Optimized copywriting generation method and device based on user satisfaction, equipment and medium

Abstract

The embodiment of the invention provides an optimized document generation method and device based on user satisfaction, equipment and a medium, belongs to the technical field of artificial intelligence, and is suitable for the fields of financial science and technology and medical science and technology. The method comprises the steps of obtaining user feedback data of a target product; wherein the user feedback data is feedback data of the target user group on the target product; performing feature extraction on the user feedback data to obtain target feedback data; performing emotion recognition on the target feedback data to obtain user emotion features of the target user group; performing semantic recognition on the target feedback data to obtain user attention features of the target user group; performing causal inference based on the user emotion features and the user attention features to obtain user satisfaction influence features; and generating a product optimization copywriting of the target product based on the user satisfaction influence feature. According to the embodiment of the invention, the efficiency and accuracy of product optimization can be improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and is applicable to the fields of fintech and medical technology. In particular, it relates to a method, apparatus, device, and medium for optimizing copywriting generation based on user satisfaction. Background Technology

[0002] User satisfaction analysis is a research method for systematically evaluating user satisfaction with products or services. It optimizes user experience and improves business outcomes through feedback collection and data analysis. User satisfaction analysis can be applied to various scenarios. For example, in fintech, it analyzes customer satisfaction with financial products (such as insurance and wealth management products) to optimize product design and customer service processes. In healthcare technology, it analyzes patient feedback on medical services (such as in-person visits and online consultations) to improve healthcare quality and patient satisfaction.

[0003] Traditional user satisfaction analysis methods primarily rely on questionnaires and telephone follow-ups, which suffer from limited data collection scope, low analysis efficiency, and difficulty in real-time monitoring and dynamic optimization. Currently, customer satisfaction analysis systems based on artificial intelligence models are mainly used to analyze collected user feedback. However, in practical use, the sentiment analysis and semantic understanding capabilities of artificial intelligence models are limited, making it difficult to accurately capture customers' true needs and emotions, and thus unable to accurately analyze user satisfaction. Consequently, it is impossible to achieve product optimization based on user satisfaction, affecting the efficiency and accuracy of product optimization.

[0004] Therefore, improving the efficiency and accuracy of product optimization has become an urgent technical problem to be solved. Summary of the Invention

[0005] The main objective of this application is to propose a method, apparatus, device, and medium for generating optimized copy based on user satisfaction. This aims to solve the technical problem that product optimization based on user satisfaction is impossible due to the inability to accurately analyze user satisfaction, thereby improving the efficiency and accuracy of product optimization.

[0006] To achieve the above objectives, a first aspect of this application proposes a product optimization method based on user satisfaction, the method comprising: Obtain user feedback data for the target product; wherein, the user feedback data is feedback data from the target user group regarding the target product; Feature extraction is performed on the user feedback data to obtain the target feedback data; Sentiment recognition is performed on the target feedback data to obtain the user sentiment characteristics of the target user group; Semantic recognition is performed on the target feedback data to obtain the user attention characteristics of the target user group; Based on the user's emotional characteristics and the user's attention characteristics, causal inference is performed to obtain the user satisfaction impact characteristics; Based on the user satisfaction impact characteristics, generate product optimization copy for the target product.

[0007] In some embodiments, the step of performing causal inference based on the user's emotional characteristics and the user's attention characteristics to obtain user satisfaction influence characteristics includes: Based on the user's emotional characteristics and the user's attention characteristics, feature extraction is performed to obtain causal feature factors; Causal analysis is performed based on the user's emotional characteristics, the user's attention characteristics, and the causal characteristic factors to obtain the satisfaction causal relationship, wherein the satisfaction causal relationship is used to characterize the causal relationship between the causal characteristic factors; Factor screening is performed based on the aforementioned causal relationship of satisfaction to obtain the characteristics influencing user satisfaction.

[0008] In some embodiments, performing emotion recognition on the target feedback data to obtain the user emotion characteristics of the target user group includes: The target feedback data is subjected to sentiment classification to obtain the original sentiment classification probability; The original sentiment classification probabilities are filtered to obtain the target sentiment classification probabilities; The user sentiment characteristics of the target user group are determined based on the target sentiment classification probability.

[0009] In some embodiments, performing semantic recognition on the target feedback data to obtain the user attention characteristics of the target user group includes: The target feedback data is segmented into words to obtain segmented feedback data. Named entity recognition is performed on the segmented feedback data to obtain feedback entity information; Relationship extraction is performed on the word segmentation feedback data to obtain entity relationship information; The target feedback data is subjected to intent recognition to obtain feedback intent data; The feedback entity information, the entity relationship information, and the feedback intent data are fused to obtain the user attention characteristics of the target user group.

[0010] In some embodiments, generating product optimization copy for the target product based on the user satisfaction impact features includes: Obtain the original product functions of the target product; Based on the user satisfaction impact characteristics, the original product functions are screened to obtain defective product functions. Based on a pre-set expert knowledge graph, suggestions are generated for the defective product functions to obtain the product optimization copy for the target product.

[0011] In some embodiments, the user feedback data includes user comment information, user rating information, and user conversation records; the step of extracting features from the user feedback data to obtain target feedback data includes: Feature extraction is performed on the user comment information to obtain structured comment data; Feature extraction is performed on the user rating information to obtain structured rating data; Feature extraction is performed on the user dialogue records to obtain structured dialogue data; The target feedback data is obtained by feature fusion of the structured comment data, the structured rating data, and the structured dialogue data.

