Picture generation method and device

Abstract

The invention discloses a picture generation method and device, and relates to the technical field of computers. A specific embodiment of the method comprises the following steps: extracting a user question and response information corresponding to the user question from session data of an article; matching the user question with each preset appeal category to determine the appeal category to which the user question belongs; and generating an introduction picture of the article according to the appeal category to which the user question belongs and the response information corresponding to the appeal category. According to the embodiment, on one hand, manpower can be liberated, article maintenance and shop operation efficiency can be improved, on the other hand, the purchase decision point of the user can be timely acquired and the introduction picture of the article can be updated according to the recent session data of the user, so that the introduction picture of the article is closer to the purchase decision point of the user, and the transaction conversion rate is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and device for generating an image. Background Art

[0002] In the field of e-commerce, the introduction pictures of items have the function of introducing item information and matching transactions. For example, merchants use the main picture of the item to introduce the color, style, main functions, marketing information, etc. of the item, and use the detailed picture of the item to introduce the detailed parameters and specifications of the item. The above information is an important reference for users to choose items. At present, the introduction pictures of items are all maintained manually by merchants.

[0003] In the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art:

[0004] Manually maintaining the introductory images of items is labor-intensive and inefficient, and it is difficult to obtain users' purchase decision points in a timely manner. The introductory images of items are difficult to meet users' demands. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method and device for generating pictures, which can, on the one hand, liberate manpower and improve the efficiency of item maintenance and store operation; on the other hand, can timely obtain the user's purchase decision point and update the introduction picture of the item according to the user's recent session data, so that the introduction picture of the item is closer to the user's purchase decision point, thereby improving the transaction conversion rate.

[0006] To achieve the above object, according to one aspect of an embodiment of the present invention, a method for generating an image is provided, comprising:

[0007] Extract user questions and their corresponding answer information from the item's session data;

[0008] Matching the user question with each preset demand category to determine the demand category to which the user question belongs;

[0009] According to the demand category to which the user's question belongs and its corresponding answer information, an introduction picture of the item is generated.

[0010] Optionally, extracting user questions and their corresponding answer information from the conversation data of the item includes: converting the conversation data into a conversation text vector, determining the grammatical structure of the conversation text vector through syntactic analysis, and extracting user questions and their corresponding answer information from the conversation text vector according to the grammatical structure of the conversation text vector.

[0011] Optionally, before matching the user question with each preset demand category, the method further includes at least one of the following:

[0012] Determining the semantic similarity between the user question and the corresponding answer information, and determining whether the semantic similarity between the user question and the corresponding answer information meets a preset matching condition;

[0013] The conversation data is subjected to sentiment analysis to determine the user's satisfaction with the response information, and to determine whether the user's satisfaction with the response information meets a preset matching condition.

[0014] Optionally, an introduction picture of the item is generated according to the demand category to which the user question belongs and its corresponding answer information, including: determining the distribution information of each preset demand category according to the demand category to which each user question belongs and its corresponding answer information; determining a core demand category from the preset demand categories according to the distribution information of each preset demand category; and generating an introduction picture of the item according to the core demand category and its corresponding answer information.

[0015] Optionally, generating an introduction picture of the item includes: generating a text description picture of the item according to the determined demand category and its corresponding response information; and fusing the text description picture with the item picture of the item to obtain an introduction picture of the item.

[0016] Optionally, a text description image of the object is generated based on the determined claim category and its corresponding response information, including: inputting the determined claim category and its corresponding response information into a pre-trained conditional generative adversarial network so that the conditional generative adversarial network outputs a text description image of the object.

[0017] Optionally, merging the text description picture with the item picture of the item includes: overlaying the item picture on the text description picture and adjusting the transparency and / or position of the item picture, or placing the text description picture at a preset position of the item picture.

[0018] According to a second aspect of an embodiment of the present invention, there is provided a device for generating a picture, including:

[0019] The extraction module extracts user questions and their corresponding answer information from the item's session data;

[0020] A matching module, matching the user question with various preset demand categories to determine the demand category to which the user question belongs;

[0021] The generation module generates an introduction picture of the item according to the demand category to which the user's question belongs and the corresponding response information.

