Chat-type commodity guide method and device, equipment and medium

Abstract

The application relates to a chat-type commodity guide method and device, equipment and medium in the computer technology field, the method comprises the following steps: obtaining chat text in a chat window, identifying keywords in the chat text by using a text recognition model, the keywords comprising description information of commodity characteristics; determining whether a purchase intention expressed by the chat text is clear by using a text classification model; for the chat text expressing a clear purchase intention, retrieving commodity information of a commodity matched with the keywords in the chat text from a commodity database and pushing the commodity information to the chat window; for the chat text expressing a fuzzy purchase intention, determining a preset commodity label matched with the keywords in the chat text by using a commodity recommendation model, obtaining commodity information of a commodity corresponding to the commodity label, and pushing the commodity information to the chat window. The application can push chat-related commodities in a chat process in a timely manner, ensure the efficiency of guide services, and improve shopping satisfaction.

Description

Technical Field

[0001] This application relates to the field of e-commerce technology, and in particular to a chat-based product guide method and its corresponding apparatus, computer equipment, and computer-readable storage medium. Background Technology

[0002] In e-commerce shopping guide scenarios using instant messaging, buyers can communicate with sellers instantly, describing their purchase intentions. Sellers then provide corresponding sales pitches for reference, allowing buyers to efficiently and conveniently find desired products that match their purchase intentions. During this process, buyers can request products from sellers that correspond to their purchase intentions mentioned in the chat for their reference. Sellers, typically possessing extensive knowledge of their store's numerous products, are usually able to identify suitable items. Clearly, this places high demands on sellers; however, traditional technologies currently lack solutions to address this business challenge.

[0003] In view of this, the applicant has long been engaged in research in related fields and has made corresponding explorations to solve industry problems in the e-commerce sector. Summary of the Invention

[0004] The primary objective of this application is to address at least one of the aforementioned problems by providing a chat-based product shopping guide method and corresponding apparatus, computer equipment, and computer-readable storage medium.

[0005] To achieve the various objectives of this application, the following technical solution is adopted:

[0006] A chat-based product recommendation method provided for one of the purposes of this application includes the following steps:

[0007] The chat text in the chat window is obtained, and the keywords in the chat text are identified using a text recognition model. The keywords include descriptive information about product features.

[0008] A text classification model is used to determine whether the purchase intent expressed in the chat text is clear;

[0009] For chat text that clearly expresses a purchase intention, retrieve product information that matches the keywords in the chat text from the product database and push it to the chat window;

[0010] For chat text that expresses a vague purchase intention, a product recommendation model is used to determine the preset product tags that match the keywords in the chat text, obtain the product information of the products corresponding to the product tags, and push them to the chat window.

[0011] In a further embodiment, the step of identifying keywords in the chat text using a text recognition model includes the following steps:

[0012] The chat text is segmented into words to obtain the corresponding segmented text;

[0013] The segmented text is input into a text recognition model to determine the entity identifiers corresponding to each word element in the segmented text.

[0014] Select the words corresponding to entity identifiers used to represent product characteristics as keywords.

[0015] In a further embodiment, the step of retrieving product information from the product database that matches the keywords in the chat text and pushing it to the chat window includes the following steps:

[0016] Based on the keywords in the chat text, a search engine is used to retrieve a set of products that match the keywords from the product database;

[0017] Obtain user behavior data corresponding to each product in the product set, and calculate the ranking score corresponding to each product. The user behavior data includes any one or more of the following: clicks, favorites, add-to-cart, and purchases.

[0018] The product with the highest ranking score is selected from the product set, and its product information is pushed to the chat window.

[0019] In a further embodiment, the step of using a product recommendation model to determine the preset product tags that match the keywords in the chat text includes the following steps:

[0020] The keywords in the chat text are paired with preset product tags to form data pairs, which are then input into the product recommendation model for encoding, thereby obtaining the encoded feature vectors corresponding to the keywords and product tags in the data pairs.

[0021] The encoded feature vectors corresponding to the keywords and product tags are respectively subjected to average pooling, and the similarity between the pooling results of the two is calculated.

[0022] Filter out product tags whose similarity exceeds a preset threshold.

[0023] In a further embodiment, in the step of using a text classification model to determine whether the purchase intention expressed in the chat text is clear, it is determined whether the purchase intention expressed in the chat text is clear or ambiguous, wherein:

[0024] The keywords of the chat text that clearly expresses the intention to purchase include the keywords in the currently identified chat text, as well as the keywords corresponding to the various chat texts that also clearly express the intention to purchase that precede this chat text.

[0025] The keywords corresponding to the chat text that expresses an ambiguous purchase intention include the keywords in the currently identified chat text, as well as the keywords corresponding to the various chat texts that also express an ambiguous purchase intention preceding this chat text.

[0026] In a further embodiment, before obtaining the chat text in the chat window, the following steps are also included:

[0027] A single training sample and its supervision label are obtained from the prepared training set. The training sample includes data pairs consisting of keywords in chat text that express ambiguous purchase intentions and preset product labels. The supervision label characterizes whether the descriptions corresponding to the keywords and product labels in the data pairs of the training sample are similar.

[0028] The training samples are input into the product recommendation model to extract the deep semantic information corresponding to the keywords and product tags in the data pairs and obtain the corresponding encoded feature vectors.

[0029] Average pooling is performed on the encoded feature vectors corresponding to keywords and product labels in the training samples, and the similarity between the two pooling results is calculated.

[0030] Using the supervised labels of the training samples, the loss value of the similarity is determined. When the loss value does not reach the preset threshold, the weights of the product recommendation model are updated, and other training samples are called to carry out iterative training until the model converges.

[0031] In a further embodiment, before obtaining the chat text in the chat window, the following steps are also included:

[0032] Based on the text information of each product in the product database, a clustering algorithm is used to classify each product and label the products in each category with corresponding product tags. The text information includes the product description information.

