Enterprise service recommendation method for smart city

Abstract

The invention discloses an enterprise service recommendation method for a smart city, and the method comprises the following steps: collecting the data of an enterprise, including the name of the enterprise, the establishment time, the registration capital, the operation range, the organization structure, the revenue flow, the profit level, the cost composition, the market share, the sales channel, the customer group distribution, and the industry classification; credit rating and historical service requirements are included at the same time, and the historical service requirements comprise detailed records of various services accepted by enterprises in the past, such as service types, service frequencies, service satisfaction and regional characteristics of smart cities, and cover regional industry planning, policy orientation, infrastructure conditions and talent resource distribution. According to the method, by collecting the multi-dimensional data of the enterprise, the operation condition and demand characteristics of the enterprise can be insighted, a solid foundation is laid for targeted service recommendation, the enterprise portrait is constructed by utilizing a deep learning algorithm fused with multiple advanced technologies, and the depth and precision of enterprise understanding are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of smart city service technology, and more specifically, to a smart city enterprise service recommendation method and method. Background Art

[0002] In today's era of rapid development of digitalization and intelligence, the concept of smart city has emerged. Under the framework of smart city, enterprises, as important economic entities, face many challenges and opportunities in their development and operation.

[0003] With the increasingly fierce market competition and the continuous diversification of corporate needs, the traditional enterprise service recommendation methods have gradually revealed many shortcomings. Previous recommendation methods are usually based on simple enterprise information and limited data dimensions, which makes it difficult to fully and deeply understand the real situation and complex needs of enterprises.

[0004] The collection of corporate data is often limited to some basic static information, such as company name, establishment time, etc., but lacks the integration of dynamic and in-depth operating data such as revenue flow, cost structure, market share, etc., resulting in an incomplete and inaccurate understanding of the company.

[0005] In terms of data analysis and portrait construction, traditional methods mostly use a single analysis model, which is unable to effectively extract deep features and time series features in the data, difficult to capture key dependencies in corporate data, and difficult to accurately reflect the company's true situation and changing trends.

[0006] In terms of service recommendation algorithms, the application of content-based filtering algorithms and collaborative filtering algorithms is relatively isolated and simple, without sufficient integration and optimization, and cannot accurately match the diverse needs of enterprises. Moreover, recommendation strategies often lack real-time and dynamic adjustment capabilities, and cannot respond to market and policy changes and the development needs of enterprises in a timely manner.

[0007] At the same time, there is a lack of a comprehensive system and effective technical means for evaluating and optimizing the effectiveness of service recommendations. It is impossible to comprehensively consider various factors, such as policy impact, market dynamics, and corporate feedback, making it difficult to continuously improve the accuracy and effectiveness of recommendations.

[0008] In addition, there are problems such as security, compliance, and insufficient data value mining in the process of data processing and use. It is impossible to fully utilize advanced means such as natural language processing technology and knowledge graph technology to deeply process and analyze unstructured data, which limits the improvement of corporate portraits and the improvement of recommendation effects.

[0009] In summary, in order to better meet the service needs of enterprises in smart cities and improve the accuracy, real-time and effectiveness of service recommendations, an innovative and comprehensive enterprise service recommendation method is urgently needed. Summary of the invention

[0010] 1. Technical issues to be solved

[0011] The purpose of the present invention is to provide a method and method for recommending enterprise services in a smart city to solve the problems raised in the above-mentioned background technology.

[0012] 2. Technical solution

[0013] A method for recommending enterprise services in a smart city, comprising the following steps:

[0014] S1. Collect enterprise data, including: enterprise name, establishment date, registered capital, business scope, organizational structure, revenue flow, profit level, cost structure, market share, sales channels, customer group distribution, and industry classification, and also include credit rating and historical service needs. The historical service needs include detailed records of various services received by the enterprise in the past, such as service type, service frequency, service satisfaction, and regional characteristics of the smart city where the enterprise is located, covering regional industrial planning, policy orientation, infrastructure status, and talent resource distribution;

[0015] S2. Analyze the collected multi-dimensional data through deep learning algorithms to build corporate portraits. The deep learning algorithms integrate multi-layer convolutional neural networks, recurrent neural networks, and self-attention mechanisms to extract deep features and time series features from the data, capture key dependencies in the data, and use generative adversarial networks to enhance the expressiveness of data features to more accurately reflect the true status of the company.