[0022] Optionally, the extraction module extracts user questions and their corresponding answer information from the conversation data of the item, including: converting the conversation data into a conversation text vector, determining the grammatical structure of the conversation text vector through syntactic analysis, and extracting user questions and their corresponding answer information from the conversation text vector according to the grammatical structure of the conversation text vector.

[0023] Optionally, the matching module is further used for at least one of the following:

[0024] Before matching the user question with each preset demand category, determining the semantic similarity between the user question and the corresponding answer information, and determining whether the semantic similarity between the user question and the corresponding answer information meets the preset matching condition;

[0025] Before matching the user question with each preset demand category, sentiment analysis is performed on the session data to determine the user's satisfaction with the response information and to determine whether the user's satisfaction with the response information meets preset matching conditions.

[0026] Optionally, the generating module generates an introduction picture of the item according to the demand category to which the user's question belongs and the corresponding answer information, including:

[0027] Determine the distribution information of each preset demand category according to the demand category to which each user question belongs and the corresponding answer information;

[0028] According to the distribution information of each preset demand category, determining a core demand category from each preset demand category;

[0029] According to the core demand category and its corresponding response information, an introduction picture of the item is generated.

[0030] Optionally, the generation module generates an introduction picture of the item, including: generating a text description picture of the item according to the determined demand category and its corresponding response information; and fusing the text description picture with the item picture of the item to obtain an introduction picture of the item.

[0031] Optionally, the generation module generates a text description picture of the object based on the determined claim category and its corresponding response information, including: inputting the determined claim category and its corresponding response information into a pre-trained conditional generative adversarial network so that the conditional generative adversarial network outputs the text description picture of the object.

[0032] Optionally, the generation module merges the text description picture with the item picture of the item, including: overlaying the item picture on the text description picture and adjusting the transparency and / or position of the item picture, or placing the text description picture at a preset position of the item picture.

[0033] According to a third aspect of an embodiment of the present invention, there is provided an electronic device for generating a picture, including:

[0034] one or more processors;

[0035] a storage device for storing one or more programs,

[0036] When the one or more programs are executed by the one or more processors, the one or more processors implement the method provided in the first aspect of the embodiment of the present invention.

[0037] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided. When the program in the computer-readable medium is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

[0038] One embodiment of the above invention has the following advantages or beneficial effects: by parsing the conversation data of the item, the demand category that the user is concerned about can be determined, and by generating an introduction picture of the item based on the demand category and its corresponding response information, on the one hand, manpower can be liberated and the efficiency of item maintenance and store operation can be improved; on the other hand, the user's purchase decision point can be obtained in time and the introduction picture of the item can be updated based on the user's recent conversation data, so that the introduction picture of the item is closer to the user's purchase decision point and the transaction conversion rate is improved.

[0039] The further effects of the above-mentioned non-conventional optional manner will be described below in conjunction with the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings are used to better understand the present invention and do not constitute an improper limitation of the present invention.

[0041] Figure 1 is a schematic diagram of the main process of the method for generating an image according to an embodiment of the present invention;

[0042] Figure 2 is a schematic diagram of distribution information of various preset demand categories in an optional embodiment of the present invention;

[0043] Figure 3 is a schematic diagram of the main process of the method for generating an image according to an embodiment of the present invention;

[0044] Figure 4 is a schematic diagram of main modules of a device for generating a picture according to an embodiment of the present invention;

[0045] Figure 5 is an exemplary system architecture diagram to which embodiments of the present invention may be applied;

[0046] Figure 6 It is a schematic diagram of the structure of a computer system of a terminal device or a server suitable for implementing an embodiment of the present invention. DETAILED DESCRIPTION

[0047] The following is a description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0048] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, storage and other aspects of user personal information involved in the technical solution of this disclosure are in compliance with the provisions of relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken for user personal information to prevent illegal access to user personal information data and maintain the security of user personal information, network security and national security. For example: After collecting session data, we will de-identify the data through technical means.