[0033] On the other hand, a chat-based product guide device provided to meet one of the purposes of this application includes a text acquisition module, an intent determination module, a first push module, and a second push module. The text acquisition module is used to acquire chat text in a chat window and identify keywords in the chat text using a text recognition model. The keywords include descriptive information about product features. The intent determination module is used to determine whether the purchase intent expressed in the chat text is clear using a text classification model. The first push module is used to retrieve product information matching the keywords in the chat text from a product database for chat text expressing a clear purchase intent and push it to the chat window. The second push module is used to determine preset product tags matching the keywords in the chat text using a product recommendation model for chat text expressing an ambiguous purchase intent, acquire product information corresponding to the product tags, and push it to the chat window.

[0034] In a further embodiment, the text acquisition module includes: a text segmentation submodule, used to segment the chat text to obtain the corresponding segmented text; a named entity recognition submodule, used to input the segmented text into a text recognition model to determine the entity identifiers corresponding to each word in the segmented text; and a keyword filtering submodule, used to filter out the word elements corresponding to the entity identifiers used to represent product features as keywords.

[0035] In a further embodiment, the first push module includes: a product retrieval submodule, used to retrieve a set of products matching the keywords in the chat text from a product database using a search engine; a rating calculation submodule, used to obtain user behavior data corresponding to each product in the product set and calculate the ranking rating corresponding to each product, wherein the user behavior data includes any one or more of clicks, favorites, add-to-cart, and purchases; and a product filtering submodule, used to filter the product with the highest ranking rating from the product set and push its product information to the chat window.

[0036] In a further embodiment, the second push module includes: a text encoding submodule, used to form data pairs by combining keywords in the chat text with preset product tags, inputting them into a product recommendation model for encoding, and obtaining encoded feature vectors corresponding to the keywords and product tags in the data pairs; an average pooling submodule, used to perform average pooling on the encoded feature vectors corresponding to the keywords and product tags respectively, and calculate the similarity between the pooling results of the two; and a tag filtering submodule, used to filter out product tags whose similarity exceeds a preset threshold.

[0037] In a further embodiment, the intent determination module determines whether the purchase intent expressed by the chat text is explicit or implicit, wherein: a first keyword submodule is used for chat text expressing explicit purchase intent, the corresponding keywords of which include the keywords in the currently identified chat text, as well as the keywords corresponding to each chat text preceding the current chat text that also expresses explicit purchase intent; a second keyword submodule is used for chat text expressing implicit purchase intent, the corresponding keywords of which include the keywords in the currently identified chat text, as well as the keywords corresponding to each chat text preceding the current chat text that also expresses implicit purchase intent.

[0038] In a further embodiment, before the text acquisition module, the system further includes: a sample acquisition module, used to acquire a single training sample and its supervision label from a prepared training set, wherein the training sample includes data pairs consisting of keywords in chat text expressing ambiguous purchase intentions and preset product labels, and the supervision label characterizes whether the descriptions corresponding to the keywords and product labels in the data pairs of the training sample are similar; a text encoding module, used to input the training sample into the product recommendation model, extract the deep semantic information corresponding to the keywords and product labels in the data pairs, and obtain the corresponding encoded feature vectors; an average pooling module, used to perform average pooling on the encoded feature vectors corresponding to the keywords and product labels in the training sample, and calculate the similarity between the two pooling results; and an iterative training module, used to use the supervision label of the training sample to determine the loss value of the similarity, and when the loss value does not reach a preset threshold, to update the weights of the product recommendation model and continue to call other training samples to perform iterative training until the model converges.

[0039] In a further embodiment, before the text acquisition module, there is also a tag labeling module, which is used to classify each product according to the text information of each product in the product database using a clustering algorithm, and label the corresponding product tags of each category of products, wherein the text information includes product description information.

[0040] In another aspect, a computer device provided for one of the purposes of this application includes a central processing unit and a memory, the central processing unit being used to invoke and run a computer program stored in the memory to perform the steps of the chat-style product guide method described in this application.

[0041] On another aspect, a computer-readable storage medium is provided to suit another purpose of this application, which stores, in the form of computer-readable instructions, a computer program implemented according to the chat-style product guide method, which, when invoked by a computer, executes the steps included in the method.

[0042] The technical solution of this application has many advantages, including but not limited to the following aspects:

[0043] This application employs a text recognition model to identify keywords in chat text entered in the chat window, and a text classification model to determine whether the purchase intent expressed in the chat text is clear. Then, for chat text expressing a clear purchase intent, it retrieves product information matching the keywords in the product database and pushes it to the chat window. For chat text expressing a vague purchase intent, it uses a product recommendation model to determine preset product tags that match the chat text, obtains product information corresponding to the product tags, and pushes it to the chat window. Therefore, this approach achieves intelligent functionality without human intervention, providing suitable products based on the purchase intent expressed in the chat text, ensuring timeliness and high efficiency. It effectively lowers the barrier for sellers to understand the numerous products available online, efficiently assisting sellers in providing high-quality shopping guide services and improving customer satisfaction. Attached Figure Description

[0044] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0045] Figure 1 This is a flowchart illustrating a typical embodiment of the chat-based product shopping guide method of this application;

[0046] Figure 2 This is a flowchart illustrating the process of identifying keywords in chat text in an embodiment of this application;

[0047] Figure 3 This is a schematic diagram illustrating the process of pushing product information of products that match keywords in chat text to the chat window in an embodiment of this application;

[0048] Figure 4 This is a schematic diagram illustrating the process of determining product tags that match keywords in chat text in an embodiment of this application;

[0049] Figure 5 This is a flowchart illustrating the training process of the product recommendation model in the embodiments of this application;

[0050] Figure 6 This is a schematic diagram illustrating the process of labeling different categories of goods in an embodiment of this application;

[0051] Figure 7 This is a schematic diagram of the chat-based product guide device of this application;

[0052] Figure 8 This is a schematic diagram of the structure of a computer device used in this application. Detailed Implementation