[0012] In some embodiments, the step of extracting features from the user comment information to obtain structured comment data includes: The user comment information is cleaned to obtain cleaned comment data; The cleaned comment data is deduplicated to obtain deduplicated comment data; The deduplicated comment data is labeled to obtain labeled comment data; Information is extracted from the labeled comment data to obtain the structured comment data.

[0013] To achieve the above objectives, a second aspect of this application proposes an optimized copywriting generation apparatus based on user satisfaction, the apparatus comprising: The feedback data acquisition module is used to acquire user feedback data of the target product; wherein, the user feedback data is feedback data of the target user group on the target product; The feature extraction module is used to extract features from the user feedback data to obtain target feedback data; The emotion recognition module is used to perform emotion recognition on the target feedback data to obtain the user emotion characteristics of the target user group. A semantic recognition module is used to perform semantic recognition on the target feedback data to obtain the user attention characteristics of the target user group. The causal inference module is used to perform causal inference based on the user's emotional characteristics and the user's attention characteristics to obtain the user satisfaction impact characteristics. The copywriting generation module is used to generate optimized product copy for the target product based on the user satisfaction impact characteristics.

[0014] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0015] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0016] This application proposes a method, apparatus, device, and medium for generating optimized copy based on user satisfaction. It utilizes user feedback data from a target user group regarding a target product, extracting features from this data to obtain target feedback data. This allows for focusing on key information, removing redundancy, and improving analysis efficiency. Next, it performs sentiment recognition on the target feedback data to obtain the user sentiment characteristics of the target user group, understanding users' emotional inclinations towards the target product. Semantic recognition is then performed on the target feedback data to obtain the user attention characteristics of the target user group, clarifying users' focus points on the target product. Furthermore, causal inference is performed based on user sentiment and attention characteristics to obtain user satisfaction impact characteristics, accurately identifying key factors influencing satisfaction. Finally, optimized product copy is generated based on these user satisfaction impact characteristics, making the copy more targeted and effectively addressing key issues affecting user satisfaction, thus contributing to improved efficiency and accuracy of product optimization. Attached Figure Description

[0017] Figure 1 This is a flowchart of the optimized copywriting generation method based on user satisfaction provided in the embodiments of this application; Figure 2 yes Figure 1 The flowchart of step S102 in the document; Figure 3 yes Figure 2 The flowchart of step S201 in the text; Figure 4 yes Figure 1 The flowchart of step S103 in the process; Figure 5 yes Figure 1 The flowchart of step S104 in the process; Figure 6 yes Figure 1 The flowchart of step S105 in the process; Figure 7 yes Figure 1 The flowchart of step S106 in the process; Figure 8This is a schematic diagram of the structure of the optimized copywriting generation device based on user satisfaction provided in the embodiments of this application; Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0021] First, let's analyze some of the terms used in this application: Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0022] Natural Language Processing (NLP): NLP uses computers to process, understand, and utilize human language (such as Chinese and English). It is a branch of artificial intelligence and an interdisciplinary field of computer science and linguistics, often referred to as computational linguistics. NLP includes syntactic analysis, semantic analysis, and discourse understanding. It is commonly used in machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, intent recognition, information extraction and filtering, text classification and clustering, sentiment analysis, and opinion mining. It involves data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computation.

[0023] User satisfaction analysis is a research method for systematically evaluating user satisfaction with products or services. It optimizes user experience and improves business outcomes through feedback collection and data analysis. User satisfaction analysis can be applied to various scenarios. For example, in fintech, it analyzes customer satisfaction with financial products (such as insurance and wealth management products) to optimize product design and customer service processes. In healthcare technology, it analyzes patient feedback on medical services (such as in-person visits and online consultations) to improve healthcare quality and patient satisfaction.

[0024] Traditional user satisfaction analysis methods primarily rely on questionnaires and telephone follow-ups, which suffer from limitations such as limited data collection scope, low analysis efficiency, and difficulty in real-time monitoring and dynamic optimization. Currently, customer satisfaction analysis systems based on artificial intelligence models are mainly used to analyze collected user feedback information. However, in practical use, user feedback information contains a large amount of unstructured information (such as social media comments and customer service dialogue records). The sentiment analysis and semantic understanding capabilities of artificial intelligence models are limited, making it difficult to accurately capture customers' true needs and emotions, and thus unable to accurately analyze user satisfaction. Consequently, it is impossible to achieve product optimization based on user satisfaction, affecting the efficiency and accuracy of product optimization.

[0025] Based on this, embodiments of this application provide a method, apparatus, device, and medium for generating optimized copy based on user satisfaction, aiming to solve the technical problem that product optimization based on user satisfaction is impossible due to the inability to accurately analyze user satisfaction, thereby improving the efficiency and accuracy of product optimization.

[0026] The method, apparatus, device, and medium for generating optimized copy based on user satisfaction provided in this application are specifically described through the following embodiments. First, the method for generating optimized copy based on user satisfaction in this application embodiment is described.