[0049] According to one aspect of an embodiment of the present invention, a method for generating a picture is provided.

[0050] Figure 1 is a schematic diagram of the main process of the method for generating an image according to an embodiment of the present invention, such as Figure 1 As shown, the method for generating a picture includes step S101, step S102 and step S103.

[0051] Step S101: extract user questions and their corresponding answer information from the item's session data.

[0052] Session data refers to the conversation data generated when users consult customer service. For example, when users purchase milk powder, they can see the applicable age group, brand, taste, price, preferential policies and other information of the milk powder from the product details page. However, if users want to know in depth whether the milk powder is suitable for the baby's physique, they can establish a conversation with a professional nanny through the health consultation service to get answers.

[0053] User questions refer to questions raised by users in a conversation, and response information refers to answers given by customer service to user questions. For example, in a conversation, a user asked the question "Is this milk powder suitable for babies during the milk-changing period?", and customer service gave an answer "It is very suitable. You can use the new and old mixing method to change milk. The operation method is to use milk powder for three days and then milk powder for three days, and repeat it for 2 to 3 times."

[0054] In actual application, the texts that are questions among the texts sent by the user in the conversation can be regarded as user questions. For example, the content of a conversation between a user and customer service is as follows. In the following conversation, the user sends two texts "Hello" and "Is this milk powder suitable for babies during the milk change period?", of which the latter text is a question and can be regarded as a user question, and the corresponding customer service response content is regarded as the response information:

[0055] User: Hello.

[0056] Customer Service: Hello, how can I help you?

[0057] User: Is this milk powder suitable for babies during the milk transition period?

[0058] Customer Service: It is very suitable. You can use the new and old mixing method to change milk. The operation method is to use stage one milk powder for three days and then stage two milk powder for three days, and repeat 2 to 3 times.

[0059] Those skilled in the art can also extract user questions and their corresponding answer information from the conversation data of the item through semantic analysis. Of course, user questions and their corresponding answer information can also be extracted from the conversation data of the item through syntactic analysis, including: converting the conversation data into a conversation text vector, determining the grammatical structure of the conversation text vector through syntactic analysis, and extracting user questions and their corresponding answer information from the conversation text vector according to the grammatical structure of the conversation text vector. Exemplarily: For the question sentence "Is this milk powder suitable for babies in the milk-changing period?", syntactic analysis can identify "this milk powder" as the subject, "suitable" as the predicate, "take" as the object, and "babies in the milk-changing period" as a specific group of people. Through syntactic analysis, on the one hand, it helps to understand the questions raised by users and the answers given by customer service, and on the other hand, it can also better understand user needs, so that customer service can give more accurate answers.

[0060] Step S102: Match the user question with each preset demand category to determine the demand category to which the user question belongs.

[0061] The embodiment of the present invention pre-sets multiple appeal categories, and the division method of the appeal categories and the number of appeal categories can be selectively set according to actual conditions. Taking a certain milk powder as an example, the preset appeal categories may include: product consultation, feeding problems, health consultation, milk change problems, special population product consultation, product recommendation (such as age-appropriate recommendation), nutritional supplements, etc. In the actual application process, in order to more accurately analyze the issues of concern to users, multiple levels of appeal categories can be set. For example, the appeal category corresponding to the above-mentioned milk powder is set as a first-level appeal category, and the second-level appeal category is further refined under the first-level appeal category. Exemplarily, the product consultation category can be further subdivided into subcategories such as ingredients, comparison, applicable population, usage and dosage, adverse reactions, etc., the health consultation category can be further subdivided into subcategories such as eczema, diarrhea, side learning, allergies, jaundice, etc., and the special population product consultation category can be further subdivided into subcategories such as lactose intolerance, allergic infants, favism, etc., and product recommendations (age-appropriate recommendations).

[0062] After extracting the user question, the embodiment of the present invention does not directly generate an introduction picture of the item based on the user question and its corresponding answer content, but first matches the user question with each preset demand category and generates an introduction picture of the item based on the determined demand category. This can reduce the workload of generating the introduction picture and improve the efficiency of generating the introduction picture.

