Vending machine intelligent shopping guide method and system based on voice interaction

Abstract

The invention discloses a vending machine intelligent shopping guide method and system based on voice interaction. The objective of the invention is to realize accurate recognition of user demands through voice recognition and natural language processing technologies, and perform intelligent recommendation in combination with commodity big data. According to the method, through integration of a high-sensitivity microphone and voice recognition, a voice instruction of a user is converted into text information, and a purchase intention behind the user is analyzed. Meanwhile, detailed information of commodities sold by the vending machine is collected, and after standardization processing and feature extraction, the detailed information serves as training data of the deep learning model. Personalized commodity recommendation can be generated according to user requirements through a trained and optimized model, and the personalized commodity recommendation is fed back to a user through a high-resolution display screen or voice output. According to the invention, the shopping experience of the user is improved, and the purchase conversion rate is improved through intelligent recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of vending machines, and in particular to an intelligent shopping guide method and system for vending machines based on voice interaction. Background Art

[0002] In the traditional vending machine shopping scenario, users usually need to manually browse the product list and make selections, which is inefficient and lacks personalized recommendations.

[0003] In recent years, with the development of artificial intelligence technology, there have emerged artificial intelligence shopping guide methods and terminal devices as described in Patent CN111724236A. The specific solutions are as follows: The artificial intelligence shopping guide method is applied to a terminal device, and the terminal device has a display interface. The method includes: when the user is within the preset range of the terminal device, retrieving the user behavior information within a preset time range from the big data service platform, and displaying a list of available products according to the user behavior information. The order of the available product list is sorted from high to low according to the degree of user intention; when there is a target product in the available product list, in response to the user's order operation for the target product, displaying the target on the product order interface; having basic information and delivery time information; providing a payment channel for the user; in response to the user's payment success operation, when the delivery time information indicates immediate delivery, sending the target product to the delivery window; when the delivery time information indicates delayed delivery, sending a task order including the target product order to the first client according to the basic order information, and sending a logistics reminder information to the second client. The task order including the target product order is used for the user to instruct the deliveryman to deliver the target product; the logistics reminder information is used to instruct the user to query the logistics information of the target product.

[0004] This technology recommends products by analyzing the user's behavior information within the range of the terminal. However, this method has some limitations. First, it relies on the user's behavior data and cannot directly obtain the user's real-time needs. Second, the terminal device lacks visual information, and users cannot perceive that the product recommendation is generated according to their personal preferences. In addition, this technology cannot meet the flexible needs of users and lacks an interactive means for users to express their expectations. Summary of the Invention

[0005] Based on this, the embodiments of the present application provide an intelligent shopping guide method and system for vending machines based on voice interaction, which can directly interact with users and provide accurate product recommendations according to their needs.

[0006] In a first aspect, an intelligent shopping guide method for vending machines based on voice interaction is provided. The method includes:

[0007] Obtain the user's voice command through the voice input device integrated in the vending machine, and perform recognition processing on the user's voice command to obtain user demand information;

[0008] Extract the information of the goods on sale in the vending machine, and perform preprocessing and feature extraction on the information of the goods on sale to obtain the product features of each good on sale; wherein, the information of the goods on sale at least includes the product name, selling points, ingredient list, nutritional components, price, and inventory;

[0009] After corresponding the user demand information with the product features, use it as training data to train the deep learning model, deploy the trained deep learning model, and perform intelligent shopping guidance on the newly obtained user demand information.

[0010] Optionally, obtaining the user's voice command through the voice input device integrated in the vending machine, and performing recognition processing on the user's voice command to obtain user demand information, specifically including:

[0011] Obtain the user's voice command through the highly sensitive microphone integrated in the vending machine, convert the voice command into text information using voice recognition technology, and parse the text information through natural language processing technology to extract keywords and semantics to obtain user demand information; at the same time, maintain a list of hot words and sensitive words.

[0012] Optionally, preprocessing and feature extraction of the information of the goods on sale to obtain the product features of each good on sale, including:

[0013] Unify the measurement units of the ingredient list and nutritional components into international standard units;

[0014] Unify the product prices into currency units;

[0015] Standardize the naming of the product name to ensure the consistency of the name of the same product among different suppliers;

[0016] Further, through feature extraction, obtain the keywords in the product selling points, calculate the proportion of nutritional components, determine the product price range, and analyze the real-time status of the inventory.