[0027] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0028] Foundational technologies in artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0029] The optimized copywriting generation method based on user satisfaction provided in this application relates to the field of artificial intelligence technology. This optimized copywriting generation method based on user satisfaction can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the optimized copywriting generation method based on user satisfaction, but is not limited to the above forms.

[0030] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0031] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0032] Figure 1 This is an optional flowchart of the optimized copywriting generation method based on user satisfaction provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S106.

[0033] Step S101: Obtain user feedback data for the target product; wherein, the user feedback data is the feedback data of the target user group on the target product; Step S102: Extract features from user feedback data to obtain target feedback data; Step S103: Perform sentiment recognition on the target feedback data to obtain the user sentiment characteristics of the target user group; Step S104: Perform semantic recognition on the target feedback data to obtain the user attention characteristics of the target user group; Step S105: Based on user emotional characteristics and user attention characteristics, perform causal inference to obtain the characteristics affecting user satisfaction; Step S106: Generate product optimization copy for the target product based on user satisfaction impact characteristics.

[0034] Steps S101 to S106, as illustrated in this embodiment, involve using user feedback data from the target user group regarding the target product and extracting features from this data to obtain target feedback data. This process focuses on key information, removes redundancy, and improves analysis efficiency. Next, sentiment recognition is performed on the target feedback data to obtain the user sentiment characteristics of the target user group, understanding users' emotional inclinations towards the target product. Semantic recognition is then performed on the target feedback data to obtain the user attention characteristics of the target user group, clarifying users' focus points on the target product. Furthermore, causal inference is performed based on the user sentiment and attention characteristics to obtain user satisfaction impact characteristics, accurately identifying key factors affecting satisfaction. Finally, product optimization copy for the target product is generated based on the user satisfaction impact characteristics, making the copy more targeted and effectively addressing key issues affecting user satisfaction, thus improving the efficiency and accuracy of product optimization.

[0035] Step S101 involves acquiring user feedback data for the target product, providing comprehensive foundational material for subsequent analysis. Step S102 extracts features from the feedback data to obtain target feedback data, enabling the focus on key information, removal of redundancy, and improved analysis efficiency. Step S103 performs sentiment recognition to obtain user emotional characteristics, providing a clear understanding of users' emotional inclinations towards the product. Step S104 conducts semantic recognition to obtain user attention characteristics, clarifying the product aspects that users are focusing on. Step S105 uses causal inference based on sentiment and attention characteristics to derive user satisfaction-influencing features, accurately identifying factors affecting satisfaction. Finally, step S106 generates product optimization copy based on these features, making the copy more targeted, effectively addressing key issues affecting user satisfaction, improving product optimization results, and enhancing product competitiveness and user loyalty.

[0036] In step S101 of some embodiments, the target product refers to a specific product that the enterprise hopes to understand user feedback and optimize and improve. The target product can be a physical product (such as an insurance product or a wealth management product) or a virtual service product (such as an online wealth management platform, a hospital portal website, offline medical treatment, online consultation services, etc.), and is not limited to these.

[0037] The target user group is a group of users with similar characteristics, needs and behaviors who use or are potential users of the target product, such as users who have purchased / are about to purchase critical illness insurance.

[0038] User feedback data is a collection of information expressed by the target user group regarding their views, opinions, suggestions, and experiences with the product through various channels (such as online reviews, customer service communication, questionnaires, social media platforms, etc.). User feedback data truly reflects the user's acceptance of the product and their actual needs.

[0039] By obtaining user feedback data from the target user group on the target product, we can understand the advantages and disadvantages of the target product in actual use from multiple angles and levels, gain a deeper understanding of users' expectations and actual needs for the product, help to discover the gap between the product and user needs, and point the way for product optimization and improvement.

[0040] In some embodiments, user feedback data includes user comments, user ratings, and user conversation logs; User reviews are written evaluations of a target product by the target user group, which may include the product's advantages, disadvantages, and user experience. For example, after purchasing critical illness insurance, a user might review it as saying, "This critical illness insurance has broad coverage, but the premium is relatively high."

[0041] User ratings are quantitative evaluations of products by the target user group through rating, usually using star ratings or numerical scores, which directly reflect the target user group's overall satisfaction with the target product. For example, a user might give an online financial management platform 3 stars (out of 5).

[0042] User chat logs are the conversations between target users and customer service representatives, reflecting the problems, questions, and expectations they encounter while using the product. For example, a user might ask in a financial platform's customer service window, "Is the return on this financial product stable?"

[0043] For example, insurance companies can publish satisfaction surveys on critical illness insurance through online questionnaire platforms, inviting customers who have purchased critical illness insurance to fill them out; set up a dedicated feedback channel in their official customer service hotline to record customer inquiries and feedback; and monitor user discussions on critical illness insurance on social media platforms to collect relevant evaluations and opinions.

[0044] Please see Figure 2 In some embodiments, step S102 may include, but is not limited to, steps S201 to S204: Step S201: Extract features from user comment information to obtain structured comment data; Step S202: Extract features from user rating information to obtain structured rating data; Step S203: Extract features from user dialogue records to obtain structured dialogue data; Step S204: Feature fusion is performed on the structured comment data, structured rating data, and structured dialogue data to obtain the target feedback data.