[0016] S3. Generate a preliminary service recommendation list based on the constructed enterprise portrait and the recommendation model pre-trained with large-scale enterprise data. The recommendation model adopts a hybrid recommendation algorithm that combines a content-based filtering algorithm with a collaborative filtering algorithm. The content-based filtering algorithm makes recommendations by deeply analyzing the precise matching degree between enterprise features in the enterprise portrait and various service features. The collaborative filtering algorithm makes recommendations based on the service selection preferences and behavior patterns of enterprises with similar portraits. At the same time, a reinforcement learning mechanism is introduced to continuously optimize the recommendation strategy based on the feedback of the recommendation results.

[0017] S4. Establish a real-time dynamic adjustment mechanism. By deploying sensors and data collection equipment at key nodes of smart cities, continuously monitor policy changes, market dynamics and real-time feedback from enterprises in smart cities. When monitoring the update and adjustment of policies and regulations, significant changes in market supply and demand, major changes in the industry competition landscape or clear feedback from enterprises, use the event-driven architecture to trigger in-depth evaluation and fine optimization of the preliminary service recommendation list, and generate a service recommendation list. The optimization formula is:

[0018] F=ω1×I+ω2×P+ω3×M+ω4×E+ω5×R+ω6×S

[0019] Among them, F represents the final precise service recommendation list, I represents the preliminary service recommendation list, P represents the impact of policy changes, M represents market dynamic trends, E represents enterprise feedback, R represents the random adjustment factor, S represents the service quality evaluation factor, ω1, ω2, ω3, ω4, ω5, and ω6 are weight coefficients, and ω1+ω2+ω3+ω4+ω5+ω6=1.

[0020] Preferably, the basic information of the enterprise also includes equity structure and branch distribution; operating conditions also include cash flow status and asset-liability status; credit ratings also consider the company's fulfillment of social responsibilities and environmental performance; historical service needs also cover the specific content of the services and the implementation effects.

[0021] Preferably, the content-based filtering algorithm also considers the reputation and word-of-mouth of the service provider, and the collaborative filtering algorithm also makes recommendations based on potential associations and similarities between enterprises.

[0022] Preferably, the real-time dynamic adjustment mechanism is also combined with an artificial intelligence prediction model to predict emerging policy changes and market dynamics and adjust the recommendation list.

[0023] Preferably, the blockchain technology also introduces smart contracts to automatically execute rules for data access and use to ensure compliance with data use.

[0024] Preferably, it also includes using natural language processing technology and knowledge graph technology to process and analyze the company's unstructured data to improve the company's portrait.

[0025] Preferably, the recommendation model performs online real-time learning and dynamic updating by continuously receiving new enterprise data and real-time service feedback, and uses transfer learning and meta-learning techniques to adapt to new data distribution and changes in enterprise needs.

[0026] Preferably, big data analysis technology and machine learning algorithms are used to accurately predict the service demand trends of enterprises in smart cities, and macroeconomic data, industry research reports and social media data are combined to provide guidance for service recommendations.

[0027] Preferably, the recommendation results are presented to the enterprise in the form of multi-dimensional dynamic charts and virtual reality interactive interfaces through visualization technology, incorporating virtual reality and augmented reality elements to facilitate enterprise understanding and selection.

[0028] Preferably, a comprehensive and complete service evaluation system is established to comprehensively collect enterprises' evaluation of the use of recommended services, satisfaction feedback, and quantitative indicators of service effects, and use sentiment analysis and text mining technology to optimize the recommendation model.

[0029] 3. Beneficial effects

[0030] Compared with the prior art, the advantages of the present invention are:

[0031] By extensively and deeply collecting multi-dimensional data of enterprises, the present invention can comprehensively and accurately understand the operating conditions and demand characteristics of enterprises, laying a solid foundation for providing personalized and targeted service recommendations.