[0053] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0054] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0055] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0056] Those skilled in the art will understand that the terms "client," "terminal," and "terminal device" as used herein include both devices that receive wireless signals, devices that only possess wireless signal receiver capabilities without transmission capabilities, and devices with receiving and transmitting hardware, devices that have receiving and transmitting hardware capable of bidirectional communication over a bidirectional communication link. Such devices may include: cellular or other communication devices such as personal computers or tablets, having single-line displays, multi-line displays, or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service) that can combine voice, data processing, fax, and / or data communication capabilities; PDAs (Personal Digital Assistants) that may include radio frequency receivers, pagers, internet / intranet access, web browsers, notebooks, calendars, and / or GPS (Global Positioning System) receivers; and conventional laptops and / or handheld computers or other devices that have and / or include radio frequency receivers. As used herein, "client," "terminal," and "terminal device" can be portable, transportable, installed in a means of transportation (air, sea, and / or land), or suitable and / or configured to operate locally and / or in a distributed manner, operating in any other location on Earth and / or in space. "Client," "terminal," and "terminal device" as used herein can also be a communication terminal, an internet access terminal, or a music / video playback terminal, such as a PDA, a MID (Mobile Internet Device), and / or a mobile phone with music / video playback capabilities, or a smart TV, set-top box, etc.

[0057] The hardware referred to by the names "server," "client," and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer. It is a hardware device with the necessary components revealed by the von Neumann architecture, such as a central processing unit (including an arithmetic logic unit and a control unit), memory, input devices, and output devices. The computer program is stored in its memory, and the central processing unit loads the program stored in the secondary storage into the main memory to run it, execute the instructions in the program, and interact with the input and output devices to complete specific functions.

[0058] It should be noted that the concept of "server" used in this application can also be extended to the case of server clusters. Based on the network deployment principles understood by those skilled in the art, the servers should be logically divided. Physically, these servers can be independent of each other but accessible through interfaces, or they can be integrated into a single physical computer or a computer cluster. Those skilled in the art should understand this flexibility and should not use it to constrain the implementation of the network deployment method in this application.

[0059] One or more of the technical features of this application, unless explicitly specified herein, can be deployed on a server and accessed by a client remotely calling the online service interface provided by the server, or can be directly deployed and run on a client to access the service.

[0060] Unless otherwise specified, the neural network models referenced or potentially referenced in this application may be deployed on a remote server and invoked remotely on the client, or deployed on a client with the capability to invoke directly. In some embodiments, when running on the client, the corresponding intelligence may be acquired through transfer learning in order to reduce the requirements on the client's hardware resources and avoid excessive consumption of the client's hardware resources.

[0061] Unless otherwise specified, all data involved in this application may be stored remotely on a server or on a local terminal device, as long as it is suitable for use by the technical solution of this application.

[0062] Those skilled in the art will understand that although the various methods in this application are described based on the same concept and thus present commonality among them, they can be performed independently unless otherwise specified. Similarly, the various embodiments disclosed in this application are all based on the same inventive concept; therefore, concepts expressed in the same way, as well as concepts that are appropriately changed for convenience but are expressed differently, should be understood equivalently.

[0063] Unless otherwise expressly stated, the various embodiments disclosed in this application can be combined in a cross-cutting manner to flexibly construct new embodiments, as long as such combination does not depart from the inventive spirit of this application and can meet the needs of the prior art or solve a certain deficiency in the prior art. Those skilled in the art should be aware of such modifications.

[0064] The chat-based product shopping guide method of this application can be programmed into a computer program product and deployed on a client or server to run. For example, in the exemplary application scenario of this application, it can be deployed on the server of an e-commerce platform. In this way, the method can be executed by human-computer interaction with the process of the computer program product through a graphical user interface by accessing the interface opened after the computer program product runs.

[0065] Please see Figure 1 The chat-based product shopping guide method of this application, in its typical embodiment, includes the following steps:

[0066] Step S1100: Obtain the chat text in the chat window, and use a text recognition model to identify the keywords in the chat text. The keywords include descriptive information about the product features.

[0067] Generally speaking, cross-border e-commerce platforms provide independent website services, where goods are rented from the platform's independent website, online stores are set up, and product information is listed on the online stores for buyers on the platform to select the products they need.

[0068] Cross-border e-commerce platforms typically set up chat windows for each independent site, with each site assigned its own chat program entry point to facilitate communication between online stores and buyers. Buyers can access the chat window of the corresponding online store through these entry points and chat with the seller. For example, in common e-commerce shopping guide scenarios, buyers can type text describing their needs and send it to the chat window. After reviewing the information, if the seller can directly determine the user's needs, they can push relevant products from the store to the chat window as a response. If they cannot directly determine the user's needs, they can further send text containing descriptions of the product characteristics, such as product attributes, product category, product brand, product name, etc., to the chat window as a response. After reviewing the response, the buyer can determine whether the product characteristics described in the response match their needs and then input corresponding text to send to the chat window. Through this chat, the seller can continuously provide text containing descriptions of product characteristics to the buyer, confirming whether the provided descriptions match the buyer's needs based on the buyer's input. This process is repeated until the seller can finally determine the user's product requirements.

[0069] Accordingly, the chat text in the corresponding chat window can be obtained through a pre-encapsulated data interface, which can be flexibly implemented by those skilled in the art.

[0070] The text recognition model described herein is suitable for named entity recognition tasks. Specific model selections can include BiLSTM+CRF, CNN+CRF, etc. Those skilled in the art can choose one model as needed. After training to convergence, the text recognition model can learn to recognize the entity identifiers corresponding to the entity identifier set of each word in the chat text input to the model. The entity identifier set can be pre-set by those skilled in the art as needed. For example, the entity identifier set includes various entity identifiers representing product characteristics, such as product attributes, product categories, product brands, product names, etc., as well as entity identifiers representing non-product characteristics.

[0071] Therefore, the chat text can be used as input to the text recognition model to identify the entity identifiers corresponding to each word in the chat text, and select the word identifiers corresponding to the entity identifiers representing product features as keywords.

[0072] Step S1200: Use a text classification model to determine whether the purchase intention expressed in the chat text is clear;

[0073] The text classification model described herein is suitable for text classification tasks. Specific model selections may include the BERT model, HAN model, LEAM model, TextCNN, TextRNN, and TextRCNN. Those skilled in the art can select one model to implement as needed.

