model converges according to the result of the loss calculation, thereby obtaining a trained annotation model.

[0120] It can be understood that the number of classifiers of the annotation model is the same as the number of sentences input into the annotation model in each batch. Figure 3 In the given embodiment of the annotation model, the number of classifiers is three, and in other embodiments, the number of classifiers may also be four or five. The more classifiers there are, the more sentences the annotation model can annotate in a single batch, and the higher the annotation efficiency. However, the data processing volume of the intermediate coding layer of the annotation model is also larger, and the data processing capability of the computer device is also higher. When actually constructing the annotation model, the user can design the structure of the annotation model based on actual needs and in combination with the data processing capability of the computer used for sentence annotation.

[0121] This embodiment uses a pre-trained annotation model to intelligently extract highlight description sentences from the product description abstract, and uses the highlight description sentences as part of the product recommendation language, so that the finally generated product recommendation language can concisely point out the advantages of the recommended product, allowing users to understand the recommended product in a shorter time, thereby increasing the purchase rate.

[0122] In one embodiment, the splicing module 16 is specifically used to use the topic description as the beginning of the recommendation words, the highlight description sentence as the end of the recommendation words, and the transfer words as the middle part to combine and splice into the recommendation words for the product.

[0123] In this embodiment, using the topic description as the beginning of the recommendation language allows users to quickly understand the application scenarios and application requirements of the recommended products, using the highlight description sentence as the end of the recommendation language allows users to concisely understand the characteristics of the recommended product, and splicing the transfer language between the topic description and the highlight description sentence makes the final product recommendation language more fluent and the semantic transition more natural.

[0124] Taking the critical illness insurance category as an example, the recommended words obtained according to an application scenario of this embodiment are as follows:

[0125] {February 4 is World Cancer Day. Data from the World Health Organization shows that in 2020, the number of cancer patients diagnosed worldwide reached 19.3 million, and the number of people who died from cancer increased to 10 million.}{Major diseases come unexpectedly. Only by being prepared at all times and preventing risks can you live in peace!}{Ping An Fu 20, comprehensive coverage for mild, moderate and severe diseases, and high insurance benefits}.

[0126] Among them, the topic description is "February 4 is World Cancer Day. Data from the World Health Organization show that in 2020, the number of diagnosed cancer patients worldwide reached 19.3 million, and the number of people who died from cancer increased to 10 million." The transfer script is "Major diseases come unexpectedly. Only by being prepared at all times and guarding against risks can you live in peace!", and the highlight description sentence is "Ping An Fu 20, comprehensive coverage of light, moderate and severe diseases, and high insurance growth and payment."

[0127] Among them, the terms "including" and "having" and any variations of them are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The division of modules appearing in this application is merely a logical division, and there may be other division methods when implemented in actual applications.

[0128] The device for generating product recommendation scripts based on artificial intelligence proposed in this embodiment first obtains the category of the product, then collects Internet hot articles related to the products of the category, extracts the topic description from the Internet hot articles, and then obtains the transfer script of the topic description of the product category that is pre-edited and stored, and then obtains the descriptive abstract of the products of the category, and marks the highlight description sentences of the products of the category from the descriptive abstract through a pre-trained annotation model, and finally splices and combines the topic description, the transfer script and the highlight description sentences to obtain the recommendation script of the products of the category. The method for generating product recommendation scripts based on artificial intelligence proposed in this application can be applied to product promotion scripts in all popular fields, has the characteristics of wide versatility, and the entire generation process of the product recommendation script is completed by artificial intelligence, and also has the effect of high production efficiency of the product recommendation script.

[0129] For the specific limitations of the device for generating product recommendation speech based on artificial intelligence, please refer to the limitations of the method for generating product recommendation speech based on artificial intelligence above, which will not be repeated here. Each module in the above-mentioned device for generating product recommendation speech based on artificial intelligence can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0130] In one embodiment, a computer device is provided, which may be a server, and includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a storage medium and an internal memory. The storage medium includes a non-volatile storage medium and / or a volatile storage medium, and the storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the storage medium. The database of the computer device is used to store data involved in a method for generating product recommendation speech based on artificial intelligence. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for generating product recommendation speech based on artificial intelligence is implemented.

[0131] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 6As shown. The computer device includes a processor, a memory, a network interface, a display screen and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a storage medium and an internal memory. The storage medium includes a non-volatile storage medium and / or a volatile storage medium, and the storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When the computer program is executed by the processor, a method for generating product recommendation words based on artificial intelligence is implemented.

[0132] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method for generating product recommendation words based on artificial intelligence in the above embodiment are implemented, for example: Figure 2 Steps 101 to 106 and other extensions of the method and related steps are shown. Alternatively, when the processor executes the computer program, the functions of each module / unit of the device for generating product recommendation speech based on artificial intelligence in the above embodiment are realized, for example Figure 5 The functions of modules 11 to 16 are shown in Figure 1. To avoid repetition, they will not be described here.