[0063] Optionally, before matching the user question with each preset demand category, the method further includes at least one of the following: (1) determining the semantic similarity between the user question and its corresponding response information, and determining whether the semantic similarity between the user question and its corresponding response information satisfies a preset matching condition; (2) performing sentiment analysis on the session data to determine the user's satisfaction with the response information, and determining whether the user's satisfaction with the response information satisfies the preset matching condition.

[0064] Semantic similarity is used to measure the semantic similarity between different texts, that is, it can determine the degree of similarity between the meanings of two sentences or phrases. This is very useful for understanding user needs, matching related questions and answers, and generating personalized recommendations. For example: for the user question "Is this milk powder suitable for babies during the milk-changing period?", the similarity can be calculated to find the relevant response information, such as an answer "It is very suitable. You can use the new and old mixing method to change milk. The operation method is to consume three days of stage one milk powder and then three days of stage two milk powder, and repeat 2 to 3 times." By comparing the semantic similarity between the user's question and the response information, it can be determined whether the customer service's answer matches the user's question. The embodiment of the present invention can pre-set matching conditions corresponding to the semantic similarity, such as the semantic similarity is greater than or equal to the preset semantic similarity threshold.

[0065] Sentiment analysis is used to analyze the emotions and emotional tendencies conveyed in the text, which helps to understand the user's opinions, emotions, etc., and helps generate answers or recommendations that are more suitable for user needs. For example, the user's comment "Very good, thank you for your patient answer!" can be identified through sentiment analysis to have a positive emotional tendency, thereby further understanding the user's satisfaction with the item and providing the user with more accurate feedback or recommendations. The embodiment of the present invention can pre-set matching conditions corresponding to the sentiment analysis, such as satisfaction being greater than or equal to a preset satisfaction threshold, satisfaction indicating that the user has a positive emotional tendency, etc.

[0066] By screening the question and answer data (a user question and its corresponding answer information is a question and answer data) according to semantic similarity and satisfaction, the user's purchase decision point can be further accurately understood. In the actual application process, the question and answer data whose satisfaction meets the preset matching conditions can be screened out, and classified through step S102 to determine the corresponding demand category; the question and answer data whose semantic similarity meets the preset matching conditions can also be screened out, and classified through step S102 to determine the corresponding demand category; the question and answer data whose satisfaction and semantic similarity both meet the preset matching conditions can also be screened out, and classified through step S102 to determine the corresponding demand category.

[0067] Step S103: Generate an introduction picture of the item according to the demand category to which the user's question belongs and its corresponding response information.

[0068] In the embodiment of the present invention, analysis can be performed on a session data. For example, for a session that is currently being conducted or has been conducted by a certain user, the user's question and the corresponding answer information are extracted from it, the demand category to which the user's question belongs is determined, and the introduction picture of the corresponding item is generated according to the demand category to which the user's question belongs and the corresponding answer information, so that the introduction picture of the item can be updated in a personalized manner according to the consultation content of the user. When the session data in step S101 comes from the user's ongoing session, the generated introduction picture of the item can also reflect the user's real-time needs, so as to be closer to the user's purchase decision point.

[0069] Multiple user question and answer data may be extracted from one session data. In addition, the embodiment of the present invention can also analyze multiple session data of one user or multiple session data of multiple users, so the extracted question and answer data may also be multiple. When multiple user question and answer data are extracted, an introduction picture of the item is generated according to the demand category to which the user question belongs and its corresponding answer information, including: determining the distribution information of each preset demand category according to the demand category to which each user question belongs and its corresponding answer information; determining the core demand category from each preset demand category according to the distribution information of each preset demand category; generating an introduction picture of the item according to the core demand category and its corresponding answer information.