[0017] Optionally, after corresponding the user demand information with the product features, use it as training data to train the deep learning model, specifically including:

[0018] Through natural language processing technology, extract the keywords, semantic information, and constraints in the user demand information, and label the parsed user demand information as a structured data format; wherein, the constraints include, for example, price range, nutritional component requirements,

[0019] Match the labeled user demand information with the product features to generate training samples.

[0020] Optionally, match the labeled user requirement information with the product features, specifically including:

[0021] Match the keywords in the user requirements with the keywords in the product features;

[0022] For the constraint conditions in the user requirements, check whether the product features meet the constraint conditions;

[0023] According to the matching results, generate training samples labeled with the corresponding relationship between user requirements and product features;

[0024] Use the generated matching samples to train the deep learning model so that the model can learn the mapping relationship between user requirements and product features.

[0025] Use the generated matching samples to train the deep learning model, specifically including:

[0026] Select the deep learning model architecture for natural language processing and recommendation tasks; among them, the deep learning model architecture at least includes a recurrent neural network, a long short-term memory network, and a Transformer architecture;

[0027] Convert the matching samples into an input format acceptable to the model, including converting text information into word embedding vectors;

[0028] Use the labeled matching samples to perform supervised training on the model, optimize the model parameters, so that it can generate accurate product recommendations according to user requirements;

[0029] The method further includes:

[0030] Display the high-definition image of the virtual shopping guide through a high-resolution display screen, and output natural voice synthesis sounds using a high-quality sound amplification device;

[0031] When the user issues a voice command, the virtual shopping guide immediately makes a matching response through preset actions and expressions, and at the same time answers the user's questions by voice; among them, the preset actions at least include waving and pointing at the product, and the expressions at least include smiling.

[0032] In the second aspect, a vending machine intelligent shopping guide system based on voice interaction is provided, and the system includes:

[0033] A voice recognition and natural language processing module, configured to obtain a user voice command through a voice input device integrated in the vending machine, and perform recognition processing on the user voice command to obtain user requirement information;

[0034] A product information extraction and preprocessing module, which is used to extract the on-sale product information of the vending machine, preprocess and extract features from the on-sale product information to obtain the product features of each on-sale product; wherein, the on-sale product information at least includes product name, selling points, ingredient list, nutritional components, price, and inventory.

[0035] A large model training and optimization module, which is used to use the user demand information and product features as training data to train a deep learning model after corresponding them, deploy the trained deep learning model, and conduct intelligent shopping guidance for newly obtained user demand information.

[0036] In a third aspect, an electronic device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the method for intelligent shopping guidance of a vending machine based on voice interaction according to any one of the first aspects above.

[0037] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the method for intelligent shopping guidance of a vending machine based on voice interaction according to any one of the first aspects above.

[0038] In a fifth aspect, a computer program product is provided, including a computer program / instructions. When the computer program / instructions are executed by a processor, it implements the method for intelligent shopping guidance of a vending machine based on voice interaction according to any one of the first aspects above.

[0039] The beneficial effects brought by the technical solutions provided in the embodiments of the present application at least include:

[0040] (1) Through voice interaction technology, users can naturally communicate with the vending machine without manual operation, greatly improving the convenience and efficiency of shopping. At the same time, the system can provide personalized product recommendations according to the real-time needs of users, enhancing the shopping experience of users.

[0041] (2) Analyze the user's needs through a deep learning model and generate intelligent recommendations in combination with product features. This data-driven recommendation method can accurately match the user's needs, improving user satisfaction and purchase conversion rate.

[0042] (3) The integrated virtual shopping guide image and voice interaction function enable the vending machine to interact with users in a more vivid and interesting way. Description of the Drawings

[0043] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only exemplary. For those of ordinary skill in the art, without creative efforts, other implementation drawings can also be obtained based on the provided drawings.