[0032] Building a corporate portrait using deep learning algorithms that integrate multiple advanced technologies can not only more accurately reflect the true situation of the company, but also uncover the deep value and potential connections hidden in the data, greatly improving the depth and accuracy of understanding of the company.

[0033] The hybrid recommendation algorithm that combines content-based filtering algorithm with collaborative filtering algorithm and introduces reinforcement learning mechanism can achieve more accurate service recommendations, meet the diversified and dynamically changing needs of enterprises, and improve the satisfaction and adoption rate of enterprises with recommended services.

[0034] A real-time dynamic adjustment mechanism is established to promptly capture key information such as policy changes, market trends, and corporate feedback in smart cities, and quickly optimize the service recommendation list so that the recommended service always maintains a high degree of adaptability and timeliness.

[0035] The application of optimization formulas can scientifically and rationally integrate multiple influencing factors, ensuring that the final service recommendation list generated is more accurate and reliable, and providing enterprises with service recommendations that best suit their current conditions and future development needs.

[0036] The introduction of natural language processing technology and knowledge graph technology to process unstructured data has further enriched and improved the corporate portrait, and improved data utilization efficiency and the accuracy of recommendations.

[0037] The recommendation model can continuously learn and update, using transfer learning and meta-learning techniques to adapt to new data distribution and changes in enterprise needs, ensuring that the recommendation service is always in the best state.

[0038] Big data analysis technology and machine learning algorithms are used to make accurate predictions, and multi-source data are combined to provide comprehensive guidance for service recommendations, enhancing the foresight and strategic nature of the recommendations.

[0039] By displaying recommendation results through innovative visualization technology and incorporating virtual reality and augmented reality elements, the user experience is significantly improved, making it easier for companies to understand and choose recommendation services.

[0040] We have established a comprehensive and complete service evaluation system, and used a variety of technologies to optimize the recommendation model, forming a virtuous cycle of service recommendations and continuously improving service quality and effectiveness.

[0041] In summary, the present invention can significantly improve the quality and efficiency of enterprise service recommendations in smart cities, provide strong support for the development of enterprises, and promote the economic prosperity and sustainable development of smart cities. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a schematic diagram of the enterprise service recommendation method for a smart city of the present invention. DETAILED DESCRIPTION

[0043] Example: See Figure 1 , a method for recommending enterprise services in a smart city, comprising the following steps:

[0044] S1. When collecting enterprise data, the following aspects must be fully and carefully covered:

[0045] The basic information of an enterprise includes the name of the enterprise, which represents the enterprise's logo and brand image; the date of establishment, which reflects the development history and experience accumulation of the enterprise; the registered capital, which reflects the initial investment and financial strength of the enterprise; the business scope, which clarifies the business areas and activity boundaries of the enterprise; and the organizational structure, which shows the internal division of functions and management levels of the enterprise.

[0046] In terms of financial status, it is necessary to obtain revenue flow, which is a direct reflection of the company's operating results; profit level, which is used to measure the profitability of the company; cost structure, which analyzes the distribution of various costs of the company to provide a basis for cost control, and its calculation formula can be expressed as: total cost = fixed cost + variable cost, where fixed costs include rent, equipment depreciation, etc., and variable costs cover raw materials, labor, etc.; market share, to evaluate the company's competitive position in the market, the calculation method can be: company sales ÷ total market sales × 100%; sales channels, to understand the ways and means by which products or services reach customers.

[0047] Customer-related aspects include the distribution of customer groups, understanding the customer's region, age, consumption capacity and other characteristics; historical service needs, including detailed records of various types of services the company has received in the past, such as service types, covering consulting, training, technical support, etc.; service frequency, counting the number of times the service is used in units of months, quarters or years; service satisfaction, measured by scores (such as 1-5 points) or evaluation levels (such as very satisfied, satisfied, dissatisfied, etc.), the calculation formula can be: Service satisfaction = (number of very satisfied × 5 + number of satisfied × 3 + number of dissatisfied × 1) ÷ total number of evaluations × 100%.

