Internet information service quality evaluation method and system based on artificial intelligence

Abstract

The invention discloses an Internet information service quality evaluation method and system based on artificial intelligence, and relates to the technical field of information service evaluation, and the method comprises the steps: collecting user behavior data, user feedback data and system performance data through a network, carrying out the preprocessing, and extracting data features; constructing a service evaluation model, evaluating the service quality based on the data features, and optimizing the Internet information service according to an evaluation result; the optimization result is applied and stored in a database; and carrying out self-learning optimization on the service evaluation model through reinforcement learning. According to the method, the user behavior data, the user feedback data and the system performance data are collected to respectively extract features and combine the features to form the data features, so that the accuracy and correlation of feature extraction are greatly improved, and the data authenticity of service quality evaluation is ensured; and the accuracy of service quality evaluation is improved by constructing the service quality evaluation model, and the problem of service quality optimization is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of information service evaluation, and particularly to an Internet information service quality evaluation method and system based on artificial intelligence. Background Art

[0002] In recent years, with the rapid development of Internet technology, the complexity and diversity of information services have increased significantly, and users' requirements for service quality have become more stringent. Traditional Internet information service quality evaluation methods mainly rely on static index calculation or rule-based analysis means. For example, by statistically analyzing single indicators such as page views, response time, and user satisfaction, a preliminary conclusion about service quality is obtained. However, these methods cannot dynamically adapt to the ever-changing Internet environment, and it is difficult to comprehensively capture multi-dimensional information of user behavior, feedback data, and system performance. This leads to one-sided or lagging conclusions when traditional evaluation models face complex service scenarios. In addition, existing methods often lack the ability to deeply mine historical data and are difficult to optimize future service quality based on existing data. In recent years, with the wide application of artificial intelligence (AI) technology, especially the emergence of intelligent means such as deep learning and reinforcement learning, new ideas have been provided to solve this problem. However, existing technologies often only focus on a single data source (such as user feedback or system performance) when evaluating service quality, and fail to fully integrate multi-dimensional data characteristics of users and systems. This results in limitations in the integrity and accuracy of evaluation results, and there is a lack of optimization for service quality, and the organic combination of service evaluation and service optimization cannot be achieved. Summary of the Invention

[0003] In view of the above existing problems, the present invention is proposed.

[0004] Therefore, the present invention provides an Internet information service quality evaluation method and system based on artificial intelligence, which solves the problems of failure to fully integrate multi-dimensional data characteristics of users and systems and lack of optimization for service quality.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides an Internet information service quality evaluation method based on artificial intelligence, which includes:

[0007] Collecting user behavior data, user feedback data, and system performance data through the network for preprocessing and extracting data features;

[0008] Constructing a service evaluation model to evaluate service quality based on data features and optimizing Internet information services according to the evaluation results;

[0009] Applying the optimization results and storing them in a database;

[0010] Self - learning optimization of the service evaluation model through reinforcement learning.

[0011] As a preferred embodiment of the method for evaluating the quality of Internet information services based on artificial intelligence according to the present invention, wherein: the pre - processing of collecting user behavior data, user feedback data, and system performance data through the network means collecting user behavior data, user feedback data, and system performance data through the network, supplementing missing values using the K - nearest neighbor interpolation method, and normalizing the collected data after denoising using a Kalman filter.

[0012] As a preferred embodiment of the method for evaluating the quality of Internet information services based on artificial intelligence according to the present invention, wherein: the extraction of data features includes extracting user behavior features, user emotion features, and system performance features respectively according to the pre - processed user behavior data, user feedback data, and system performance data;

[0013] The extraction of user behavior features means constructing a behavior feature matrix X from the collected user behavior data:

[0014]

[0015] where T is the time series and D is the data type in the user behavior data;

[0016] Using a recurrent neural network (RNN) to construct a behavior feature extraction model, setting the model input as the behavior feature matrix and the output as the user behavior features, training the behavior feature extraction model based on the training data, and inputting the behavior feature matrix X into the trained behavior feature extraction model to obtain the user behavior features;

[0017] The extraction of user emotion features means using a pre - trained Word2Vec model to convert the user feedback data into word vectors y, combining all the word vectors y to form a feature vector Y of the user feedback data, and constructing a wave function based on the feature vector Y :

[0018]

[0019] where is the mean vector of the word vectors, is the standard deviation;

[0020] According to the wave function defining a super - potential function :

[0021]

[0022] where q is an adjustment coefficient, is the adjustment frequency, is the wave function derivative;

[0023] Through the superpotential function Finally, the potential function is obtained:

[0024]

[0025] where is the derivative of the superpotential function derivative;

[0026] Construct a bidirectional LSTM network and train it. Respectively input into the forward and backward directions of the bidirectional LSTM network to obtain the forward hidden state and the backward hidden state, and combine the forward hidden state and the backward hidden state to obtain the bidirectional hidden state as the user's emotional feature;

[0027] The extraction of the system performance feature means representing the collected system performance data as a time series matrix S:

[0028]

[0029] where T is the time series and I is the data type of the system performance data;

[0030] Define the time window size as t*I. Based on the time window, randomly generate a dimensional vector s from the time series matrix as the initial individual, form an initial population with the initial individuals, and calculate the fitness of the initial individuals in the initial population:

[0031]

[0032] where is the fitness of the i-th initial individual, is the classification error of the i-th individual using the KNN classifier, is the weight factor, is the i-th initial individual, and t is the time window length;

[0033] Use the Aquila optimizer to optimize the initial population, and select the initial individuals for update iteration according to the fitness. Stop the iteration when the change in fitness converges, and use the population individual with the highest fitness in the population after stopping the iteration as the system performance feature;

[0034] Concatenate the user behavior feature, the user emotional feature, and the system performance feature to form the data feature Z.

[0035] As a preferred solution of the Internet information service quality assessment method based on artificial intelligence according to the present invention, wherein: the construction of the service evaluation model for evaluating the service quality based on data features means mapping the data features Z through a non-linear function:

[0036]

[0037] where is the mapping result, Q is the weight matrix, and b is the bias vector;

[0038] Define a performance evaluation function based on queuing theory for the mapping result :

[0039]

[0040] Construct a service evaluation model P through the performance evaluation function and the mapping result:

[0041]

[0042] where and are weight coefficients;

[0043] Use the gradient descent method to optimize Q, b, and , and calculate the scoring error of the service evaluation model as the optimization target for iterative optimization. When the scoring error converges, stop the iteration and output the optimal Q, b, and Apply them to the non-linear function and the service evaluation model;

[0044] Input the data feature Z into the trained service evaluation model to obtain the service quality score, and compare the service quality score with the set scoring threshold. If the service quality score is higher than the scoring threshold, the service quality is good; otherwise, the service quality is poor.

[0045] As a preferred embodiment of the method for evaluating the quality of Internet information services based on artificial intelligence according to the present invention, wherein: optimizing the Internet information service according to the evaluation result means that if the service quality is poor, a genetic population is formed based on the system performance data, a CNN model is constructed, the input of the CNN model is set as the system performance data, and the output is the corresponding user behavior data and user feedback data. The CNN model is trained and the service quality score is calculated by obtaining the user behavior data and user feedback data corresponding to the system performance data in the genetic population through the CNN model. The service quality score is used as the genetic fitness, and the roulette method is used to select genetic individuals from the genetic population for crossover and mutation, and the iteration is repeated until the genetic fitness converges, and the system performance data with the highest fitness in the iterated genetic population is output as the final system performance data, and the performance parameters in the final system performance data are extracted as the optimization scheme.

[0046] As a preferred embodiment of the method for evaluating the quality of Internet information services based on artificial intelligence according to the present invention, wherein: applying the optimization result and storing it in the database means comparing the fitness of the final system performance data with the fitness of the current system performance data. If the fitness of the final system performance data is higher than the fitness of the current system performance data, the optimization scheme is applied; otherwise, optimization is performed again.

[0047] After the optimization scheme is applied, an application record is generated and stored in the database synchronously with the system performance data.

[0048] As a preferred embodiment of the method for evaluating the quality of Internet information services based on artificial intelligence according to the present invention, wherein: self-learning and optimizing the service evaluation model through reinforcement learning means regularly evaluating the accuracy of the service evaluation model. When the accuracy is lower than the preset standard, the reinforcement learning algorithm is used to optimize the service evaluation model until the accuracy is higher than the preset standard.

[0049] In a second aspect, the present invention provides an Internet information service quality evaluation system based on artificial intelligence, including

[0050] A data processing module, configured to collect user behavior data, user feedback data, and system performance data for preprocessing and extract data features;

[0051] An evaluation and optimization module, configured to construct a service evaluation model and perform service quality evaluation based on the extracted data features, and optimize the service quality according to the evaluation result;

[0052] A record storage module, configured to record and store the optimization scheme and system performance data;

[0053] An adaptive learning module, configured to evaluate the accuracy of the service evaluation model and optimize the service evaluation model through a reinforcement learning algorithm.

[0054] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the method for evaluating the quality of Internet information services based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0055] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the method for evaluating the quality of Internet information services based on artificial intelligence as described in the first aspect of the present invention is implemented.

[0056] The beneficial effects of the present invention are as follows: by collecting user behavior data, user feedback data, and system performance data, extracting features respectively and combining them to form data features, the accuracy and relevance of feature extraction are greatly improved, ensuring the authenticity of data for service quality evaluation, and by constructing a service quality evaluation model, the accuracy of service quality evaluation is improved, effectively solving the problem of service quality optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0058] Figure 1 It is a flowchart of the method for evaluating the quality of Internet information services based on artificial intelligence in Embodiment 1.

[0059] Figure 2 It is a structural diagram of the system for evaluating the quality of Internet information services based on artificial intelligence in Embodiment 1. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will give a detailed description of the specific embodiments of the present invention in conjunction with the drawings in the specification.