[0074] Sufficient chat texts expressing clear purchase intentions can be collected, such as "Recommend a dark red lipstick for me, as a gift," or "Do you have a red thermos?", as well as chat texts expressing vague purchase intentions, such as "Baby is afraid of the heat, do you have any suitable summer clothes to recommend?", "What products are suitable for oily skin?", or "I am currently pregnant, do you have any suitable products to recommend?", etc., as training samples. These training samples are labeled with supervisory tags that represent whether the purchase intentions expressed are clear, thus constructing a training set. The text classification model is trained using this training set. During training, a single training sample is input into the model, encoded, and corresponding deep semantic features are extracted to obtain an encoded feature vector. This vector is then mapped to a corresponding classification space. The classification space includes a positive class space representing explicit purchase intent expressed in the chat text, and a negative class space representing ambiguous purchase intent expressed in the chat text. Corresponding classification probabilities are obtained. Based on the supervision label corresponding to the training sample, a loss value for this classification probability is calculated. When the loss value reaches a preset threshold, the text classification model is considered converged, and training terminates. Otherwise, the next training sample is used to continue iterative training of the text classification model. Training terminates once the text model is determined to have converged based on the loss value. It can be understood that after training to convergence, the text classification model learns the ability to identify whether the purchase intent expressed in the input chat text is explicit. The preset threshold can be flexibly set by those skilled in the art based on prior knowledge or experimental data.

[0075] Therefore, the chat text can be used as input to the text classification model trained to convergence. The chat text is encoded, and corresponding deep semantic features are extracted to obtain encoded feature vectors. These vectors are then mapped to the corresponding classification space. The classification space corresponding to the highest classification probability is determined. If the classification space is a positive class space, it indicates that the purchase intention expressed by the chat text is clear; otherwise, it indicates that the purchase intention expressed by the chat text is ambiguous. Furthermore, in this process, the keywords corresponding to chat texts expressing clear purchase intentions include the keywords in the currently identified chat text, as well as the keywords corresponding to other chat texts that also express clear purchase intentions before this chat text.

[0076] The keywords corresponding to the chat text that expresses an ambiguous purchase intention include the keywords in the currently identified chat text, as well as the keywords corresponding to the various chat texts that also express an ambiguous purchase intention preceding this chat text.

[0077] Step S1300: For the chat text that clearly expresses the intention to purchase, retrieve the product information of the products that match the keywords in the chat text from the product database and push it to the chat window;

[0078] It is understandable that each independent website can build and maintain its own product database to store product information corresponding to each product listed in the online store. This product information includes, but is not limited to, textual information and descriptive information about product features. The textual information may include any one or more of the following: product title, product details text, product attributes, category tags, etc.

[0079] A search engine is used to retrieve multiple products from a product database that match the keywords in the chat text. For each product title, the TF-IDF algorithm is used to calculate the TF*IDF corresponding to each keyword as a search score. The products are then sorted according to their search scores from highest to lowest, and the top-ranked products are selected to form a product set. The search engine can be Elasticsearch, which is a distributed search and analytics engine located at the core of the Elastic Stack.

[0080] Furthermore, user behavior data corresponding to each product in the product set is obtained. This user behavior data includes one or more of the following: clicks, favorites, add-to-cart, and purchases. Each user behavior data point is matched with a corresponding preset weight and then summed to calculate the ranking score for each product. The preset weights can be set according to the magnitude of the benefits generated by each user behavior. For example, in the order of Click > Favorite > Add to Cart > Purchase, the specific corresponding weights could be: Clicks: 1, Favorites: 2, Add-to-Cart: 3, Purchases: 4. Those skilled in the art can flexibly preset the weights corresponding to each user behavior as disclosed herein. Thus, the product with the highest ranking score can be filtered from the product set, its product information can be obtained, and it can be packaged into any form of data such as a product page link, product push poster image, or QR code, and pushed to the chat window.

[0081] Step S1400: For chat text that expresses an ambiguous purchase intention, a product recommendation model is used to determine the preset product tags that match the keywords in the chat text, obtain the product information of the product corresponding to the product tags, and push it to the chat window.

[0082] The product recommendation model is suitable for text similarity tasks. The model structure can be a text encoder followed by a similarity calculation module. Specifically, the text encoder can be a deep learning model from the NLP (Natural Language Processing) field, suitable for extracting semantic features from text. Specifically, the open-source framework Sentence Transformers is used, which provides a large number of pre-trained and converged Transformer models, such as BERT, RoBERTa, XLM-RoBERTa, MPNet, etc. Those skilled in the art can choose one to implement as needed. The similarity calculation module is used to calculate the vector distance between the vectorized representations of the semantics of two texts as the similarity, representing the degree of semantic similarity between the two texts.

[0083] Based on the text information of each product in the product database, corresponding product tags can be manually labeled for each product. These product tags are used to establish a connection between the product and the buyer's needs for that product mentioned in the chat text. Therefore, product tags must be set and labeled to cater to the needs of the corresponding buyer. For example, if the product title is "Calendula Plant Toner Oil Control Moisturizing Soothes Redness and Shrinks Pores," the product tags could be "Oily Skin," "Dry Skin," "Large Pores," "Oil Control," "Moisturizing," etc. The text information can include any one or more of the following: product title, product details text, product attributes, category tags, etc.

[0084] It is understood that data pairs consisting of keywords in chat text expressing ambiguous purchase intentions and preset product tags can be collected as training samples. Based on whether the descriptions corresponding to the keywords and product tags in the data pairs of the training samples are similar, the training samples are labeled with supervision tags accordingly. This process is used to train the product recommendation model until it converges, enabling it to learn the ability to determine the similarity between product tags and the descriptions of keywords in chat text expressing ambiguous purchase intentions.

[0085] Accordingly, keywords in the chat text can be paired with various product tags to form multiple data pairs, which are then input into the product recommendation model that has been trained and converged. This determines the similarity between each product tag and the keywords, thereby filtering out product tags with similarity exceeding a preset threshold. The product information corresponding to each product tag is then obtained, packaged into any form of data such as a product page link, product push poster image, or QR code, and pushed to the chat window. The preset threshold can be set as needed by those skilled in the art as disclosed herein.