[0045] Steps S201 to S204, as illustrated in this embodiment, extract features from user comment information to obtain structured comment data, which can accurately identify the key points of the comments. Extracting features from user rating information to obtain structured rating data clearly presents rating-related features. Extracting features from user dialogue records to obtain structured dialogue data effectively extracts the core content of the dialogue. Fusing features from structured comment data, structured rating data, and structured dialogue data comprehensively integrates multi-dimensional user feedback information, thereby obtaining more complete and accurate target feedback data.

[0046] Please see Figure 3 In some embodiments, step S201 may include, but is not limited to, steps S301 to S304: Step S301: Clean the user comment information to obtain cleaned comment data; Step S302: Perform deduplication on the cleaned comment data to obtain deduplicated comment data; Step S303: Label the deduplicated comment data to obtain labeled comment data; Step S304: Extract information from the labeled comment data to obtain structured comment data.

[0047] Steps S301 to S304, as illustrated in this embodiment, involve cleaning user comment information to obtain cleaned comment data, removing noise and erroneous data to ensure data quality. Deduplication of the cleaned comment data reduces interference from duplicate data, improving data validity and processing efficiency. Annotation of the deduplicated comment data clarifies its meaning and improves the accuracy of data processing. Finally, information extraction from the annotated comment data yields structured comment data, facilitating storage, retrieval, and use, further enhancing data processing efficiency.

[0048] In step S301 of some embodiments, natural language processing technology is used to filter, correct, and standardize the acquired user comment information to obtain cleaned comment data. This includes removing special symbols, garbled text, and irrelevant advertising information from the comments, standardizing data formats (such as date formats and text encoding formats), and correcting spelling and grammatical errors. This improves data quality and avoids biases in analysis results due to data errors or noise.

[0049] In step S302 of some embodiments, techniques such as hash algorithms and similarity comparisons can be used to compare the cleaned comment data one by one, identify comments with completely identical or highly similar content, and delete them, keeping only one copy. This can reduce data redundancy, lower data storage costs, and improve data processing efficiency.

[0050] In step S303 of some embodiments, annotation rules are first formulated based on business needs and analysis objectives. Then, the deduplicated comment data is annotated manually or using automated annotation tools to obtain annotated comment data. The annotation types can include labeling comments as positive, negative, or neutral, or indicating that the comments relate to product features, service quality, price, etc. This makes the data more meaningful and facilitates subsequent classification analysis, sentiment analysis, etc., providing more meaningful data for product optimization.

[0051] In step S304 of some embodiments, after data cleaning, deduplication, and annotation, annotated comment data is obtained. Natural language processing techniques and information extraction algorithms are then used to identify and extract key entities, attributes, relationships, and other information from the annotated comment data, and this information is stored in a database or data table to form a structured data format, resulting in structured comment data. This facilitates data storage, retrieval, and management, enables rapid location and analysis of specific information, improves data utilization efficiency, and provides a more convenient data foundation for data mining, decision support, and other applications.

[0052] For example: Extract information such as the insurance product name, review type, key advantages, and key disadvantages from the labeled review data related to a certain insurance product, and form a structured data table. For example: [Insurance Product Name: Critical Illness Insurance; Evaluation Type: Positive Evaluation; Key Advantages: Broad Coverage; Key Disadvantages: Low Coverage Amount].

[0053] Information such as product name, return details, risk level, and transaction fees are extracted from labeled review data related to financial service platforms to form structured data. For example: [Financial product name: Certain wealth management product; Return: Stable return; Risk level: Low risk; Transaction fee: Relatively high].

[0054] In some embodiments, step S202 may include, but is not limited to, the following steps: User rating information is cleaned to obtain cleaned rating data; The cleaning score data is deduplicated to obtain deduplicated score data; Information is extracted from the deduplicated scoring data to obtain structured scoring data.

[0055] By following the steps above, users' subjective evaluations can be transformed into objective numerical indicators, which facilitates quantitative analysis and comparison, and intuitively reflects the level of user satisfaction with different products or aspects.

[0056] Specifically, the specific implementation methods of the above-mentioned sub-steps of step S202 are basically the same as those of the above-mentioned steps S301, S302 and S304, and will not be repeated here.

[0057] In some embodiments, step S203 may include, but is not limited to, the following steps: The user dialogue information is cleaned to obtain cleaned dialogue data. The cleaned dialogue data is deduplicated to obtain deduplicated dialogue data. The deduplicated dialogue data is labeled to obtain labeled dialogue data. Information is extracted from the labeled dialogue data to obtain structured dialogue data.

[0058] By following the steps above, we can gain a deeper understanding of the issues and needs that users are concerned about during the consultation process, thereby better meeting their expectations.

[0059] Specifically, the specific implementation methods of the above-mentioned sub-steps of step S203 are basically the same as those of the above-mentioned steps S301 to S304, and will not be repeated here.

[0060] In step S204 of some embodiments, a data splicing method can be used to fuse structured review data, structured rating data, and structured dialogue data through multi-source feature fusion to form target feedback data. This integrates information from multiple aspects such as user comments, ratings, and dialogues, and can more comprehensively and accurately reflect users' feedback and needs for the product, avoiding the limitations of a single data source.

[0061] Please see Figure 4 In some embodiments, step S103 may include, but is not limited to, steps S401 to S403: Step S401: Perform sentiment classification on the target feedback data to obtain the original sentiment classification probability; Step S402: Filter the original sentiment classification probabilities to obtain the target sentiment classification probabilities; Step S403: Determine the user sentiment characteristics of the target user group based on the target sentiment classification probability.

