Retail industry service system based on NLP
By applying an NLP-based service system in the retail industry, analyzing customer willing data and using semantic analysis algorithms to recommend products, the product recommendation error problem caused by inaccurate customer description is solved, and the purchasing experience and recommendation accuracy are improved.
Patent Information
- Application Number
- CN202411002835.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-12-06
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
During retail sales, customers may not be able to accurately describe the product name or purpose, resulting in large errors between the products recommended by the NLP model and the customer's intentions, reducing the customer's purchasing experience.
A retail industry service system based on NLP was designed, including customer intention acquisition module, intention identification module, product matching module, product function recognition module, use matching module and product selection module. The pre-trained NLP model is used to analyze customer intention data. Extract product information and purpose information, and recommend matching products based on semantic analysis algorithms, support voice and text input, and update the model to improve recommendation accuracy.
It improves the customer's purchasing experience, accurately matches recommended products, reduces the difficulty of customers' searches by themselves, and updates the model according to customer preferences, improving the accuracy and personalization of recommendations.
Smart Images

Figure CN119090527A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of retail service technology, and in particular to a retail industry service system based on NLP. Background Art
[0002] The full name of NLP is Natural Language Processing, which means natural language processing in Chinese. It is an important direction in the field of artificial intelligence. It mainly focuses on how computers can accurately and quickly process large natural language corpora. Natural language corpora refer to language learning expressed in real-world languages. It is a comprehensive method of understanding a set of abstract rules from text and the relationship between language and another language.
[0003] In the actual retail sales process, there are customers who may not say the official name of the product during the purchase process, or do not know the name of the product, and only intend to purchase the product. This increases the difficulty of recommending products to customers through NLP models, and can easily lead to large errors between the products recommended by the model and the customer's target products, or the model cannot recognize the customer's intentions, greatly reducing the customer's purchasing experience. Summary of the invention
[0004] The purpose of the present invention is to provide a retail industry service system based on NLP to solve the above-mentioned deficiencies in the prior art.
[0005] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: a retail industry service system based on NLP, comprising a customer intention acquisition module, an intention recognition module, a commodity matching module, a commodity function recognition module, a usage matching module, a commodity selection module, and a model updating module;
[0006] The customer intention acquisition module is used to acquire the customer's commodity purchase intention and obtain customer intention data, wherein the customer's commodity purchase intention can be voice information or text information;
[0007] The intention recognition module is used to analyze the customer intention data based on the NLP model, and extract the product information and usage information therein, wherein the product information is information referring to the product, such as the customer intention data is "I want a bottle of Coke", and the product information is "a bottle of Coke"; the customer intention data is "the toilet is clogged, is there anything that can unclog the toilet", and the usage information is "unclog the toilet"; wherein the NLP model can adopt a pre-trained NLP model;
[0008] The product matching module is used to obtain the product input information of the products on sale and the product information extracted by the intention recognition module, and analyze the product input information matching the input product information based on the NLP model, and recommend the products on sale corresponding to the product input information matching the input product information to the customer to obtain recommended products; for example, if the product information is "a bottle of Coke", and the matching product input information includes "Coca-Cola" and "Pepsi-Cola", then various specifications of "Coca-Cola" and "Pepsi-Cola" are recommended to the customer, including 330ml, 500ml, 1.5L, 2L, etc.;
[0009] The product function identification module is used to analyze the instructions of the products on sale based on the NLP model to obtain the purpose of the products;
[0010] The purpose matching module is used to analyze the purpose of the goods that matches the purpose information in the customer intention data based on the NLP model, and recommend the corresponding goods on sale to the customer to obtain the recommended goods;
[0011] The commodity selection module is used for customers to select the commodity on sale from the recommended commodities to obtain the purchased commodity;
[0012] The commodity selection module is also used to select the purchase quantity of the purchased commodity;
[0013] The model updating module is used to update the NLP model based on purchased goods and corresponding product information, usage information, and recommended goods.
[0014] Furthermore, the customer intention acquisition module includes an audio recognition module and a text acquisition module;
[0015] The audio recognition module is used to obtain the voice information of the customer and convert the voice information of the customer into text information to obtain the customer's intention data; wherein, the audio recognition module can use a microphone to obtain the customer's voice and convert the obtained customer's voice into text information through voice recognition technology; further, a control switch of the microphone can be set to facilitate the customer to control the microphone to turn on through the control switch, and turn on to obtain the customer's voice information when the customer needs it;
[0016] The text acquisition module is used to acquire the text input by the customer and obtain the customer's intention data.