[0044] Figure 1 It is a flowchart of the steps of an intelligent shopping guide method for a vending machine based on voice interaction provided by an embodiment of the present application;

[0045] Figure 2 It is a block diagram of an intelligent shopping guide system for a vending machine based on voice interaction provided by an embodiment of the present application;

[0046] Figure 3 It is a schematic diagram of an electronic device provided by an embodiment of the present application. Specific Embodiments

[0047] In order to make the purpose, technical solutions and advantages of the present application more clear, the following further details the present application in conjunction with the drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present application and are not used to limit the present application.

[0048] In the description of the present invention, the terms "including", "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units does not necessarily limit to the clearly listed steps or units, but may also include other steps or units inherent to these processes, methods, products or devices that are not clearly listed, or steps or units added by further optimized solutions based on the concept of the present invention.

[0049] In this application, the problems of how to enable the vending machine to identify the purchase needs of users through voice interaction and how to enable the vending machine to intelligently generate product recommendations through product big data are mainly solved.

[0050] In this application, by integrating multimedia interaction devices such as high-sensitivity microphones and high-definition cameras, combined with advanced speech recognition technology and image recognition algorithms, the vending machine is enabled to have the ability to conduct natural language conversations with humans. At the same time, key attribute information of products (such as selling points, ingredient lists, nutritional components, etc.) is extracted from a massive product database as the training data and inference basis for the machine learning large model to achieve accurate product information matching and recommendation.

[0051] In addition, a vivid and lifelike virtual shopping guide image is created by using a high-resolution large-screen display and high-quality sound amplification equipment. This image can not only interact in real time according to the user's voice commands, but also enhance the authenticity and fun of communication through visual elements such as expressions and actions, thereby improving the user's shopping experience and satisfaction.

[0052] On the basis of traditional vending machines, we introduce advanced speech recognition and natural language processing technologies to ensure that the vending machines can accurately capture the user's voice commands and further understand the purchase intentions behind them, such as distinguishing different needs such as asking for product information, checking prices, placing orders, etc., thereby improving user experience and purchasing efficiency. Figure 1 , which shows a flow chart of a vending machine intelligent shopping guide method based on voice interaction provided by an embodiment of the present application. The method may include the following steps:

[0053] S1, obtaining user voice commands through a voice input device integrated in the vending machine, and recognizing and processing the user voice commands to obtain user demand information.

[0054] In this step, accurate recognition of user voice commands is achieved, and user intentions are analyzed through natural language processing technology, such as keyword extraction and semantic understanding. In addition, information such as hot words and sensitive words that are easier to recognize and should not be recognized is maintained to ensure that the vending machine can accurately respond to user needs.

[0055] In this application, a high-sensitivity microphone is integrated into the vending machine to capture the user's voice commands. The microphone needs to have a noise reduction function to ensure that the voice signal can be clearly obtained even in a noisy environment. When the user approaches the vending machine and issues a voice command, the microphone collects the voice signal and converts it into a digital audio format (such as PCM encoding).

[0056] The collected voice signal is transmitted to the speech recognition module, which converts the voice signal into text information based on deep learning technology (such as recurrent neural network RNN ​​or Transformer architecture). For example, if the user says "I want to buy a bottle of low-sugar drink", the speech recognition module converts it into text: "I want to buy a bottle of low-sugar drink".

[0057] The converted text information is processed by natural language processing to extract keywords and semantic information. The system identifies the core needs of users, such as "low sugar", "beverages", and "purchase" through word segmentation, part-of-speech tagging, and semantic analysis. At the same time, the hot word and sensitive word lists are maintained to optimize recognition accuracy.

[0058] According to the analysis results, structured user demand information is generated. For example, the user's demand is expressed as: {"Intent": "Buy goods", "Keyword": "Low-sugar beverages"}. This information will serve as the basis for subsequent intelligent shopping guides.

[0059] S2. Extract the information of the products on sale in the vending machine, and preprocess and extract features from the information of the products on sale to obtain the product features of each product on sale.

[0060] Among them, the information of the products on sale includes at least the product name, selling points, ingredient list, nutritional components, price, and inventory.

[0061] This step includes:

[0062] Data collection: First, we collect the detailed information of the products on sale in the vending machine, including but not limited to the product name, selling points, ingredient list, nutritional components, price, inventory, etc. This information will be obtained from channels such as product suppliers and e-commerce platforms through manual entry or automated data scraping tools.