[0062] Steps S401 to S403, as illustrated in this embodiment, involve classifying the target feedback data to obtain the original sentiment classification probability. Next, the original sentiment classification probability is filtered to obtain the target sentiment classification probability, eliminating uncertain and inaccurate classification results and improving data reliability. Finally, the target sentiment classification probability is used to determine the user sentiment characteristics of the target user group, enabling a more accurate grasp of user sentiment tendencies and providing support for product optimization and service improvement.

[0063] In step S401 of some embodiments, a pre-trained sentiment classification model (such as BERT, GPT, etc.) can be used to classify the target feedback data to obtain the original sentiment classification probabilities corresponding to multiple sentiment types. This can comprehensively mine the sentiment information in the target feedback data, transforming vague user feedback into specific, quantifiable sentiment probabilities. For example, taking a critical illness insurance policy as an example, collecting and processing user reviews yields the target feedback data "This critical illness insurance has broad coverage, reasonable premiums, and is worth recommending." Using a BERT-based sentiment classification model to classify the sentiment of this target feedback data, for the review "This application has a beautiful interface and smooth operation," the model might output a positive sentiment probability of 0.85, a negative sentiment probability of 0.05, and a neutral sentiment probability of 0.1.

[0064] In step S402 of some embodiments, the data with the highest probability value is selected from the original sentiment classification probabilities as the target sentiment classification probability, which helps to improve the accuracy and reliability of sentiment classification.

[0065] In step S403 of some embodiments, after obtaining the target sentiment classification probability, the sentiment type corresponding to the target sentiment classification probability is used as the user sentiment feature of the target user group.

[0066] Please see Figure 5 In some embodiments, step S104 may include, but is not limited to, steps S501 to S505: Step S501: Perform word segmentation on the target feedback data to obtain word segmentation feedback data; Step S502: Perform named entity recognition on the word segmentation feedback data to obtain feedback entity information; Step S503: Extract relationships from the word segmentation feedback data to obtain entity relationship information; Step S504: Perform intent recognition on the target feedback data to obtain feedback intent data; Step S505: The feedback entity information, entity relationship information and feedback intent data are fused to obtain the user attention characteristics of the target user group.

[0067] Steps S501 to S505, as illustrated in this embodiment, involve segmenting the target feedback data into words to obtain segmented feedback data, transforming complex text into easily processed word units. Named entity recognition is then performed on the segmented feedback data to obtain feedback entity information, accurately extracting key entities and clarifying the target audience. Relationship extraction is then performed on the segmented feedback data to obtain entity relationship information, clarifying the inherent connections between entities and enriching the information dimensions. Intent recognition is then performed on the target feedback data to obtain feedback intent data, clarifying the user's intent. Finally, the feedback entity information, entity relationship information, and feedback intent data are fused to obtain the user attention characteristics of the target user group. This allows for in-depth analysis of the target user group's concerns, needs, and other user attention characteristics, providing data support for product optimization.

[0068] In step S501 of some embodiments, word segmentation algorithms (such as dictionary-based word segmentation methods, statistical word segmentation methods, etc.) can be used to segment the target feedback data to obtain word segmentation feedback data, decomposing the long text into words, which facilitates more detailed analysis of the text in the future, such as named entity recognition, relation extraction, etc., and can improve the accuracy and efficiency of the analysis.

[0069] In step S502 of some embodiments, a named entity recognition model (such as a rule-based model, a machine learning-based model, etc.) is used to analyze the word segmentation feedback data, accurately identify the key entities in the word segmentation feedback data, and obtain feedback entity information, which helps to focus on the core objects that users care about.

[0070] For example, taking a critical illness insurance policy as an example, collecting and processing user reviews yields word-segmented feedback data such as "this," "critical illness insurance," "coverage," "broad," "premium," "suitable," "worth it," and "recommended." Named entity recognition then yields the feedback entity information as "critical illness insurance," "coverage," and "premium."

[0071] In step S503 of some embodiments, relation extraction algorithms (such as pattern matching-based methods, deep learning-based methods, etc.) can be used to analyze word segmentation feedback data, find the relationships between entities, obtain entity relationship information, reveal the inherent connections between entities, more comprehensively understand the semantics of the text, and help to deeply analyze users' views and needs on different combinations of entities.

[0072] For example, if there is a relationship between the feedback entity information "critical illness insurance", "coverage", and "premium", then the entity relationship information is: critical illness insurance - coverage - premium.

[0073] In step S504 of some embodiments, an intent recognition model (such as a text classification-based model) can be used to identify the intent of the target feedback data to obtain the user's intent category, i.e., the feedback intent data. This can accurately grasp the user's core needs, help to respond to the user in a targeted manner, improve the corresponding defects of the product, and increase user satisfaction.

[0074] For example, taking hospital offline medical services as an example, if a patient reports that "the medical experience is terrible, there is a lack of guidance, it takes a long time to figure things out, and the queuing time is too long", after intention recognition, the feedback intention data obtained is "complaint".

[0075] In step S505 of some embodiments, after obtaining feedback entity information, entity relationship information and feedback intent data, the feedback entity information, entity relationship information and feedback intent data are feature-concatenated to obtain the user attention characteristics of the target user group, so as to comprehensively and systematically understand the focus of attention and demand tendency of the target user group.