[0017] Further, the product matching module is used to obtain the product input information of the products on sale and the product information extracted by the intention recognition module, and analyze the product input information matching the input product information based on the NLP model, and recommend the products on sale corresponding to the product input information matching the input product information to the customer to obtain the recommended products, including the following steps:
[0018] Obtaining product input information of the product on sale, wherein the product input information includes product name, product specifications and product image;
[0019] Obtain product information extracted by the intent recognition module;
[0020] Analyze the similarity between the product information and the product name based on a semantic analysis algorithm, wherein the semantic analysis algorithm may adopt a word embedding model (Word Embeddings), a semantic similarity calculation method, and the like;
[0021] Filter out product entry information corresponding to product names whose similarity is greater than a set first similarity threshold;
[0022] The filtered product entry information is generated into different options and displayed visually to customers.
[0023] Furthermore, generating different options for the filtered product input information and visually displaying them to the customer includes the following steps:
[0024] Obtain the similarity between the product name and product information in each product entry information;
[0025] Sort the product input information in order of similarity.
[0026] Furthermore, the model updating module is used to update the NLP model based on the purchased goods and the corresponding customer intention data, the product information and the usage information, including the following steps:
[0027] Check whether there is a new purchase item;
[0028] If not, search again; if yes, obtain the purchased product and its corresponding product information and recommended products to obtain a first updated sample;
[0029] Classifying the first update samples according to different commodity information to obtain multiple first update sample sets;
[0030] Determine whether the number of first update samples in each first update sample set is greater than a set first update threshold;
[0031] If so, the order of the recommended commodities matching the corresponding commodity information is updated according to the order of the quantity of each purchased commodity in the first updated sample set, and new recommended commodities are added; in one embodiment, if the purchased commodity is not in the recommended commodities, the purchased commodity and the corresponding input commodity information are marked, and if the number of marks for the same purchased commodity and the corresponding input commodity information is greater than the set mark number threshold, the purchased commodity is added to the recommended commodities of the corresponding commodity information.
[0032] Furthermore, the model updating module is used to update the NLP model based on the purchased goods and the corresponding customer intention data, the product information and the usage information, including the following steps:
[0033] Check whether there is a new purchase item;
[0034] If not, search again; if yes, obtain the purchased product and its corresponding usage information and recommended products to obtain a second updated sample;
[0035] Analyze the similarity of the usage information of each second update sample based on the semantic analysis algorithm, and group the second update samples corresponding to the usage information with similarity greater than a set second similarity threshold into one group to obtain multiple second update sample sets;
[0036] Determine whether the number of second updated samples in each second updated sample set is greater than a set second update threshold;
[0037] If so, according to the order of the quantity of each purchased commodity in the second updated sample set, the order of the recommended commodities corresponding to the commodity information of the purpose matching module is updated, and new recommended commodities are added.
[0038] Furthermore, the product selection module is also used to obtain product entry information of products on sale and generate a list of products on sale for customers to select and purchase products in the list of products on sale, wherein the product entry information of the list of products on sale is sorted by the first letter of the pinyin of the product name, and the list of products on sale supports keyword search to improve the efficiency of customers in finding products.
[0039] Furthermore, the product selection module is also used to identify quantity information in the product information based on the NLP model, obtain the default product quantity, and set the purchase quantity as the default product quantity. For example, if the product information is "two bottles of Coke", after the customer selects and confirms to purchase the product, the purchase quantity of the purchased product will be initially set to two, and the customer can modify it based on the initial setting of two, thereby facilitating the customer's purchasing process.
[0040] 1. Compared with the prior art, the NLP-based retail industry service system provided by the present invention, by setting a customer intention acquisition module, an intention recognition module, a commodity matching module, a commodity function recognition module, a purpose matching module, and a commodity selection module, can match and recommend corresponding commodities and purchase quantities for customers based on the customer's description based on the NLP model, eliminating the need for customers to search for commodities by themselves, thereby improving the customer's purchasing experience.
[0041] 2. Compared with the prior art, the NLP-based retail industry service system provided by the present invention can collect customers' preferences when selecting goods according to the stores and customers using the system by setting up a customer intention acquisition module, an intention recognition module, a product matching module, a product function recognition module, a purpose matching module, a product selection module, and a model update module, and update the NLP model, so that the products recommended by the system and the sorting of the products are more in line with the preferences of store customers, further improving the customer's shopping experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.