[0063] Information standardization: Standardize the collected product information to ensure the consistency of the data format for subsequent data analysis and model training. For example, convert the ingredient list and nutritional components into unified measurement units and formats.

[0064] Feature extraction: Based on the standardized processing of the product information, we extract the features that have a key impact on product recommendation and shopping guide answers, such as selling point keywords, nutritional component ratios, price ranges, etc. These features will be used as the input data for large model training.

[0065] Specifically, in an optional embodiment, extract the detailed information of the products on sale from the product database of the vending machine, including the product name, selling points, ingredient list, nutritional components, price, inventory, etc. This information can be obtained through manual entry or automatically scraped from the data interface provided by the supplier.

[0066] Standardize the collected product information to ensure consistent data formats. For example, unify the measurement units of nutritional components into international standard units (such as "sugar content: 5g / 100ml"), unify the price into currency units (such as "8 yuan"), and standardize the naming of product names.

[0067] Based on the standardized product information, extract the features that have a key impact on intelligent shopping guides. For example:

[0068] Product name: Used to quickly identify products.

[0069] Selling point keywords: Such as "low sugar", "high fiber", "no additives", etc.

[0070] Nutritional component ratios: Such as "sugar content", "fat content", etc.

[0071] Price range: Used to screen products that meet the user's budget.

[0072] Inventory: Ensure that the recommended products are in stock.

[0073] Integrate the extracted features into a feature vector for each product. These features will serve as the input data for the deep learning model.

[0074] S3. After corresponding the user demand information with the product features, use it as training data to train the deep learning model, deploy the trained deep learning model, and conduct intelligent shopping guidance for newly obtained user demand information.

[0075] In this step, select a suitable deep learning model as the basic architecture of the large model according to the features of the product information and the requirements of the shopping guide answer. These models have powerful natural language processing capabilities and context understanding capabilities, and can accurately capture the user's intention and generate logical shopping guide answers. Secondly, use the extracted product information features as training data to feed the large model for training.

[0076] Specifically, in an optional embodiment, perform semantic matching between the user demand information and the product features. For example, if the user's demand is "low-sugar beverage", the system matches this demand with the "selling points" and "sugar content" in the product features to screen out eligible products.

[0077] Generate labeled data as training samples according to the matching results. For example, label the user demand "low-sugar beverage" with the product features of a certain brand of low-sugar beverage to form a training sample:

[0078] Select a suitable deep learning model architecture (such as Transformer or LSTM), and use the generated training data to perform supervised training on the model. During the training process, the model learns the mapping relationship between user demands and product features, and optimizes the parameters to improve the matching accuracy and recommendation effect.

[0079] Deploy the trained deep learning model to the vending machine system. When a new user issues a voice command, the system parses the user's demand through voice recognition and natural language processing, inputs it into the model to generate an intelligent shopping guide result, and feeds it back to the user through a high-resolution display screen and a voice output device.

[0080] The system dynamically generates a recommendation list according to the user's demand. For example, when the user says "Recommend a low-sugar beverage", the system matches eligible products through the model and displays the recommendation result in a voice and visual manner through a virtual shopping guide image, such as "Recommend a certain brand of low-sugar beverage for you, with a sugar content of only 5g / 100ml and a price of 8 yuan".

[0081] In an optional embodiment of this application, the method further includes:

[0082] Using a high-resolution large-screen display and high-quality sound amplification equipment, a vivid and lifelike virtual shopping guide image is created. The core lies in parsing the user's voice commands through speech recognition and natural language processing technologies, and combining with an action control engine and an expression generation algorithm to drive the virtual image to make corresponding reactions in real time. The specific implementation details include: First, the high-resolution display shows a high-definition image of the virtual shopping guide to ensure a realistic visual effect; Second, the high-quality sound amplification equipment outputs natural synthesized speech sounds to enhance the auditory experience. When the user issues a voice command, the system quickly parses the intention, and the virtual shopping guide then makes a matching reaction through a preset action library (such as waving, pointing to the product) and an expression library (such as smiling, concentrating), and at the same time answers the user's questions in voice. This multi-modal interaction method not only realizes the synchronous coordination of voice and vision, but also enhances the authenticity and interest of communication through actions and expressions, thus significantly improving the user's shopping experience and satisfaction.