[0086] As can be seen from the typical embodiments of this application, the technical solution of this application has many advantages, including but not limited to the following aspects:

[0087] This application employs a text recognition model to identify keywords in chat text entered in the chat window, and a text classification model to determine whether the purchase intent expressed in the chat text is clear. Then, for chat text expressing a clear purchase intent, it retrieves product information matching the keywords in the product database and pushes it to the chat window. For chat text expressing a vague purchase intent, it uses a product recommendation model to determine preset product tags that match the chat text, obtains product information corresponding to the product tags, and pushes it to the chat window. Therefore, this approach achieves intelligent functionality without human intervention, providing suitable products based on the purchase intent expressed in the chat text, ensuring timeliness and high efficiency. It effectively lowers the barrier for sellers to understand the numerous products available online, efficiently assisting sellers in providing high-quality shopping guide services and improving customer satisfaction.

[0088] Please see Figure 2 In a further embodiment, step S1100, the step of identifying keywords in the chat text using a text recognition model, includes the following steps:

[0089] Step S1110: Segment the chat text to obtain the corresponding segmented text;

[0090] The chat text can be segmented into its finest granularity using various word segmentation algorithms such as WordPiece, jieba, and n-gram to obtain the corresponding segmented text, which contains each word element from the chat text.

[0091] In one embodiment, a 1-gram segmentation algorithm is used, with a word-capturing window of 1 and a movement step of 1, to segment the chat text and obtain segmented text containing each word element in the chat text.

[0092] Step S1120: Input the segmented text into the text recognition model to determine the entity identifiers corresponding to each word element in the segmented text;

[0093] In one embodiment, a text recognition model implemented using BiLSTM+CRF is employed. The segmented text is input into the word embedding layer of the text recognition model. A pre-trained embedding matrix is ​​used to map each word in the segmented text into a low-dimensional dense vector, obtaining the embedding representation sequence corresponding to the segmented text. BiLSTM (two-layer LSTM) is used to encode the embedding representation sequence, extracting corresponding deep semantic features. One LSTM network calculates the forward latent features corresponding to the embedding representation sequence, and the other LSTM network calculates the backward latent features. The latent features obtained from the two LSTM networks are then concatenated to obtain the encoded feature sequence of the embedding representation sequence. The encoded feature sequence is input into a linear layer and mapped to various classification spaces. Each classification space represents a corresponding entity identifier in the entity identifier set, obtaining a score matrix corresponding to each word in the segmented text. The score matrix represents the scores mapped to each classification space. Based on this, for each word, the entity identifier represented by the classification space with the largest score in its score matrix is ​​determined as the entity identifier corresponding to each word or phrase. Furthermore, the entity identifiers corresponding to each word element in the segmented text are used as input to the CRF layer. Based on the transition matrix learned during training in this CRF layer, the transition score corresponding to each word element in the segmented text is calculated. The transition score represents the probability that the current entity identifier will transition to another entity identifier. Finally, based on the score matrix corresponding to each word element in the segmented text and the transition scores, the path score is calculated. An exemplary formula is as follows:

[0094]

[0095] The word segmentation text is determined by the above formula. Its corresponding path score The sum of scores for each word element is given, and the scores consist of two parts:

[0096] First half The score matrix corresponding to the word output by BiLSTM Decide.

[0097] The second half The transition matrix A corresponding to the word element calculated by CRF is determined, where Indicates from Entity identity transfer to Transfer score of entity identifier.

[0098] The entity identifier sequence with the highest path score is thus determined as the entity identifier sequence corresponding to the segmented text.

[0099] Step S1130: Select the word units corresponding to the entity identifiers used to represent product characteristics as keywords.

[0100] Based on the entity identifier sequence corresponding to the segmented text, the entity identifier corresponding to each word in the segmented text is determined, and the word corresponding to the entity identifier used to represent product features is selected as keywords.

[0101] In this embodiment, a text recognition model is used to accurately identify the entity identifiers corresponding to each word in the chat text. Based on the accurate entity identifiers, entity identifiers representing product characteristics can be determined as keywords, making the process efficient and convenient.

[0102] Please see Figure 3 In a further embodiment, step S1300, retrieving product information from the product database that matches the keywords in the chat text and pushing it to the chat window, includes the following steps:

[0103] Step S1310: Based on the keywords in the chat text, use a search engine to retrieve a set of products that match the keywords from the product database;

[0104] A search engine is used to retrieve multiple products from a product database that match the keywords in the chat text. For each product title, the TF-IDF algorithm is used to calculate the TF*IDF corresponding to each keyword as a search score. The products are then sorted according to their search scores from highest to lowest, and the top-ranked products are selected to form a product set. The search engine can be Elasticsearch, which is a distributed search and analytics engine located at the core of the Elastic Stack.

[0105] Step S1320: Obtain user behavior data corresponding to each product in the product set, and calculate the ranking score corresponding to each product. The user behavior data includes any one or more of the following: clicks, favorites, add-to-cart, and purchases.

[0106] Furthermore, user behavior data corresponding to each product in the product set is obtained. The user behavior data includes any one or more of the following: clicks, favorites, add-to-cart, and purchases. Each user behavior data is matched with a corresponding preset weight and then summed to calculate the ranking score corresponding to each product. The preset weights can be set according to the magnitude of the benefits brought by each user behavior. For example, clicks > favorites > add-to-cart > purchases, the specific corresponding weights can be: clicks: 1, favorites: 2, add-to-cart: 3, purchases: 4. Those skilled in the art can flexibly preset the weights corresponding to each user behavior according to the disclosure herein.

[0107] Step S1330: Select the product with the highest ranking score from the product set, obtain its product information and push it to the chat window.

[0108] Based on the sorting scores from high to low, the products in the product set are sorted accordingly. The product with the highest sorting score is selected, its product information is obtained, and it is packaged into any form of data such as a product page link, product push poster image, or QR code, and then pushed to the chat window.