[0070] When determining the core demand category from the various preset demand categories, several demand categories with a higher magnitude or a demand category with a magnitude greater than a preset threshold can be selected. The magnitude mentioned here refers to the demand category with more corresponding user questions. Exemplarily, after determining the demand category to which the user question belongs, the number of user questions corresponding to the corresponding demand category is increased by 1. After all user questions have been classified, the total number of user questions or the proportion of user questions corresponding to the demand category can be regarded as the magnitude of the demand category. The embodiment of the present invention processes a large number of user questions under the items and classifies the questions with strong user consultation demands, so as to capture the core issues, that is, to mine the core demands. The embodiment of the present invention can also reflect the distribution information of each preset demand category in the form of a graph. Figure 2 : This is a schematic diagram of the distribution information of each preset demand category in an optional embodiment of the present invention, specifically the distribution information of user questions under infant formula. The percentage in the figure represents the proportion of each demand category, that is, the magnitude of each demand category. Through magnitude evaluation, the core demand points of users are: infant formula product consultation and feeding problems.

[0071] In an embodiment of the present invention, a text description picture of the item can be generated according to the determined appeal category and its corresponding response information, and the text description picture can be directly spliced ​​with the item picture of the item to obtain an introduction picture of the item. A text description picture refers to a picture containing a text description of the item, and an item picture refers to an item-related picture, such as an item appearance picture, an item raw material picture, an item production process picture, an item applicable population picture, etc. In an optional embodiment, generating an introduction picture of the item includes: generating a text description picture of the item according to the determined appeal category and its corresponding response information, and merging the text description picture with the item picture of the item to obtain an introduction picture of the item.

[0072] The specific implementation method of fusing the text description image with the item image of the item can be selectively set. For example, the item image is overlaid on the text description image, and the transparency and / or position of the item image is adjusted. Exemplarily, image processing software or libraries, such as OpenCV or PIL libraries, are used to implement layer superposition or blending of two images and appropriate position adjustments. For another example, the text description image is placed at a preset position of the item image, such as added above or below the item image. By generating an introduction image of an item through image fusion, the design sense of the introduction image can be enhanced, thereby enhancing the user experience.

[0073] After the introduction picture is generated, the generated introduction picture can be further adjusted to a fixed-size picture. The embodiment of the present invention can use image processing software or a library to adjust the size of the image, such as scaling or cropping the image to a desired size, and it should be ensured that the quality and proportion of the image are maintained during the size adjustment process so that the finally generated introduction picture meets the expected size requirements.

[0074] In an embodiment of the present invention, a pre-trained model can be used to generate a text description image of an object. The network structure of the model can be selectively set. Generative adversarial networks (GANs) are a neural network model composed of a generator and a discriminator. The generator is responsible for generating realistic images, while the discriminator is responsible for evaluating whether the generated image is real or fake. The two networks compete with each other, and the ability of the generator to generate realistic images is gradually improved during the training process. In an embodiment of the present invention, the generator can be trained into a conditional generative adversarial network (Conditional GANs), and the determined claim category and its corresponding response information are input into a pre-trained conditional generative adversarial network, so that the conditional generative adversarial network outputs a text description image of the object.

[0075] Figure 3 : is a schematic diagram of the main process of the method for generating an image in an embodiment of the present invention. In this embodiment, the method for generating an image mainly includes three stages:

[0076] 1. Data preprocessing.

[0077] You can collect and organize conversation data on e-commerce platforms, such as consultations in online medical scenarios, product-related consultations in online transaction scenarios, etc., and clean and annotate the data for subsequent analysis. Cleaning mainly removes noise from the data. For example, annotating includes converting gender "male" and "female" to 0 and 1, marking the demand category of user questions in the sample, etc.

[0078] 2. Natural language processing: Use natural language processing technology to perform text analysis on question and answer data and extract user purchase decision points from it.

[0079] (1) Word vector representation: Convert words or phrases in natural language into vector representations so that computers can understand and process them. For example, algorithms such as Word2Vec and GloVe can represent words as vectors in a high-dimensional space. Through these word vectors, we can calculate the similarity between words and find words related to purchase decisions in product question and answer data.

[0080] For example, for infant formula products, purchasing decisions may be closely related to "nutritional ingredients", "suitable age" and "suitable physique". Using word vector representation, we can compare the similarity between the words in the user's question and these keywords to find out what aspects the user is concerned about when asking questions.