[0083] Please refer to Figure 2 , which shows a block diagram of an intelligent shopping guide system for a vending machine based on voice interaction provided by an embodiment of the present application. As Figure 2 shown, the system may include:

[0084] A speech recognition and natural language processing module, configured to obtain a user voice command through a voice input device integrated in the vending machine, and perform recognition processing on the user voice command to obtain user demand information;

[0085] By combining speech recognition technology with a vending machine, the system can achieve efficient recognition and accurate response to user voice commands. Using advanced natural language processing technology, the system can parse the user's intention, including but not limited to keyword extraction, semantic understanding, sentiment analysis, etc. To improve the accuracy of speech recognition, the system specifically maintains a hot word library and a sensitive word library, so as to ensure efficient recognition of common vocabulary, while blocking sensitive or prohibited content that should not be recognized, ensuring that the vending machine can make accurate responses according to user needs.

[0086] In the process of applying this technology, several practical problems are specifically solved:

[0087] The problem of sound collection in a noisy environment: By optimizing the microphone array and noise reduction technology, even in a noisy environment, the system can still effectively capture clear voice commands to ensure that the user's needs are responded to in a timely manner.

[0088] Long-distance speech recognition: Using far-field speech recognition technology, users can smoothly issue commands at a certain distance, and the system can accurately capture and recognize these commands without recognition failure due to the distance.

[0089] Priority Recognition in Multi-Person Conversations: When multiple users speak simultaneously, the system can, through sound source localization and voice separation technologies, only recognize the voice commands of the first speaker, avoiding voice interference from multiple people and ensuring the clarity and effectiveness of the interaction process.

[0090] Blocking Sensitive and Prohibited Words, Easier Recognition of Popular Words: During the recognition process, through the maintained popular word library and sensitive word library, the system can accurately block sensitive words and prohibited words, preventing inappropriate information from being recognized and responded to by the system. At the same time, the system has also optimized for common popular words to ensure more efficient recognition and improve the user experience.

[0091] The commodity information extraction and preprocessing module is used to extract the on-sale commodity information of the vending machine, and preprocess and extract features from the on-sale commodity information to obtain the commodity features of each on-sale commodity; among them, the on-sale commodity information includes at least the commodity name, selling points, ingredient list, nutritional components, price, and inventory.

[0092] 1. Data Collection: First, we collect detailed information on the on-sale commodities of the vending machine, including but not limited to commodity names, selling points, ingredient lists, nutritional components, prices, inventory, etc. This information will be obtained from channels such as commodity suppliers and e-commerce platforms through manual entry or automated data scraping tools.

[0093] During the data collection process, it is necessary to ensure the integrity and accuracy of the information and reduce the occurrence of missing values and abnormal data. The commodity information collection formula can be expressed as:

[0094] D={(ID1,Name1,Price1,Ingredients1,Nutrients1,Stock1),(ID2

[0095] ,Name2,Price2,Ingredients2,Nutrients2,Stock2),…}

[0096] Among them, D represents all the commodity data collected, and each commodity consists of multiple features, such as ID, name, price, ingredients, nutritional components, inventory, etc.

[0097] Information Standardization: Standardize the collected commodity information to ensure the consistency of the data format for subsequent data analysis and model training. For example, convert the ingredient list and nutritional components into unified measurement units and formats.

[0098] Unit Standardization: Uniformly convert the weight units of different commodities to grams (g), and the price to the unit price per single item (such as the price per bottle):

[0099] Standardized Price = Raw Price / Unit Conversion Factor

[0100] Among them, Raw Price is the original price, and Unit Conversion Factor is the unit conversion factor.

[0101] 2. Standardization of text information: The ingredient names and nutritional components in the ingredient list need to be in a unified format, redundant or variant names removed, and converted into a unified standard:

[0102] Standardized Ingredients = Normalize(Ingredients)

[0103] Among them, Normalize() refers to the operation of converting the original ingredient list into a standardized format, including removing redundant product information, removing spaces, unifying unit names, etc.