[0109] In this embodiment, based on the keywords in the product matching chat text, the user behavior data corresponding to the product is referenced, so that the pushed products can attract buyers to a certain extent.

[0110] Please see Figure 4 In a further embodiment, step S1400, which involves using a product recommendation model to determine the preset product tags that match the keywords in the chat text, includes the following steps:

[0111] Step S1410: Combine the keywords in the chat text with preset product tags to form a data pair, input it into the product recommendation model for encoding, and obtain the encoded feature vectors corresponding to the keywords and product tags in the data pair;

[0112] In one embodiment, the product recommendation model is a dual-tower model, which includes two text encoders with shared weights, a pooling layer following each of the two text encoders, and a similarity calculation module following both pooling layers. The two text encoders are both BERT and share weights, and the two pooling layers both perform average pooling.

[0113] The keywords in the chat text are combined with each product tag to form multiple data pairs, which are then input into the product recommendation model that has been trained and converged. For each data pair, one text encoder of the model receives the keywords in the data pair, and another text encoder receives the product tags for encoding. Based on this, the deep semantic features corresponding to the product tags and keywords in each data pair are extracted to obtain the corresponding encoded feature vector.

[0114] Step S1420: Perform average pooling on the encoded feature vectors corresponding to the keywords and product tags respectively, and calculate the similarity between the pooling results of the two.

[0115] The product recommendation model employs two pooling layers to perform average pooling on the encoded feature vectors corresponding to the product tags and keywords, respectively. The two pooling results are then input into a similarity calculation module to calculate the vector distance between the two pooling results as the similarity. The similarity calculation module can be implemented using any of the large-scale vector retrieval engines such as Faiss, ElasticSearch, and Milvus, or it can use any existing algorithm such as cosine similarity, dot product, Manhattan distance, or Euclidean distance.

[0116] Step S1430: Filter out product labels with similarity exceeding a preset threshold.

[0117] Based on the above steps, the similarity between each product tag and the keywords in the chat text can be calculated. It can be understood that the similarity represents the degree of semantic similarity between each product tag and the keywords. Accordingly, product tags with similarity exceeding a preset threshold are selected. The product tags are descriptions that are close to the keywords. The preset threshold can be set as needed by those skilled in the art as disclosed herein.

[0118] In this embodiment, a product recommendation model is used to accurately calculate the similarity between preset product tags and keywords in chat text. This allows for the selection of product tags that closely match the keyword descriptions based on accurate similarity, resulting in efficient and convenient execution.

[0119] Please see Figure 5 In a further embodiment, before step S1100, obtaining the chat text in the chat window, the following steps are also included:

[0120] Step S1010: Obtain a single training sample and its supervision label from the prepared training set. The training sample includes data pairs consisting of keywords in chat text that express ambiguous purchase intentions and preset product labels. The supervision label characterizes whether the descriptions corresponding to the keywords and product labels in the data pairs of the training sample are similar.

[0121] Sufficient chat texts expressing ambiguous purchase intentions across different product categories can be collected. Referring to steps S1110-1130, the text recognition model is used to identify keywords in each chat text. For each chat text, its keywords and preset product tags are combined to form multiple data pairs as training samples. Each training sample is labeled with a supervisory label indicating whether the descriptions corresponding to the keywords and product tags in its data pairs are similar. A training set is constructed using each training sample and its supervisory label. The product categories include, but are not limited to, cosmetics, clothing, electronics, and food.

[0122] Step S1020: Input the training samples into the product recommendation model, extract the deep semantic information corresponding to the keywords and product tags in the data pairs, and obtain the corresponding encoded feature vectors;

[0123] In one embodiment, the product recommendation model is a dual-tower model, which includes two text encoders with shared weights, a pooling layer following each of the two text encoders, and a similarity calculation module following both pooling layers. The two text encoders are both BERT and share weights, and the two pooling layers both perform average pooling.

[0124] Two text encoders using a product recommendation model extract deep semantic information corresponding to keywords and product tags in data pairs, respectively, and obtain corresponding encoded feature vectors. The deep semantic information contains each word in the text. Based on attention interaction in its context, the semantic information obtained is corresponding to the semantic information. Each element in the encoded feature vector represents the semantic representation of the corresponding word in the text.

[0125] Step S1030: Perform average pooling on the encoded feature vectors corresponding to keywords and product labels in the training samples, and calculate the similarity between the two pooling results;

[0126] The product recommendation model employs two pooling layers to perform average pooling on the encoded feature vectors corresponding to keywords and product tags in the training samples, respectively, to obtain two pooling results. Then, a similarity calculation module is used to calculate the similarity between the two pooling results. The similarity calculation module can be implemented using any of the large-scale vector retrieval engines such as Faiss, ElasticSearch, and Milvus, or it can use any existing algorithm such as cosine similarity, dot product, Manhattan distance, or Euclidean distance.

[0127] Step S1040: Using the supervision labels of the training samples, determine the loss value of the similarity. When the loss value does not reach the preset threshold, update the weights of the product recommendation model and continue to call other training samples to carry out iterative training until the model converges.

[0128] The preset cross-entropy loss function is invoked. This function can be flexibly set by those skilled in the art based on prior knowledge or experimental experience. The cross-entropy loss value of the similarity is calculated based on the supervision label of the training samples. When the loss value reaches a preset threshold, it indicates that the product recommendation model has been trained to a convergent state, and the model training can be terminated. If the loss value does not reach the preset threshold, it indicates that the model has not converged. Therefore, the model is updated with gradients based on the loss value. Typically, the weight parameters of each part of the model are corrected through backpropagation to make the model further approach convergence. Then, the next training sample in the training set is called to iteratively train the product recommendation model until the product recommendation model is trained to a convergent state.

[0129] This embodiment reveals the training process of the product recommendation model. After training to convergence, the text classification model learns the ability to accurately calculate the similarity between two texts in the input data pair.

[0130] Please see Figure 6 In a further embodiment, before step S1100, obtaining the chat text in the chat window, the following steps are also included:

[0131] Step S1000: Based on the text information of each product in the product database, a clustering algorithm is used to classify each product and label the corresponding product tags for each category. The text information includes the product description information.