[0048] In addition, credit ratings must also be included to comprehensively consider the company's debt repayment ability, contract performance, etc.; at the same time, the regional characteristics of the smart city where the company is located must be considered, covering regional industrial planning, and clarifying the key industrial development directions of the region; policy orientation, grasping the government's support or restriction policies for enterprises; infrastructure conditions, such as transportation, communications, energy supply, etc.; distribution of talent resources, understanding the quantity, quality and distribution of various types of professional talents in the region.

[0049] Through such comprehensive and in-depth data collection, we can provide rich and accurate basic information for subsequent corporate portrait construction, service recommendations, and other work;

[0050] S2. In the process of analyzing the collected multi-dimensional enterprise data to build enterprise portraits, we use advanced and powerful deep learning algorithms. This deep learning algorithm is not a single structure, but a fusion of multi-layer convolutional neural networks (CNN), recurrent neural networks (RNN) and self-attention mechanisms to form a comprehensive deep analysis model.

[0051] Multi-layer convolutional neural networks (CNNs) perform well in processing images and grid-shaped data. They can extract local, spatially correlated deep features from corporate data. For example, for a company's revenue flow chart, CNN can automatically identify revenue patterns and trends over different time periods.

[0052] Recurrent Neural Networks (RNNs) are good at processing sequence data, especially data with time series characteristics. For data that changes over time, such as a company's profit level and cost structure, RNNs can capture the long-term dependencies and dynamic changes in them.

[0053] The self-attention mechanism allows the model to pay more attention to the key parts of the data and assign different weights according to the importance of the data. In complex data such as the company's organizational structure and customer group distribution, the self-attention mechanism can accurately focus on the information that plays a key role in building the company's profile.

[0054] At the same time, we also use generative adversarial networks (GANs) to enhance the expressiveness of data features. GANs consist of a generator and a discriminator. The generator tries to generate realistic data, while the discriminator determines the authenticity of the data. Through the continuous confrontation and optimization of the two, the extracted data features are clearer, more accurate and more representative.

[0055] The specific solution can be: First, preprocess the collected multi-dimensional enterprise data, including data cleaning, normalization and other operations, to make it suitable for input into the deep learning model. Then, input the preprocessed data into the multi-layer convolutional neural network, recurrent neural network and self-attention mechanism for feature extraction. The features extracted by these different types of networks will be fused and interacted to further explore the hidden information in the data. Then, the generative adversarial network is used to enhance and optimize the fused features to improve the quality and expression ability of the features.

[0056] In this process, some evaluation indicators and formulas may be used to measure the performance and effect of the model. For example, the mean square error (MSE) can be used to evaluate the difference between the model prediction results and the actual data. The formula is: Where n is the sample size, yi is the actual value, is the predicted value. Accuracy can also be used to evaluate the accuracy of model classification or prediction. The formula is: Accuracy = (number of correct predictions / total number of predictions) × 100%.

[0057] Through the comprehensive application and continuous optimization of the above deep learning algorithms, the real situation of the enterprise can be reflected more accurately, providing solid technical support and data foundation for the construction of enterprise portraits;

[0058] S3 generates a preliminary list of service recommendations based on a carefully constructed corporate portrait and a recommendation model that has been fully trained with large-scale corporate data.

[0059] This recommendation model adopts a hybrid recommendation algorithm that combines a content-based filtering algorithm with a collaborative filtering algorithm.

[0060] In terms of content-based filtering algorithms, recommendations are made by deeply analyzing the rich corporate features in the corporate portrait and the detailed features of various services, and accurately calculating the matching degree between the two. Specifically, the company's revenue flow, profit level, industry classification and other characteristics are quantitatively compared with the service's scope of application, cost-effectiveness, industry targeting and other characteristics. For example, the cosine similarity formula can be used to calculate the matching degree:

[0061] Cosine similarity = (enterprise feature vector · service feature vector) / (|enterprise feature vector| × |service feature vector|)

[0062] The collaborative filtering algorithm provides recommendations based on the service selection preferences and behavior patterns of companies with similar profiles. By measuring the similarity of corporate profiles, it finds groups of companies with similar operating conditions, demand characteristics, and development stages. Then, it refers to the services selected by these similar companies in the past to infer the services that the current company may be interested in.