[0081] (2) Syntactic analysis: By analyzing the structure and grammatical rules of natural language sentences, we can understand the relationship between words in a sentence, identify the subject, predicate, object and other grammatical structures in the sentence, and better understand the user's questions and needs.

[0082] For example, for the question sentence "Is this milk powder suitable for babies in the milk-changing period?", syntactic analysis can identify "this milk powder" as the subject, "suitable" as the predicate, "take" as the object, and "babies in the milk-changing period" as a specific group of people. This helps to understand the user's question and better answer the user's needs.

[0083] (3) Semantic similarity calculation: It is used to measure the semantic similarity between different texts, that is, to determine the degree of similarity between the meanings of two sentences or phrases. This is very useful for understanding user needs, matching related questions and answers, and generating personalized recommendations.

[0084] For example, for a user question "Is this milk powder suitable for babies during the milk-changing period?", similarity can be calculated to find relevant question and answer data, such as an answer "It is very suitable. You can use the new and old mixing method to change milk. The operation method is to use milk powder of stage one for three days and then milk powder of stage two for three days, and repeat it 2 to 3 times." By comparing the semantic similarity between the question and the answer, it can be determined whether the answer matches the question.

[0085] (4) Sentiment analysis: used to analyze the emotions and sentiment tendencies conveyed in the text in order to understand the user's opinions, emotions, etc., and generate answers or recommendations that are more suitable for user needs.

[0086] For example, a user's comment "Very good, thank you for your patient answer!" can be identified through sentiment analysis to identify positive sentiment tendencies, thereby further understanding the user's satisfaction with the item and providing the user with more accurate feedback or recommendations.

[0087] (5) Mining core demands: By processing a large number of user questions under the items, the questions with strong user consultation demands are classified and processed to capture the core issues.

[0088] 2. Image generation: Generative adversarial networks (GANs) are used to generate product description images based on the user’s purchase decision point and product information, including: (1) text image generation, (2) image fusion, and (3) size adjustment. The specific implementation method can be found in the above-mentioned related content and will not be repeated here.

[0089] 3. Front-end display and interaction: embed the generated item introduction pictures into the item page of the e-commerce platform, for example, as the main picture of the item, or as the picture in the details page, to interact with users, provide purchase decision support and user feedback mechanism.

[0090] In actual application, the method of the embodiment of the present invention can be used to analyze the session data of one or more sessions of a user, so as to generate personalized introduction pictures according to user needs, so as to achieve different faces for different people. When analyzing the session data of a user's ongoing session, the generated introduction picture of the item can also reflect the user's real-time needs, so as to be closer to the user's purchase decision point.

[0091] Those skilled in the art may also adopt the method of the embodiment of the present invention to analyze multiple session data of multiple users, so that the generated introduction pictures of the items can meet the needs of most users and be closer to the purchase decision points of most users.

[0092] When the content on the item page cannot solve the user's purchase decision point, the user can consult customer service, ask questions to other users who purchase goods, and so on for specific consultation. In addition, when users have professional health problems, they can get answers through health consultation. For example: when a user purchases milk powder, the applicable age group, brand, taste, price, preferential policies and other information of the milk powder can be seen from the product details page, but the user wants to deeply understand whether the milk powder is suitable for the baby's physique, and can consult a professional nanny for answers. The embodiment of the present invention can determine the category of demands that the user is concerned about by parsing the conversation data of the item, and generate the introduction picture of the item according to the category of the demand and its corresponding response information. On the one hand, it can liberate manpower and improve the efficiency of item maintenance and store operation. On the other hand, it can timely obtain the user's purchase decision point and update the introduction picture of the item according to the user's recent conversation data, so that the introduction picture of the item is closer to the user's purchase decision point and improve the transaction conversion rate.

[0093] According to a second aspect of an embodiment of the present invention, a device for implementing the above method is provided.