[0076] Please see Figure 6 In some embodiments, step S105 includes, but is not limited to, steps S601 to S603: Step S601: Extract features based on user emotional features and user attention features to obtain causal feature factors; Step S602: Based on user emotional characteristics, user attention characteristics and causal characteristic factors, causal analysis is performed to obtain the satisfaction causal relationship, wherein the satisfaction causal relationship is used to characterize the causal relationship between causal characteristic factors; Step S603: Factor screening is performed based on the causal relationship of satisfaction to obtain the characteristics of user satisfaction impact.

[0077] Steps S601 to S603, as illustrated in this embodiment, extract causal feature factors based on user emotional characteristics and user attention characteristics. This allows for the precise identification of potential factors affecting user experience of the target product. Causal analysis is then performed based on user emotional characteristics, user attention characteristics, and causal feature factors to obtain satisfaction causal relationships. These satisfaction causal relationships characterize the causal relationships between causal feature factors, clearly demonstrating how each factor interacts to influence satisfaction. Finally, factor screening is performed based on the satisfaction causal relationships to obtain user satisfaction impact characteristics. This process removes irrelevant interference, ultimately focusing on core aspects and accurately grasping the factors influencing user satisfaction, thus improving the targeting and accuracy of product optimization.

[0078] In step S601 of some embodiments, data mining and machine learning algorithms are used to analyze user emotional characteristics and user attention characteristics, and feature elements with potential causal relationships are screened out to obtain causal feature factors, wherein the causal feature factors are factors that may have a causal impact on results such as user satisfaction.

[0079] In step S602 of some embodiments, causal inference methods (such as structural equation modeling, Bayesian networks, etc.) or machine learning algorithms can be used to perform causal analysis on user emotional characteristics, user attention characteristics, and causal characteristic factors. This analyzes the causal relationship between causal characteristic factors and user satisfaction, obtains the satisfaction causal relationship, and provides a deeper understanding of the influence path and degree of each factor on user satisfaction. Specifically, the satisfaction causal relationship characterizes the causal relationships between causal characteristic factors, describes how these factors interact to influence user satisfaction, and clarifies which factors cause an increase or decrease in user satisfaction.

[0080] In step S603 of some embodiments, factors with a greater degree of influence are selected from all causal feature factors based on the causal relationship of satisfaction obtained from causal analysis, and these factors are used as user satisfaction influence features. These user satisfaction influence features are key features that have a significant impact on user satisfaction, which helps to improve the targeting of product optimization, thereby improving the efficiency and accuracy of product optimization.

[0081] For example: Example 1: When analyzing user reviews of a critical illness insurance policy, it was found from user emotional characteristics that many users expressed concerns about the covered illnesses (negative emotions). From user focus characteristics, it was learned that users were particularly concerned about the types of illnesses covered and the payout conditions. This yielded causal characteristic factors such as "insufficient coverage of illnesses" and "stringent payout conditions."

[0082] Furthermore, causal analysis revealed that "insufficient coverage of diseases" leads to a decrease in user satisfaction with the scope of coverage, thereby affecting overall satisfaction; "stringent compensation conditions" directly reduce user satisfaction with claims services, negatively impacting overall satisfaction, thus clarifying the causal relationship between these causal characteristic factors and satisfaction.

[0083] Finally, through factor screening, it was found that "types of diseases covered" and "payment conditions" are the features that have the greatest impact on user satisfaction. These were identified as the features that affect user satisfaction, and insurance companies can optimize their products based on these two aspects.

[0084] Example 2: For a certain financial trading website, user sentiment characteristics show that some users feel frustrated with the operation process (negative emotion), while user attention characteristics show that users care about the simplicity of the operation steps and the speed of transaction response. Through feature extraction, causal feature factors such as "cumbersome operation steps" and "slow transaction response" are obtained.

[0085] Furthermore, causal analysis revealed that "cumbersome operation steps" can cause user dissatisfaction during operation, reduce user satisfaction with the operation experience, and thus affect overall satisfaction with the webpage; "slow transaction response" can cause user dissatisfaction with transaction efficiency, which also affects overall satisfaction, thus revealing the corresponding causal relationship of satisfaction.

[0086] Finally, after factor screening, it was found that "operation steps" and "transaction response speed" are the features with the greatest impact on user satisfaction. These were identified as the features that affect user satisfaction, and the web development team can focus on optimizing these two aspects to improve user satisfaction.

[0087] Example 3: For a healthcare service platform, user sentiment features show that many users report long service wait times (negative emotions), while user attention features show users' relationships with doctors' professional competence and communication attitude. Through feature extraction, causal feature factors such as "long wait times" and "communication attitude" are obtained.

[0088] Furthermore, causal analysis revealed that "long waiting times" can cause user dissatisfaction during the consultation process, reducing satisfaction and thus affecting overall satisfaction with the medical service platform; "poor communication attitude" can also cause user dissatisfaction with the consultation experience, similarly affecting satisfaction, thus establishing a corresponding causal relationship in satisfaction.

[0089] Finally, through factor screening, it was found that "waiting time" and "communication attitude" are the features that have the greatest impact on patient satisfaction. These were identified as the features that affect user satisfaction, and hospitals can focus on optimizing these two aspects to improve patient satisfaction.