[0043] Figure 1 A system structure block diagram provided for an embodiment of the present invention. DETAILED DESCRIPTION
[0044] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.
[0045] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.
[0046] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0047] Example embodiments will be described more fully below with reference to the accompanying drawings, but the example embodiments may be embodied in different forms and should not be construed as limited to the embodiments set forth herein. On the contrary, the purpose of providing these embodiments is to make the present disclosure thorough and complete and to enable those skilled in the art to fully understand the scope of the present disclosure.
[0048] In the absence of conflict, the various embodiments of the present disclosure and the various features therein may be combined with each other.
[0049] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.
[0050] The terms used herein are only used to describe specific embodiments and are not intended to limit the present disclosure. As used herein, the singular forms "a", "an" and "the" are also intended to include the plural forms, unless the context clearly indicates otherwise. It will also be understood that when the terms "comprising" and / or "made of" are used in this specification, the presence of the features, wholes, steps, operations, elements and / or components is specified, but the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof is not excluded.
[0051] The embodiments described herein may be described with reference to plan views and / or cross-sectional views by means of idealized schematic diagrams of the present disclosure. Therefore, the example illustrations may be modified according to manufacturing techniques and / or tolerances. Therefore, the embodiments are not limited to the embodiments shown in the accompanying drawings, but include modifications of the configurations formed based on the manufacturing process. Therefore, the regions illustrated in the accompanying drawings have schematic properties, and the shapes of the regions shown in the figures illustrate the specific shapes of the regions of the elements, but are not intended to be limiting.
[0052] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted as having an idealized or overly formal meaning unless explicitly defined as such herein.
[0053] See also Figure 1 , a retail industry service system based on NLP, including a customer intention acquisition module, an intention recognition module, a commodity matching module, a commodity function recognition module, a usage matching module, a commodity selection module, and a model update module;
[0054] The customer intention acquisition module is used to acquire the customer's willingness to purchase goods and obtain customer intention data, wherein the customer's willingness to purchase goods can be voice information or text information; the customer intention acquisition module includes an audio recognition module and a text acquisition module; the audio recognition module is used to acquire the customer's voice information, and convert the customer's voice information into text information to obtain customer intention data; wherein, the audio recognition module can use a microphone to acquire the customer's voice, and convert the acquired customer's voice into text information through voice recognition technology; further, a control switch of the microphone can be set to facilitate the customer to control the microphone to turn on through the control switch, and turn on the microphone to obtain the customer's voice information when the customer needs it; the text acquisition module is used to acquire the text entered by the customer and obtain customer intention data.
[0055] The intention recognition module is used to analyze the customer intention data based on the NLP model, and extract the product information and usage information therein, wherein the product information is information referring to the product, such as the customer intention data is "I want a bottle of Coke", and the product information is "a bottle of Coke"; the customer intention data is "the toilet is clogged, is there anything that can unclog the toilet", and the usage information is "unclog the toilet"; wherein the NLP model can adopt a pre-trained NLP model;
[0056] The product matching module is used to obtain the product input information of the products on sale and the product information extracted by the intention recognition module, and analyze the product input information matching the input product information based on the NLP model, and recommend the products on sale corresponding to the product input information matching the input product information to the customer to obtain the recommended products; for example, if the product information is "a bottle of Coke", and the matching product input information includes "Coca-Cola" and "Pepsi-Cola", then various specifications of "Coca-Cola" and "Pepsi-Cola" are recommended to the customer, including 330ml, 500ml, 1.5L, 2L, etc., including the following steps:
[0057] A1. Obtain product input information of the products on sale, including product name, product specifications and product images;
[0058] A2. Obtain product information extracted by the intent recognition module;
[0059] A3. Analyze the similarity between product information and product name based on semantic analysis algorithm, where the semantic analysis algorithm can adopt word embedding model (Word Embeddings), semantic similarity calculation and other methods;
[0060] Based on semantic analysis, the following steps can be taken:
[0061] (1) Data preprocessing: Convert the raw text data into a usable format, such as Tokenization, Lowercasing, etc.