[0094] Figure 4 FIG. 1 is a schematic diagram of the main modules of the apparatus for generating pictures according to an embodiment of the present invention. Figure 4 As shown, the device 400 for generating a picture includes:

[0095] Extraction module 401, extracting user questions and corresponding answer information from the item's session data;

[0096] A matching module 402 matches the user question with various preset demand categories to determine the demand category to which the user question belongs;

[0097] The generation module 403 generates an introduction picture of the item according to the demand category to which the user's question belongs and the corresponding response information.

[0098] Optionally, the extraction module extracts user questions and their corresponding answer information from the conversation data of the item, including: converting the conversation data into a conversation text vector, determining the grammatical structure of the conversation text vector through syntactic analysis, and extracting user questions and their corresponding answer information from the conversation text vector according to the grammatical structure of the conversation text vector.

[0099] Optionally, the matching module is further used for at least one of the following:

[0100] Before matching the user question with each preset demand category, determining the semantic similarity between the user question and the corresponding answer information, and determining whether the semantic similarity between the user question and the corresponding answer information meets the preset matching condition;

[0101] Before matching the user question with each preset demand category, sentiment analysis is performed on the session data to determine the user's satisfaction with the response information and to determine whether the user's satisfaction with the response information meets preset matching conditions.

[0102] Optionally, the generating module generates an introduction picture of the item according to the demand category to which the user's question belongs and the corresponding answer information, including:

[0103] Determine the distribution information of each preset demand category according to the demand category to which each user question belongs and the corresponding answer information;

[0104] According to the distribution information of each preset demand category, determining a core demand category from each preset demand category;

[0105] According to the core demand category and its corresponding response information, an introduction picture of the item is generated.

[0106] Optionally, the generation module generates an introduction picture of the item, including: generating a text description picture of the item according to the determined demand category and its corresponding response information; and fusing the text description picture with the item picture of the item to obtain an introduction picture of the item.

[0107] Optionally, the generation module generates a text description picture of the object based on the determined claim category and its corresponding response information, including: inputting the determined claim category and its corresponding response information into a pre-trained conditional generative adversarial network so that the conditional generative adversarial network outputs the text description picture of the object.

[0108] Optionally, the generation module merges the text description picture with the item picture of the item, including: overlaying the item picture on the text description picture and adjusting the transparency and / or position of the item picture, or placing the text description picture at a preset position of the item picture.

[0109] According to a third aspect of an embodiment of the present invention, there is provided an electronic device for generating a picture, including:

[0110] one or more processors;

[0111] a storage device for storing one or more programs,

[0112] When the one or more programs are executed by the one or more processors, the one or more processors implement the method provided in the first aspect of the embodiment of the present invention.

[0113] According to a fourth aspect of an embodiment of the present invention, a computer-readable medium is provided. When the program in the computer-readable medium is executed by a processor, the method provided by the first aspect of the embodiment of the present invention is implemented.

[0114] Figure 5 An exemplary system architecture 500 is shown to which the method or apparatus for generating a picture according to the embodiment of the present invention may be applied.

[0115] like Figure 5 As shown, system architecture 500 may include terminal devices 501, 502, 503, a network 504 and a server 505. Network 504 is used to provide a medium for communication links between terminal devices 501, 502, 503 and server 505. Network 504 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0116] Users can use terminal devices 501, 502, 503 to interact with server 505 through network 504 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 501, 502, 503, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0117] The terminal devices 501 , 502 , and 503 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0118] The server 505 may be a server that provides various services, such as a backend management server (only an example) that provides support for shopping websites browsed by users using the terminal devices 501, 502, and 503. The backend management server may analyze and process the received data such as product information query requests, and feed back the processing results (such as target push information, product information - only an example) to the terminal device.

[0119] It should be noted that the method for generating an image provided in the embodiment of the present invention is generally executed by the server 505 , and accordingly, the device for generating an image is generally arranged in the server 505 .

[0120] It should be understood that Figure 5 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0121] Reference below Figure 6 , which shows a schematic diagram of the structure of a computer system 600 of a terminal device suitable for implementing an embodiment of the present invention. Figure 6 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0122] like Figure 6 As shown, the computer system 600 includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the system 600 are also stored. The CPU 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0123] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0124] In particular, according to the embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the system of the present invention are executed.