[0090] Please see Figure 7 In some embodiments, step S106 may include, but is not limited to, steps S701 to S703: Step S701: Obtain the original product functions of the target product; Step S702: Based on the characteristics affecting user satisfaction, the original product functions are screened to obtain defective product functions; Step S703: Based on the preset expert knowledge graph, suggestions are generated for the defective product functions to obtain the product optimization copy for the target product.

[0091] Steps S701 to S703, as illustrated in this embodiment, involve obtaining the original product functions of the target product to clarify its existing functions. Next, based on user satisfaction impact characteristics, the original product functions are screened to identify defective product functions, pinpointing these defects and clarifying improvement directions. Finally, suggestions are generated based on a pre-set expert knowledge graph to obtain product optimization copy for the target product. This ensures that the product optimization copy not only addresses user pain points but also possesses professional feasibility, effectively improving the efficiency and accuracy of product optimization.

[0092] In step S701 of some embodiments, the original product functions refer to the various functional characteristics possessed by the target product, which are the basis for the product to provide services or meet the needs of users. For medical insurance, the original product functions may include reimbursement of inpatient medical expenses and reimbursement of outpatient medical expenses; for a stock trading platform, the original product functions may cover stock trading, market information inquiry, and information push; for a medical service platform, the original product functions may include online doctor consultation, uploading and viewing of medical records, drug purchase guidance, and health information push.

[0093] Specifically, original product functions can be extracted from the target product's product documents, design materials, and development records, or they can be obtained by conducting actual testing and experience summaries of the product, and are not limited to these.

[0094] In step S702 of some embodiments, the original product functions are evaluated and screened one by one according to the characteristics affecting user satisfaction. Functions that do not meet user expectations and cause user dissatisfaction are identified as defective product functions, thus avoiding blind optimization and improving the targeting and effectiveness of optimization work. Among them, defective product functions refer to those functions in the original product functions that have a negative impact on user satisfaction or fail to meet user needs. Defective product functions may have problems such as unreasonable design, poor performance, or complex operation.

[0095] In step S703 of some embodiments, suggestions for defective product functions can be generated based on a preset expert knowledge graph, generating targeted and feasible optimization suggestions, and obtaining product optimization documents for the target product.

[0096] Among them, product optimization documents are written descriptions of specific improvement plans and measures proposed for the defective functions of the target product, including optimization goals, optimization methods, implementation steps, etc.

[0097] A pre-defined expert knowledge graph is a knowledge system that structures and represents experts' knowledge and experience in a specific field (such as finance or healthcare) in the form of a graph. In the insurance field, it may include expert knowledge on insurance product design and claims process optimization; in the financial web product field, it may cover expert experience on transaction system architecture and user experience design; and in the healthcare field, it may cover expert experience on consultation processes and queuing / calling logic design.

[0098] Understandably, product optimization copy for a target product can be used as an early warning and reminder to the relevant team to optimize the target product and service processes as soon as possible, improve user satisfaction with the target product, and thus enhance the company's market competitiveness and brand value.

[0099] Following step S106 in some embodiments, the optimized copy generation method based on user satisfaction further includes: Based on the product optimization copy of the target product, similar products or product beta testing opportunities are pushed to users who have low satisfaction with the target product, and so on.

[0100] The user satisfaction-based optimized copywriting generation method, apparatus, device, and medium provided in this application can be applied not only to the fintech and medical technology fields, but also to customer satisfaction analysis and optimization in different industries, and have wide applicability and scalability.

[0101] For example: Retail industry: Analyze customer feedback on products and services to optimize the shopping experience and after-sales service.

[0102] Education sector: Analyze student and parent satisfaction with educational services to optimize teaching content and learning experience.

[0103] Tourism industry: Analyze tourist feedback on tourism products and services to improve the travel experience and customer loyalty.

[0104] Please see Figure 8 This application also provides an optimized copywriting generation device based on user satisfaction, which can implement the above-mentioned optimized copywriting generation method based on user satisfaction. The device includes: The feedback data acquisition module 801 is used to acquire user feedback data of the target product; wherein, the user feedback data is the feedback data of the target user group on the target product; The feature extraction module 802 is used to extract features from user feedback data to obtain target feedback data; The emotion recognition module 803 is used to perform emotion recognition on the target feedback data to obtain the user emotion characteristics of the target user group. The semantic recognition module 804 is used to perform semantic recognition on the target feedback data to obtain the user attention characteristics of the target user group. The causal inference module 805 is used to perform causal inference based on user emotional characteristics and user attention characteristics to obtain the influence characteristics of user satisfaction. The copywriting generation module 806 is used to generate product optimization copy for the target product based on user satisfaction impact characteristics.

[0105] The specific implementation of the optimized copywriting generation device based on user satisfaction is basically the same as the specific implementation of the optimized copywriting generation method based on user satisfaction described above, and will not be repeated here.

[0106] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned optimized copywriting generation method based on user satisfaction. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0107] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 902 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 to execute the user satisfaction-based optimized copywriting generation method of the embodiments of this application. The input / output interface 903 is used to implement information input and output; The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0108] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for generating optimized copy based on user satisfaction.