[0063] At the same time, in order to continuously improve the recommendation effect, a reinforcement learning mechanism is also introduced. The recommendation effect is evaluated based on the feedback information obtained from each recommendation result, such as user clicks, browsing time, purchase behavior, etc. A reward function is used to quantify the success of the recommendation, such as:

[0064] Reward = click-through rate × weight 1 + purchase conversion rate × weight 2 + browsing time × weight 3 Then, based on these reward feedbacks, continuously adjust the parameters and strategies of the recommendation model to optimize the recommendation results, so that the recommendation service is more in line with the actual needs and potential expectations of the enterprise.

[0065] The specific solution includes: First, digitally represent the enterprise portrait and service features to construct feature vectors. Then, use the content-based filtering algorithm to calculate the matching degree and filter out the services that are initially matched. Next, use the collaborative filtering algorithm to find out the service preferences of similar enterprises. Finally, adjust the recommendation strategy based on the feedback of the reinforcement learning mechanism. Through this cyclic optimization process, the accuracy and practicality of the recommendations are continuously improved.

[0066] S4. Establish a real-time dynamic adjustment mechanism. By deploying sensors and data collection equipment at key nodes of smart cities, continuously monitor policy changes, market dynamics and real-time feedback from enterprises in smart cities. When monitoring the update and adjustment of policies and regulations, significant changes in market supply and demand, major changes in the industry competition landscape or clear feedback from enterprises, use the event-driven architecture to trigger in-depth evaluation and fine optimization of the preliminary service recommendation list, and generate a service recommendation list. The optimization formula is:

[0067] F=ω1×I+ω2×P+ω3×M+ω4×E+ω5×R+ω6×S

[0068] Among them, F represents the final precise service recommendation list, I represents the preliminary service recommendation list, P represents the impact of policy changes, M represents market dynamic trends, E represents enterprise feedback, R represents the random adjustment factor, S represents the service quality evaluation factor, ω1, ω2, ω3, ω4, ω5, and ω6 are weight coefficients, and ω1+ω2+ω3+ω4+ω5+ω6=1.

[0069] In the present invention, the basic information of the enterprise also includes the equity structure and the distribution of branches; the operating status also includes the cash flow status and the assets and liabilities status; the credit rating also considers the fulfillment of the enterprise's social responsibility and environmental performance; the historical service demand also covers the specific content and implementation effect of the service.

[0070] In the present invention, the content-based filtering algorithm also considers the reputation and word-of-mouth of the service provider, and the collaborative filtering algorithm also makes recommendations based on the potential associations and similarities between enterprises.

[0071] In the present invention, the real-time dynamic adjustment mechanism is also combined with an artificial intelligence prediction model to predict emerging policy changes and market dynamics and adjust the recommendation list.

[0072] In the present invention, blockchain technology also introduces smart contracts to automatically execute the rules for data access and use to ensure the compliance of data use.

[0073] The present invention also includes using natural language processing technology and knowledge graph technology to process and analyze the unstructured data of the enterprise to improve the enterprise portrait.

[0074] In the present invention, the recommendation model performs online real-time learning and dynamic updating by continuously receiving new enterprise data and real-time service feedback, and uses transfer learning and meta-learning techniques to adapt to new data distribution and changes in enterprise needs.

[0075] In the present invention, big data analysis technology and machine learning algorithms are used to accurately predict the service demand trends of enterprises in smart cities, and macroeconomic data, industry research reports and social media data are combined to provide guidance for service recommendations.

[0076] In the present invention, the recommendation results are presented to the enterprise in the form of multi-dimensional dynamic charts and virtual reality interactive interfaces through visualization technology, and virtual reality and augmented reality elements are integrated to facilitate the enterprise's understanding and selection.

[0077] In the present invention, a comprehensive and perfect service evaluation system is established to comprehensively collect enterprises' evaluation of the use of recommended services, satisfaction feedback, and quantitative indicators of service effects, and use sentiment analysis and text mining technology to optimize the recommendation model.