[0104] Feature extraction: Based on the standardized processing of product information, we extract features that have a key impact on product recommendations and shopping guide answers, such as selling point keywords, nutritional component ratios, price ranges, etc. These features will be used as input data for large model training.

[0105] Extraction of selling point keywords: Important keywords in the product description are extracted through an automated data scraping tool. For example, if the selling points of a product are "low-sugar, high-fiber food, no additives", we can extract "low-sugar", "high-fiber", and "no additives" as features:

[0106] Feature Seti = {Keyword1, Keyword2, …, Keywordk}

[0107] Among them, Keyword1, Keyword2, Keywordk are the keywords extracted from the product selling points, and FeatureSet is all the product features extracted.

[0108] Nutritional component ratio: For each product, its nutritional components (such as protein, fat, carbohydrates, etc.) can be represented as a vector. Based on the ingredient list and nutritional components, a standardized nutritional component vector is constructed:

[0109] Nutrientsi = (Pi / Ti, Fi / Ti, Ci / Ti, …)

[0110]

[0111] ​Among them, Pi, Fi, and Ci are the protein, fat, and carbohydrate contents of commodity i respectively, and Ti is the total weight or total energy of commodity i, so that the nutritional components of different commodities can be compared uniformly.

[0112] 3. Price Range: The price range is a very important feature in commodity recommendation and shopping guidance. It can be divided into different ranges according to different prices. For example, if the price of a commodity is between 5 yuan and 10 yuan, it is classified into the "medium price" range:

[0113] Price Rangei = Low if Pricei ≤ 5

[0114] Price Rangei = Medium if 5 < Pricei ≤ 10

[0115] Price Rangei = High if Pricei > 10

[0116] This can help the recommendation system push suitable commodities according to the user's budget.

[0117] The large model training and optimization module is used to correspond the user demand information with the commodity features and use them as training data to train the deep learning model, deploy the trained deep learning model, and conduct intelligent shopping guidance for the newly obtained user demand information.

[0118] According to the characteristics of the commodity information and the requirements of the shopping guidance answer, select a suitable deep learning model as the basic architecture of the large model. These models have powerful natural language processing capabilities and context understanding capabilities, and can accurately capture the user's intention and generate logical shopping guidance answers.

[0119] Secondly, extract the multi-dimensional feature data from the commodity information as training data and input it into the large model for training. These commodity features may include various aspects such as the type, price, brand, nutritional components, and applicable scenarios of the commodity. Through in-depth learning of these data, the model can understand the specific attributes of the commodity and match them with the user's needs and questions to generate more personalized and accurate shopping guidance answers.

[0120] During the training process, use high-quality labeled data sets to ensure that the large model can effectively learn the association between commodity information and user needs. To improve the training effect, adopt techniques such as transfer learning, starting from a pre-trained model and fine-tuning according to specific shopping guidance scenarios. Finally, the trained large model will be able to provide accurate commodity recommendations, answer questions for users in real time, and optimize the user's shopping experience.

[0121] For the specific limitations of the vending machine intelligent shopping guide system based on voice interaction, reference can be made to the limitations of the vending machine intelligent shopping guide method based on voice interaction in the above text, which will not be elaborated here. Each module in the above vending machine intelligent shopping guide system based on voice interaction can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so as to facilitate the processor to call and execute the operations corresponding to each of the above modules.

[0122] In one embodiment, an electronic device is provided. The electronic device can be a computer, and its internal structure diagram can be as Figure 3 shown. The electronic device includes a processor, a memory, and a network interface connected through a system bus. Among them, the processor of the device is used to provide computing and control capabilities. The memory of the device includes a non-volatile storage medium and an internal memory. The non-volatile 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 non-volatile storage medium. The database of the computer device is used for vending machine intelligent shopping guide data based on voice interaction. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements a vending machine intelligent shopping guide method based on voice interaction.

[0123] Those skilled in the art can understand that the structure shown in Figure 3 is only a block diagram of a part of the structure related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have a different component layout.

[0124] In an embodiment of this application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above vending machine intelligent shopping guide method based on voice interaction.