[0062] (3) Word embedding: Use a pre-trained word embedding model (such as Word2Vec, GloVe, etc.) to convert vocabulary into high-dimensional vectors. The following formula can be used:
[0063] v w =f(w);
[0064] Among them, v w is the vector representation of the vocabulary w, and f(·) is a mapping function, such as Word2Vec, GloVe, etc.
[0065] (4) Build an RNN model: Use a recurrent neural network (RNN) to process sequence data and capture the contextual information of the language.
[0066] (5) Adding self-attention mechanism: Through the self-attention mechanism, the model can adaptively focus on different vocabulary and context information, which can be expressed as follows:
[0067]
[0068] Where Q is the query vector, K is the keyword vector, T is the number of matrix K, V is the value vector, and d k is the dimension of the keyword vector;
[0069] (6) Model training: Use a large-scale corpus to train the model to optimize the model’s semantic understanding capabilities.
[0070] (7) Evaluate the model: Use the test dataset to evaluate the performance of the model and make corresponding adjustments and optimizations.
[0071] A4. Filter out product entry information corresponding to product names whose similarity is greater than a set first similarity threshold;
[0072] A5. Generate different options for the selected product input information and display them visually to customers, including the following steps:
[0073] A5.1. Obtain the similarity between the product name and product information in the input information of each product;
[0074] A5.2. Sort the product input information in order of similarity.
[0075] The product function identification module is used to analyze the instructions of the products on sale based on the NLP model to obtain the purpose of the products;
[0076] The purpose matching module is used to analyze the purpose of the product that matches the purpose information in the customer intention data based on the NLP model, and recommend the corresponding products on sale to the customer to obtain the recommended products;
[0077] The commodity selection module is used for customers to select the goods on sale from the recommended goods to purchase, and obtain the purchased goods; the commodity selection module is also used to select the purchase quantity of the purchased goods; the commodity selection module is also used to obtain the commodity input information of the goods on sale, and generate a list of goods on sale for customers to select and obtain the purchased goods in the list of goods on sale, wherein the commodity input information of the list of goods on sale is sorted by the first letter of the pinyin of the commodity name, and the list of goods on sale supports keyword search to improve the efficiency of customers finding goods. The commodity selection module is also used to identify the quantity information in the commodity information based on the NLP model, obtain the default commodity quantity, and set the purchase quantity to the default commodity quantity. For example, if the commodity information is "two bottles of Coke", after the customer selects and confirms the purchase of the commodity, the purchase quantity of the purchased commodity is initially set to two, and the customer can modify it based on the initial setting of two, which provides convenience for the customer's purchase process.
[0078] The model update module is used to update the NLP model based on the purchased goods and the corresponding product information, usage information, and recommended goods, including the following steps:
[0079] B1. Check whether there is any new purchased product;
[0080] B2. If not, search again. If yes, obtain the purchased product and its corresponding product information and recommended products to obtain the first updated sample;
[0081] B3. Classify the first update samples according to different commodity information to obtain multiple first update sample sets;
[0082] B4. Determine whether the number of first update samples in each first update sample set is greater than a set first update threshold;
[0083] B5. If so, update the order of recommended commodities that match the corresponding commodity information according to the order of quantity of each purchased commodity in the first updated sample set, and add recommended commodities. In one embodiment, if the purchased commodity is not in the recommended commodities, the purchased commodity and the corresponding input commodity information are marked. If the number of marks for the same purchased commodity and the corresponding input commodity information is greater than the set mark number threshold, the purchased commodity is added to the recommended commodities of the corresponding commodity information.
[0084] The following steps are also included:
[0085] C1. Check whether there is any new purchased product;
[0086] C2. If not, search again. If yes, obtain the purchased goods and their corresponding usage information and recommended goods to obtain the second updated sample;
[0087] C3. Analyze the similarity of the usage information of each second update sample based on a semantic analysis algorithm, and group the second update samples corresponding to the usage information having a similarity greater than a set second similarity threshold into one group, to obtain a plurality of second update sample sets;
[0088] C4. Determine whether the number of second update samples in each second update sample set is greater than a set second update threshold;
[0089] C5. If yes, update the order of the recommended commodities corresponding to the commodity information of the purpose matching module according to the order of the quantity of each purchased commodity in the second updated sample set, and add new recommended commodities.
[0090] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.