[0109] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0110] The user satisfaction-based optimized copywriting generation method, apparatus, device, and medium provided in this application embodiment utilize user feedback data from a target user group regarding a target product. By extracting features from this user feedback data, target feedback data is obtained, focusing on key information, removing redundancy, and improving analysis efficiency. Next, sentiment recognition is performed on the target feedback data to obtain the user sentiment characteristics of the target user group, understanding users' emotional inclinations towards the target product. Semantic recognition is then performed on the target feedback data to obtain the user attention characteristics of the target user group, clarifying users' focus points on the target product. Furthermore, causal inference is performed based on the user sentiment characteristics and user attention characteristics to obtain user satisfaction impact characteristics, accurately identifying key factors affecting satisfaction. Finally, optimized product copywriting for the target product is generated based on the user satisfaction impact characteristics, making the copywriting more targeted and effectively addressing key issues affecting user satisfaction, thus contributing to improved efficiency and accuracy of product optimization.

[0111] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0112] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0113] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0114] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0115] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0116] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0118] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0120] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0121] The software tools or components not belonging to our company that appear in the embodiments of this application are for illustrative purposes only and do not represent actual use.

[0122] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for optimizing copywriting based on user satisfaction, characterized in that, The method includes: Obtain user feedback data for the target product; wherein, the user feedback data is feedback data from the target user group regarding the target product; Feature extraction is performed on the user feedback data to obtain the target feedback data; Sentiment recognition is performed on the target feedback data to obtain the user sentiment characteristics of the target user group; Semantic recognition is performed on the target feedback data to obtain the user attention characteristics of the target user group; Based on the user's emotional characteristics and the user's attention characteristics, causal inference is performed to obtain the user satisfaction impact characteristics; Based on the user satisfaction impact characteristics, generate product optimization copy for the target product.

2. The method according to claim 1, characterized in that, The process of performing causal inference based on the user's emotional characteristics and attention characteristics to obtain user satisfaction impact characteristics includes: Based on the user's emotional characteristics and the user's attention characteristics, feature extraction is performed to obtain causal feature factors; Causal analysis is performed based on the user's emotional characteristics, the user's attention characteristics, and the causal characteristic factors to obtain the satisfaction causal relationship, wherein the satisfaction causal relationship is used to characterize the causal relationship between the causal characteristic factors; Factor screening is performed based on the aforementioned causal relationship of satisfaction to obtain the characteristics influencing user satisfaction.

3. The method according to claim 1, characterized in that, The step of performing sentiment recognition on the target feedback data to obtain the user sentiment characteristics of the target user group includes: The target feedback data is subjected to sentiment classification to obtain the original sentiment classification probability; The original sentiment classification probabilities are filtered to obtain the target sentiment classification probabilities; The user sentiment characteristics of the target user group are determined based on the target sentiment classification probability.

4. The method according to claim 1, characterized in that, The step of performing semantic recognition on the target feedback data to obtain the user attention characteristics of the target user group includes: The target feedback data is segmented into words to obtain segmented feedback data. Named entity recognition is performed on the segmented feedback data to obtain feedback entity information; Relationship extraction is performed on the word segmentation feedback data to obtain entity relationship information; The target feedback data is subjected to intent recognition to obtain feedback intent data; The feedback entity information, the entity relationship information, and the feedback intent data are fused to obtain the user attention characteristics of the target user group.

5. The method according to any one of claims 1 to 4, characterized in that, The process of generating product optimization copy for the target product based on the user satisfaction impact characteristics includes: Obtain the original product functions of the target product; Based on the user satisfaction impact characteristics, the original product functions are screened to obtain defective product functions. Based on a pre-set expert knowledge graph, suggestions are generated for the defective product functions to obtain the product optimization copy for the target product.

6. The method according to any one of claims 1 to 4, characterized in that, The user feedback data includes user comment information, user rating information, and user conversation records; the step of extracting features from the user feedback data to obtain target feedback data includes: Feature extraction is performed on the user comment information to obtain structured comment data; Feature extraction is performed on the user rating information to obtain structured rating data; Feature extraction is performed on the user dialogue records to obtain structured dialogue data; The target feedback data is obtained by feature fusion of the structured comment data, the structured rating data, and the structured dialogue data.

7. The method according to claim 6, characterized in that, The step of extracting features from the user comment information to obtain structured comment data includes: The user comment information is cleaned to obtain cleaned comment data; The cleaned comment data is deduplicated to obtain deduplicated comment data; The deduplicated comment data is labeled to obtain labeled comment data; Information is extracted from the labeled comment data to obtain the structured comment data.

8. A device for generating optimized copy based on user satisfaction, characterized in that, The device includes: The feedback data acquisition module is used to acquire user feedback data of the target product; wherein, the user feedback data is feedback data of the target user group on the target product; The feature extraction module is used to extract features from the user feedback data to obtain target feedback data; The emotion recognition module is used to perform emotion recognition on the target feedback data to obtain the user emotion characteristics of the target user group. The semantic recognition module is used to perform semantic recognition on the target feedback data to obtain the user attention characteristics of the target user group. The causal inference module is used to perform causal inference based on the user's emotional characteristics and the user's attention characteristics to obtain the user satisfaction impact characteristics. The copywriting generation module is used to generate optimized product copy for the target product based on the user satisfaction impact characteristics.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images