[0132] Based on the text information of each product in the product database, clustering algorithms such as DBSCAN, K-means, Lappass, and PCA can be used to classify the products and distinguish them into different categories. Then, product tags are manually labeled for each category. These tags establish a connection between the product and the buyer's needs mentioned in the chat text. Therefore, product tags must be set to cater to the buyer's needs for the corresponding product. For example, if the product title is "Calendula Plant Toner, Oil Control, Moisturizing, Soothing Redness, and Pore Minimizing," the product tags could be "Oily Skin," "Dry Skin," "Large Pores," "Oil Control," "Moisturizing," etc. The text information can include any one or more of the following: product title, product details text, product attributes, and category tags.

[0133] In this embodiment, a manual annotation method based on an algorithm is used to label the corresponding product tags for each product, which is highly efficient and effective.

[0134] Please see Figure 7This invention provides a chat-based product guidance device to meet one of the purposes of this application. It is a functional embodiment of the chat-based product guidance method of this application. The device includes a text acquisition module 1100, an intent determination module 1200, a first push module 1300, and a second push module 1400. The text acquisition module 1100 is used to acquire chat text in a chat window and identify keywords in the chat text using a text recognition model. The keywords include descriptive information about product features. The intent determination module 1200 is used to determine whether the purchase intent expressed in the chat text is clear using a text classification model. The first push module 1300 is used to retrieve product information matching the keywords in the chat text from a product database for chat text expressing a clear purchase intent and push it to the chat window. The second push module 1400 is used to determine preset product tags matching the keywords in the chat text using a product recommendation model for chat text expressing an ambiguous purchase intent, acquire product information corresponding to the product tags, and push it to the chat window.

[0135] In a further embodiment, the text acquisition module 1100 includes: a text segmentation submodule, used to segment the chat text to obtain the corresponding segmented text; a named entity recognition submodule, used to input the segmented text into a text recognition model to determine the entity identifiers corresponding to each word in the segmented text; and a keyword filtering submodule, used to filter out the word elements corresponding to the entity identifiers used to represent product features as keywords.

[0136] In a further embodiment, the first push module 1300 includes: a product retrieval submodule, used to retrieve a set of products matching the keywords in the chat text from a product database using a search engine; a rating calculation submodule, used to obtain user behavior data corresponding to each product in the product set and calculate the ranking rating corresponding to each product, wherein the user behavior data includes any one or more of clicks, favorites, add-to-cart, and purchases; and a product filtering submodule, used to filter the product with the highest ranking rating from the product set and push its product information to the chat window.

[0137] In a further embodiment, the second push module 1400 includes: a text encoding submodule, used to form data pairs by combining keywords in the chat text with preset product tags, inputting them into a product recommendation model for encoding, and obtaining encoded feature vectors corresponding to the keywords and product tags in the data pairs; an average pooling submodule, used to perform average pooling on the encoded feature vectors corresponding to the keywords and product tags respectively, and calculate the similarity between the pooling results of the two; and a tag filtering submodule, used to filter out product tags whose similarity exceeds a preset threshold.

[0138] In a further embodiment, the intent determination module 1200 determines whether the purchase intent expressed by the chat text is clear or ambiguous, wherein: a first keyword submodule is used for chat text expressing a clear purchase intent, and its corresponding keywords include the keywords in the currently identified chat text, as well as the keywords corresponding to each chat text preceding this chat text that also expresses a clear purchase intent; a second keyword submodule is used for chat text expressing an ambiguous purchase intent, and its corresponding keywords include the keywords in the currently identified chat text, as well as the keywords corresponding to each chat text preceding this chat text that also expresses an ambiguous purchase intent.

[0139] In a further embodiment, before the text acquisition module 1100, the system further includes: a sample acquisition module, used to acquire a single training sample and its supervision label from a prepared training set, wherein the training sample includes data pairs consisting of keywords in chat text expressing ambiguous purchase intentions and preset product labels, and the supervision label characterizes whether the descriptions corresponding to the keywords and product labels in the data pairs of the training sample are similar; a text encoding module, used to input the training sample into a product recommendation model, extract deep semantic information corresponding to the keywords and product labels in the data pairs, and obtain corresponding encoded feature vectors; an average pooling module, used to perform average pooling on the encoded feature vectors corresponding to the keywords and product labels in the training sample, and calculate the similarity between the two pooling results; and an iterative training module, used to use the supervision label of the training sample to determine the loss value of the similarity, and when the loss value does not reach a preset threshold, to update the weights of the product recommendation model and continue to call other training samples to perform iterative training until the model converges.

[0140] In a further embodiment, before the text acquisition module 1100, a tag labeling module is further included, which is used to classify each product according to the text information of each product in the product database using a clustering algorithm, and label the corresponding product tags of each category of products, wherein the text information includes product description information.

[0141] To address the aforementioned technical problems, embodiments of this application also provide computer equipment. For example... Figure 8The diagram shows the internal structure of a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. The computer-readable storage medium stores an operating system, a database, and computer-readable instructions. The database may store a sequence of control information. When the computer-readable instructions are executed by the processor, they enable the processor to implement a chat-based product recommendation method. The processor of the computer device provides computing and control capabilities, supporting the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When these computer-readable instructions are executed by the processor, they enable the processor to execute the chat-based product recommendation method of this application. The network interface of the computer device is used for communication with a terminal. Those skilled in the art will understand that… Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0142] In this embodiment, the processor is used to execute... Figure 7 The system contains the specific functions of each module and its sub-modules, and the memory stores the program code and various data required to execute these modules or sub-modules. The network interface is used for data transmission between the user terminal and the server. In this embodiment, the memory stores the program code and data required to execute all modules / sub-modules in the chat-based product guide device of this application, and the server can call the server's program code and data to execute the functions of all sub-modules.

[0143] This application also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the chat-style product guide method of any embodiment of this application.