[0125] It should be noted that the computer-readable medium shown in the present invention may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present invention, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0126] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0127] The modules involved in the embodiments of the present invention may be implemented by software or hardware. The modules described may also be set in a processor, for example, it may be described as: a processor includes an extraction module, a matching module and a generation module. The names of these units do not constitute a limitation on the units themselves in some cases, for example, the extraction module may also be described as a "module for generating an introduction picture of the item".

[0128] As another aspect, the present invention further provides a computer-readable medium, which may be included in the device described in the above embodiment; or may exist independently without being assembled into the device. The above computer-readable medium carries one or more programs, and when the above one or more programs are executed by a device, the device includes: extracting user questions and corresponding answer information from the conversation data of the item; matching the user questions with each preset demand category to determine the demand category to which the user questions belong; and generating an introduction picture of the item according to the demand category to which the user questions belong and the corresponding answer information.

[0129] According to the technical solution of the embodiment of the present invention, by parsing the conversation data of the item, the demand category that the user is concerned about can be determined. By generating an introduction picture of the item based on the demand category and its corresponding response information, on the one hand, manpower can be liberated and the efficiency of item maintenance and store operation can be improved. On the other hand, the user's purchase decision point can be obtained in time according to the user's recent conversation data and the introduction picture of the item can be updated, so that the introduction picture of the item is closer to the user's purchase decision point and the transaction conversion rate is improved.

[0130] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may occur depending on design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for generating an image, characterized in that: include: Extract user questions and their corresponding answer information from the item's session data; Matching the user question with each preset demand category to determine the demand category to which the user question belongs; According to the demand category to which the user's question belongs and its corresponding answer information, an introduction picture of the item is generated.

2. The method according to claim 1, characterized in that Extract user questions and their corresponding answer information from the item's session data, including: The conversation data is converted into a conversation text vector, and the grammatical structure of the conversation text vector is determined through syntactic analysis, so as to extract the user question and the corresponding answer information from the conversation text vector according to the grammatical structure of the conversation text vector.

3. The method according to claim 1, characterized in that Before matching the user question with each preset demand category, the method further includes at least one of the following: Determining the semantic similarity between the user question and the corresponding answer information, and determining whether the semantic similarity between the user question and the corresponding answer information meets a preset matching condition; The conversation data is subjected to sentiment analysis to determine the user's satisfaction with the response information, and to determine whether the user's satisfaction with the response information meets a preset matching condition.

4. The method according to claim 1, characterized in that According to the demand category of the user's question and its corresponding answer information, an introduction picture of the item is generated, including: Determine the distribution information of each preset demand category according to the demand category to which each user question belongs and the corresponding answer information; According to the distribution information of each preset demand category, determining a core demand category from each preset demand category; According to the core demand category and its corresponding response information, an introduction picture of the item is generated.

5. The method according to claim 1 or 4, characterized in that: Generate an introductory image of the item, including: According to the determined demand category and its corresponding response information, a text description picture of the item is generated; and the text description picture is merged with the item picture of the item to obtain an introduction picture of the item.

6. The method according to claim 5, characterized in that According to the determined demand category and its corresponding response information, a text description image of the item is generated, including: The determined demand category and its corresponding response information are input into a pre-trained conditional generative adversarial network, so that the conditional generative adversarial network outputs a text description image of the item.

7. The method according to claim 5, characterized in that Merging the text description image with the item image of the item includes: The object picture is overlaid on the text description picture, and the transparency and / or position of the object picture is adjusted, or the text description picture is placed at a preset position of the object picture.

8. A device for generating a picture, characterized in that: include: The extraction module extracts user questions and their corresponding answer information from the item's session data; A matching module, matching the user question with various preset demand categories to determine the demand category to which the user question belongs; The generation module generates an introduction picture of the item according to the demand category to which the user's question belongs and the corresponding response information.

9. An electronic device for generating a picture, characterized in that: include: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 7.

10. A computer-readable medium, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images