[0078] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for recommending enterprise services in a smart city, characterized in that: The following steps are involved: S1. Collect enterprise data, including: enterprise name, establishment date, registered capital, business scope, organizational structure, revenue flow, profit level, cost structure, market share, sales channels, customer group distribution, and industry classification, and also include credit rating and historical service needs. The historical service needs include detailed records of various services received by the enterprise in the past, such as service type, service frequency, service satisfaction, and regional characteristics of the smart city where the enterprise is located, covering regional industrial planning, policy orientation, infrastructure status, and talent resource distribution; S2. Analyze the collected multi-dimensional data through deep learning algorithms to build corporate portraits. The deep learning algorithms integrate multi-layer convolutional neural networks, recurrent neural networks, and self-attention mechanisms to extract deep features and time series features from the data, capture key dependencies in the data, and use generative adversarial networks to enhance the expressiveness of data features to more accurately reflect the true status of the company. S3. Generate a preliminary service recommendation list based on the constructed enterprise portrait and the recommendation model pre-trained with large-scale enterprise data. The recommendation model adopts a hybrid recommendation algorithm that combines a content-based filtering algorithm with a collaborative filtering algorithm. The content-based filtering algorithm makes recommendations by deeply analyzing the precise matching degree between enterprise features in the enterprise portrait and various service features. The collaborative filtering algorithm makes recommendations based on the service selection preferences and behavior patterns of enterprises with similar portraits. At the same time, a reinforcement learning mechanism is introduced to continuously optimize the recommendation strategy based on the feedback of the recommendation results. S4. Establish a real-time dynamic adjustment mechanism. By deploying sensors and data collection equipment at key nodes of smart cities, continuously monitor policy changes, market dynamics and real-time feedback from enterprises in smart cities. When monitoring the update and adjustment of policies and regulations, significant changes in market supply and demand, major changes in the industry competition landscape or clear feedback from enterprises, use the event-driven architecture to trigger in-depth evaluation and fine optimization of the preliminary service recommendation list, and generate a service recommendation list. The optimization formula is: F=ω1×I+ω2×P+ω3×M+ω4×E+ω5×R+ω6×S Among them, F represents the final precise service recommendation list, I represents the preliminary service recommendation list, P represents the impact of policy changes, M represents market dynamic trends, E represents enterprise feedback, R represents the random adjustment factor, S represents the service quality evaluation factor, ω1, ω2, ω3, ω4, ω5, and ω6 are weight coefficients, and ω1+ω2+ω3+ω4+ω5+ω6=1.

2. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: The basic information of the enterprise also includes its equity structure and branch distribution; its operating conditions also include cash flow and assets and liabilities; its credit rating also takes into account its fulfillment of social responsibilities and environmental performance; and its historical service needs also cover the specific content of the services and their implementation effects.

3. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: The content-based filtering algorithm also considers the reputation and word-of-mouth of the service provider, and the collaborative filtering algorithm also makes recommendations based on the potential associations and similarities between enterprises.

4. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: The real-time dynamic adjustment mechanism is also combined with an artificial intelligence prediction model to predict policy changes and market dynamics, and adjust the recommendation list.

5. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: The blockchain technology also introduces smart contracts to automatically enforce rules for data access and usage, ensuring compliance with data usage.

6. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: It also includes using natural language processing technology and knowledge graph technology to process and analyze the company's unstructured data to improve the company's profile.

7. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: The recommendation model performs online real-time learning and dynamic updates by continuously receiving new enterprise data and real-time service feedback, and uses transfer learning and meta-learning techniques to adapt to new data distribution and changes in enterprise needs.

8. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: Big data analysis technology and machine learning algorithms are used to accurately predict the service demand trends of enterprises in smart cities, and macroeconomic data, industry research reports and social media data are combined to provide guidance for service recommendations.

9. The enterprise service recommendation method for a smart city according to claim 1, characterized in that: Through visualization technology, the recommendation results are presented to enterprises in the form of multi-dimensional dynamic charts and virtual reality interactive interfaces, integrating virtual reality and augmented reality elements to facilitate enterprise understanding and selection.

10. The enterprise service recommendation method of a smart city according to claim 1, characterized in that: Establish a comprehensive and complete service evaluation system, comprehensively collect enterprises' evaluation of recommended services, satisfaction feedback, and quantitative indicators of service effectiveness, and use sentiment analysis and text mining technology to optimize the recommendation model.

Images