[0133] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, and various interfaces and lines are used to connect various parts of the entire computer device.

[0134] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the computer device by running or executing the computer program and / or module stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, video data, etc.), etc.

[0135] The memory may be integrated into the processor or may be arranged separately from the processor.

[0136] In one embodiment, a computer-readable storage medium is provided on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for generating product recommendation words based on artificial intelligence in the above embodiment are implemented, for example: Figure 2 Steps 101 to 106 and other extensions of the method and related steps are shown. Alternatively, when the computer program is executed by the processor, the functions of each module / unit of the device for generating product recommendation speech based on artificial intelligence in the above embodiment are realized, for example Figure 5 The functions of modules 11 to 16 are shown in Figure 1. To avoid repetition, they will not be described here.

[0137] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile and / or volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0138] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0139] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A method for generating product recommendation scripts based on artificial intelligence, characterized in that: include: Get the product category; According to the category of the product, search for Internet hot articles related to the corresponding product, and extract topic descriptions from the Internet hot articles by keyword matching; Obtaining the transfer words of the topic description of the category, wherein the transfer words are pre-edited and stored; Obtain description summaries of products in the category; Annotating highlight description sentences of the products in the category from the description abstract using a pre-trained annotation model; The topic description, the transfer words and the highlight description sentences are combined to obtain the recommendation words for the products of the category.

2. The method for generating product recommendation words based on artificial intelligence according to claim 1, characterized in that: The step of obtaining the category of the product further comprises: Inputting the electronic manual of the product into a pre-trained product classification model; The category of the product is outputted by the classifier of the product classification model.

3. The method for generating product recommendation words based on artificial intelligence according to claim 1, characterized in that: The step of searching for Internet hot articles related to the corresponding product according to the category of the product and extracting topic descriptions from the Internet hot articles by keyword matching further includes: Acquire a web article whose pageview volume is greater than a preset value, and determine the acquired web article as the Internet hot article; Classify the Internet hot articles by using a pre-trained article classification model to obtain the category to which each Internet hot article belongs; The topic description is extracted from the hot Internet articles in the same category as the product by keyword matching.

4. The method for generating product recommendation words based on artificial intelligence according to claim 3, characterized in that: The step of extracting the topic description from the hot Internet articles of the same category as the product by keyword matching further includes: Obtaining at least one pre-configured keyword corresponding to the category of the product; Identifying at least one topic segment containing at least one of the keywords in the Internet hot article; Obtaining the number of data values ​​contained in each of the topic fragments; The topic segment containing the largest number of data values ​​is determined as the topic description.

5. The method for generating product recommendation words based on artificial intelligence according to claim 1, characterized in that: The step of annotating highlight description sentences of the products of the category from the description abstract by using the pre-trained annotation model further comprises: Sequentially splitting the explanatory abstract into a plurality of consecutive sentences; Converting the plurality of continuous sentences into corresponding continuous text feature vectors in sequence; Inputting the plurality of continuous text feature vectors into the Transformer module of the trained annotation model in batches, and outputting the category of the sentence corresponding to each of the text feature vectors through the classifier of the trained annotation model, wherein the category of the sentence includes beginning, middle and end; Extract sentences with output categories of beginning and middle; The extracted sentences are concatenated and combined in sequence to obtain the highlight description sentence.

6. The method for generating product recommendation words based on artificial intelligence according to claim 1, characterized in that: The step of combining the topic description, the transfer words and the highlight description sentences to obtain the recommendation words for the products of the category further includes: The topic description is used as the beginning of the recommendation words, the highlight description sentence is used as the end of the recommendation words, and the transfer words are used as the middle part, and they are combined and spliced ​​into the recommendation words for the product.

7. The method for generating product recommendation words based on artificial intelligence according to any one of claims 1 to 6, characterized in that: The categories of the products include cosmetics, insurance, medical devices and / or real estate.

8. A device for generating product recommendation words based on artificial intelligence, characterized in that: The device comprises: Category acquisition module, used to obtain product categories; A topic extraction module is used to search for Internet hot articles related to the corresponding product according to the category of the product, and extract topic descriptions from the Internet hot articles by keyword matching; A transfer speech acquisition module, used to acquire transfer speech of the topic description of the category, wherein the transfer speech is pre-edited and stored; An abstract acquisition module, used to acquire description abstracts of products in the category; A highlight annotation module, used to annotate the highlight description sentences of the products of the category from the description abstract by using a pre-trained annotation model; The splicing module is used to splice and combine the topic description, the transfer words and the highlight description sentences to obtain the recommendation words for the products of the category.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for generating product recommendation scripts based on artificial intelligence as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for generating product recommendation scripts based on artificial intelligence as described in any one of claims 1 to 7 are implemented.

Images