[0144] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. This computer program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0145] In summary, this application adopts an intelligent approach that eliminates the need for human intervention. It can efficiently and promptly provide suitable products based on the shopping intentions expressed in the chat text, ensuring the efficiency of the shopping guide service and improving shopping satisfaction.

[0146] Those skilled in the art will understand that the steps, measures, and solutions in the various operations, methods, and processes discussed in this application can be alternated, modified, combined, or deleted. Furthermore, other steps, measures, and solutions in the various operations, methods, and processes discussed in this application can also be alternated, modified, rearranged, decomposed, combined, or deleted. Furthermore, steps, measures, and solutions in the prior art that are similar to those disclosed in this application can also be alternated, modified, rearranged, decomposed, combined, or deleted.

[0147] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A chat-based product recommendation method, characterized in that, Includes the following steps: A single training sample and its supervision label are obtained from the prepared training set. The training sample includes data pairs consisting of keywords in chat text that express ambiguous purchase intentions and preset product labels. The supervision label characterizes whether the descriptions corresponding to the keywords and product labels in the data pairs of the training sample are similar. The training samples are input into the product recommendation model to extract the deep semantic information corresponding to the keywords and product tags in the data pairs and obtain the corresponding encoded feature vectors. Average pooling is performed on the encoded feature vectors corresponding to keywords and product labels in the training samples, and the similarity between the two pooling results is calculated. Using the supervised labels of the training samples, the loss value of the similarity is determined. When the loss value does not reach the preset threshold, the weights of the product recommendation model are updated, and other training samples are called to carry out iterative training until the model converges. The chat text in the chat window is obtained, and the keywords in the chat text are identified using a text recognition model. The keywords include descriptive information about product features. A text classification model is used to determine whether the purchase intent expressed in the chat text is clear; For chat text that clearly expresses a purchase intention, retrieve product information that matches the keywords in the chat text from the product database and push it to the chat window; For chat text that expresses a vague purchase intention, a product recommendation model is used to determine the preset product tags that match the keywords in the chat text, obtain the product information of the products corresponding to the product tags, and push them to the chat window.

2. The chat-based product guidance method according to claim 1, characterized in that, The step of identifying keywords in the chat text using a text recognition model includes the following steps: The chat text is segmented into words to obtain the corresponding segmented text; The segmented text is input into a text recognition model to determine the entity identifiers corresponding to each word element in the segmented text. Select the words corresponding to entity identifiers used to represent product characteristics as keywords.

3. The chat-based product guidance method according to claim 1, characterized in that, The steps of retrieving product information from the product database that matches the keywords in the chat text and pushing it to the chat window include the following: Based on the keywords in the chat text, a search engine is used to retrieve a set of products that match the keywords from the product database; Obtain user behavior data corresponding to each product in the product set, and calculate the ranking score corresponding to each product. The user behavior data includes any one or more of the following: clicks, favorites, add-to-cart, and purchases. The product with the highest ranking score is selected from the product set, and its product information is pushed to the chat window.

4. The chat-based product guidance method according to claim 1, characterized in that, The steps for determining preset product tags that match the keywords in the chat text using a product recommendation model include the following: The keywords in the chat text are paired with preset product tags to form data pairs, which are then input into the product recommendation model for encoding, thereby obtaining the encoded feature vectors corresponding to the keywords and product tags in the data pairs. The encoded feature vectors corresponding to the keywords and product tags are respectively subjected to average pooling, and the similarity between the pooling results of the two is calculated. Filter out product tags whose similarity exceeds a preset threshold.

5. The chat-based product guidance method according to claim 1, characterized in that, In the step of determining whether the purchase intention expressed in the chat text is clear using a text classification model, it is determined whether the purchase intention expressed in the chat text is clear or ambiguous, wherein: The keywords of the chat text that clearly expresses the intention to purchase include the keywords in the currently identified chat text, as well as the keywords corresponding to the various chat texts that also clearly express the intention to purchase that precede this chat text. The keywords corresponding to the chat text that expresses an ambiguous purchase intention include the keywords in the currently identified chat text, as well as the keywords corresponding to the various chat texts that also express an ambiguous purchase intention preceding this chat text.

6. The chat-based product guidance method according to claim 1, characterized in that, Before retrieving the chat text from the chat window, the following steps are also included: Based on the text information of each product in the product database, a clustering algorithm is used to classify each product and label the products in each category with corresponding product tags. The text information includes the product description information.

7. A chat-based product guide device, characterized in that, include: The sample acquisition module is used to acquire a single training sample and its supervision label from the prepared training set. The training sample includes data pairs consisting of keywords in chat text that express ambiguous purchase intentions and preset product labels. The supervision label characterizes whether the descriptions corresponding to the keywords and product labels in the data pairs of the training sample are similar. The text encoding module is used to input the training samples into the product recommendation model, extract the deep semantic information corresponding to the keywords and product tags in the data pairs, and obtain the corresponding encoded feature vectors. The average pooling module is used to perform average pooling on the encoded feature vectors corresponding to keywords and product labels in the training samples, and calculate the similarity between the two pooling results. The iterative training module is used to determine the loss value of the similarity using the supervision labels of the training samples. When the loss value does not reach a preset threshold, the weights of the product recommendation model are updated, and other training samples are called to carry out iterative training until the model converges. The text acquisition module is used to acquire the chat text in the chat window and use a text recognition model to identify the keywords in the chat text. The keywords include descriptive information about product features. The intent determination module is used to determine whether the purchase intent expressed in the chat text is clear using a text classification model; The first push module is used to retrieve product information of products that match the keywords in the chat text that clearly express a purchase intention from the product database and push it to the chat window; The second push module is used to determine preset product tags that match the keywords in the chat text when the purchase intention is ambiguous, using a product recommendation model, obtain the product information of the product corresponding to the product tag, and push it to the chat window.

8. A computer device comprising a central processing unit and a memory, characterized in that, The central processing unit is used to invoke and run a computer program stored in the memory to perform the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, It stores, in the form of computer-readable instructions, a computer program implemented according to any one of claims 1 to 6, which, when invoked by a computer, executes the steps included in the corresponding method.

Images