Claims
1. A retail industry service system based on NLP, characterized by: It includes customer intention acquisition module, intention recognition module, product matching module, product function recognition module, purpose matching module, product selection module and model update module; The customer intention acquisition module is used to acquire the customer's commodity purchase intention and obtain customer intention data; The intention recognition module is used to analyze the customer intention data based on the NLP model and extract the product information and usage information therein; The product matching module is used to obtain the product input information of the products on sale and the product information extracted by the intention recognition module, and analyze the product input information matching the input product information based on the NLP model, and recommend the products on sale corresponding to the product input information matching the input product information to the customer to obtain recommended products; The product function identification module is used to analyze the instructions of the products on sale based on the NLP model to obtain the purpose of the products; The purpose matching module is used to analyze the purpose of the goods that matches the purpose information in the customer intention data based on the NLP model, and recommend the corresponding goods on sale to the customer to obtain the recommended goods; The commodity selection module is used for customers to select the commodity on sale from the recommended commodities to obtain the purchased commodity; The commodity selection module is also used to select the purchase quantity of the purchased commodity; The model updating module is used to update the NLP model based on purchased goods and corresponding product information, usage information, and recommended goods.
2. The NLP-based retail industry service system according to claim 1, characterized in that: The customer intention acquisition module includes an audio recognition module and a text acquisition module; The audio recognition module is used to obtain the voice information of the customer and convert the voice information of the customer into text information to obtain the customer's intention data; The text acquisition module is used to acquire the text input by the customer and obtain the customer's intention data.
3. The NLP-based retail industry service system according to claim 1, characterized in that: The product matching module is used to obtain the product input information of the products on sale and the product information extracted by the intention recognition module, and analyze the product input information matching the input product information based on the NLP model, and recommend the products on sale corresponding to the product input information matching the input product information to the customer to obtain the recommended products, including the following steps: Obtaining product input information of the product on sale, wherein the product input information includes product name, product specifications and product image; Obtain product information extracted by the intent recognition module; Analyzing the similarity between the product information and the product name based on a semantic analysis algorithm; Filter out product entry information corresponding to product names whose similarity is greater than a set first similarity threshold; The filtered product entry information is generated into different options and displayed visually to customers.
4. The NLP-based retail industry service system according to claim 3, characterized in that: The step of generating different options for the filtered product input information and visually displaying them to the customer includes the following steps: Obtain the similarity between the product name and product information in each product entry information; Sort the product input information in order of similarity.
5. The NLP-based retail industry service system according to claim 1, characterized in that: The model updating module is used to update the NLP model based on the purchased goods and the corresponding customer intention data, product information and usage information, including the following steps: Check whether there is a new purchase item; If not, search again; if yes, obtain the purchased product and its corresponding product information and recommended products to obtain a first updated sample; Classifying the first update samples according to different commodity information to obtain multiple first update sample sets; Determine whether the number of first update samples in each first update sample set is greater than a set first update threshold; If so, according to the order of the quantity of each purchased commodity in the first updated sample set, the order of the recommended commodities matching the corresponding commodity information is updated, and new recommended commodities are added.
6. The NLP-based retail industry service system according to claim 1, characterized in that: The model updating module is used to update the NLP model based on the purchased goods and the corresponding customer intention data, product information and usage information, including the following steps: Check whether there is a new purchase item; If not, search again; if yes, obtain the purchased product and its corresponding usage information and recommended products to obtain a second updated sample; Analyze the similarity of the usage information of each second update sample based on the semantic analysis algorithm, and group the second update samples corresponding to the usage information with similarity greater than a set second similarity threshold into one group to obtain multiple second update sample sets; Determine whether the number of second updated samples in each second updated sample set is greater than a set second update threshold; If so, according to the order of the quantity of each purchased commodity in the second updated sample set, the order of the recommended commodities corresponding to the commodity information of the purpose matching module is updated, and new recommended commodities are added.
7. The NLP-based retail industry service system according to claim 1, characterized in that: The commodity selection module is also used to obtain commodity input information of commodities on sale and generate a list of commodities on sale for customers to select commodities to purchase from the list of commodities on sale.
8. The NLP-based retail industry service system according to claim 1, characterized in that: The product selection module is also used to identify quantity information in the product information based on the NLP model, obtain a default product quantity, and set the purchase quantity as the default product quantity.
Citation Information
Patent Citations
Commodity recommending method and commodity recommending system based on intelligent terminal
CN111316306A
Internet-based commodity sales recommendation system and method
CN112184399A
Shopping recommendation method and online shopping platform
CN113988970A
Commodity information display method and display system
CN117541355A