[0125] The computer-readable storage medium provided in this embodiment has the same implementation principle and technical effects as the above method embodiment, which will not be elaborated here.

[0126] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in M forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Symchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0127] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0128] The above embodiments only represent several implementation manners of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A vending machine intelligent shopping guide method based on voice interaction, characterized in that: The method comprises: Acquire user voice commands through a voice input device integrated in the vending machine, and recognize and process the user voice commands to obtain user demand information; Extracting the commodity information on sale from the vending machine, and preprocessing and feature extracting the commodity information on sale to obtain commodity features of each commodity on sale; wherein the commodity information on sale includes at least commodity name, selling point, ingredient list, nutritional content, price, and inventory; The user demand information is matched with the product features and used as training data to train the deep learning model. The trained deep learning model is deployed to perform intelligent shopping guidance for the newly acquired user demand information.

2. The intelligent shopping guide method for vending machines according to claim 1, characterized in that: Acquiring a user's voice command through a voice input device integrated in the vending machine, and performing recognition processing on the user's voice command to obtain user demand information, specifically including: The user's voice commands are acquired through a high-sensitivity microphone integrated in the vending machine, which is converted into text information using speech recognition technology. The text information is parsed through natural language processing technology to extract keywords and semantics to obtain user demand information. At the same time, a list of hot words and sensitive words is maintained.

3. The intelligent shopping guide method for vending machines according to claim 1, characterized in that: The information of the products on sale is preprocessed and features are extracted to obtain product features of each product on sale, including: Standardize the measurement units of ingredient lists and nutritional content to international standard units; Unify commodity prices into monetary units; Standardize product names to ensure the consistency of names of the same product among different suppliers; Further feature extraction is used to obtain keywords in the product selling points, calculate the proportion of nutrients, determine the product price range, and analyze the real-time status of inventory.

4. The intelligent shopping guide method for vending machines according to claim 1, characterized in that: The user demand information is matched with the product features and used as training data to train the deep learning model, specifically including: Through natural language processing technology, keywords, semantic information and constraints in user demand information are extracted, and the parsed user demand information is annotated into a structured data format; wherein the constraints include price range, nutritional requirements, Match the labeled user demand information with product features to generate training samples.

5. The intelligent shopping guide method for vending machines according to claim 4, characterized in that: Match the annotated user demand information with product features, including: Match the keywords in user needs with the keywords in product features; For the constraints in user requirements, check whether the product features meet the constraints; Based on the matching results, generate training samples that annotate the correspondence between user needs and product features; Use the generated matching samples to train the deep learning model so that the model can learn the mapping relationship between user needs and product features; Use the generated matching samples to train the deep learning model, including: Select a deep learning model architecture for natural language processing and recommendation tasks; wherein the deep learning model architecture includes at least a recurrent neural network, a long short-term memory network, and a Transformer architecture; Convert matching samples into an input format acceptable to the model, including converting text information into word embedding vectors; The model is supervised and trained using labeled matching samples to optimize model parameters so that it can generate accurate product recommendations based on user needs.

6. The intelligent shopping guide method for vending machines according to claim 1, characterized in that: The method further comprises: High-definition images of virtual shopping guides are displayed on high-resolution displays, and natural speech synthesis sounds are output using high-quality sound amplification equipment; When the user issues a voice command, the virtual shopping guide immediately responds with preset actions and expressions, and answers the user's questions by voice; wherein the preset actions at least include waving and pointing at the product, and the expressions at least include smiling.

7. A vending machine intelligent shopping guide system based on voice interaction, characterized in that: The system comprises: A speech recognition and natural language processing module, which is used to obtain user voice commands through a voice input device integrated in the vending machine, and to recognize and process the user voice commands to obtain user demand information; A product information extraction and preprocessing module, used to extract the product information on sale from the vending machine, and preprocess and extract features of the product information on sale to obtain product features of each product on sale; wherein the product information on sale includes at least product name, selling point, ingredient list, nutritional content, price, and inventory; The large model training and optimization module is used to match the user demand information with the product features and use them as training data to train the deep learning model, deploy the trained deep learning model, and perform intelligent shopping guidance for the newly acquired user demand information.

8. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images