[0061] Many specific details are set forth in the following description to facilitate a thorough understanding of the present invention, but the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the spirit of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0062] Second, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation manner of the present invention. The phrase "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0063] Embodiment 1, referring to Figure 1 and Figure 2 , is the first embodiment of the present invention. This embodiment provides an artificial intelligence-based method for evaluating the quality of Internet information services, including the following steps:

[0064] S1. Collect user behavior data, user feedback data, and system performance data through the network, perform preprocessing, and extract data features;

[0065] Specifically, collecting user behavior data, user feedback data, and system performance data through the network for preprocessing means collecting user behavior data, user feedback data, and system performance data through the network. The user behavior data includes user click data, page view data, residence time data, click paths, etc. The user feedback data includes user evaluations, satisfaction survey results, etc. The system performance data includes server response time, bandwidth usage, page load time, etc. Use the K-nearest neighbor interpolation method to supplement missing values and use the Kalman filter to denoise the collected data and then perform normalization.

[0066] At the same time, collect user behavior data, user feedback data, and system performance data to form a multi-dimensional basic data set. This panoramic data collection method overcomes the limitation of single data dimension in the prior art and provides comprehensive basic data support for subsequent evaluations. Using the K-nearest neighbor interpolation method can effectively supplement missing values while retaining the local features of the data. Compared with simple mean filling or zero-value substitution, the K-nearest neighbor interpolation method significantly improves the rationality and accuracy of data completion. During the data collection process, it is inevitable to mix in noises such as network jitter and device errors. The Kalman filter effectively removes noises by dynamically estimating the true value of the data, enhancing the credibility of the data. Compared with traditional fixed filters, the Kalman filter can adapt to data characteristics and is particularly suitable for processing user behavior data and system performance data with time series characteristics. Multi-dimensional data sources have different measurement units (such as time, score, frequency, etc.). Directly using unnormalized data to input into the model may lead to evaluation biases. Normalization effectively solves this problem, making the data from different sources consistent in the model. Normalization not only improves the calculation efficiency of the model but also avoids unfair weight allocation caused by data dimension differences.

[0067] Further, extracting data features includes extracting user behavior features, user sentiment features, and system performance features from the preprocessed user behavior data, user feedback data, and system performance data respectively;

[0068] The extraction of user behavior features refers to constructing a behavior feature matrix X from the collected user behavior data:

[0069]

[0070] where T is the time series and D is the data type in the user behavior data;

[0071] Use a recurrent neural network (RNN) to construct a behavior feature extraction model. Set the model input as the behavior feature matrix and the output as the user behavior features. Train the behavior feature extraction model based on the training data, and input the behavior feature matrix X into the trained behavior feature extraction model to obtain the user behavior features;

[0072] The extraction of user sentiment features refers to using a pre-trained Word2Vec model to convert the user feedback data into word vectors y, combining all the word vectors y to form a feature vector Y of the user feedback data, and constructing a wave function based on the feature vector Y :

[0073]

[0074] where is the mean vector of the word vectors, is the standard deviation;

[0075] According to the wave function define a superpotential function :

[0076]

[0077] where q is an adjustment coefficient, is the adjustment frequency, obtained from historical data, is the wave function 's derivative;

[0078] Through the superpotential function finally obtain the potential function :

[0079]

[0080] where is the derivative of the superpotential function ;

[0081] Construct a bidirectional LSTM network and train it. Respectively, The forward and backward directions of the input bidirectional LSTM network are used to obtain the forward hidden state and the backward hidden state, and the forward hidden state and the backward hidden state are combined to obtain the bidirectional hidden state as the user's emotional feature;

[0082] The extraction of the system performance feature means representing the collected system performance data as a time series matrix S:

[0083]

[0084] where T is the time series and I is the data type of the system performance data;

[0085] Define the time window size as t * I. Randomly generate a dimensional vector s from the time series matrix based on the time window as the initial individual, form the initial population with the initial individuals, and calculate the fitness of the initial individuals in the initial population:

[0086]

[0087] where is the fitness of the i-th initial individual, is the classification error of the i-th individual using the KNN classifier, is the weight factor, is the i-th initial individual, and t is the time window length;

[0088] Use the Aquila optimizer to optimize the initial population, and select the initial individuals for update iteration according to the fitness. Stop the iteration when the change in fitness converges. Take the population individual with the highest fitness in the population after stopping the iteration as the system performance feature;

[0089] Concatenate the user behavior feature, the user emotional feature, and the system performance feature to form the data feature Z.

[0090] A recurrent neural network (RNN) is used to construct a behavioral feature extraction model to effectively capture the time-dependent relationships in user behavior data. The RNN can recognize the potential rules and patterns of user behavior by memorizing historical information in the sequence through the cyclic structure of hidden states, solving the problem of insufficient understanding of dynamic behavior data by traditional methods. Feature vectors are constructed through the Word2Vec model, and the dynamic change characteristics of user emotions are characterized by wave functions and their derived hyper-potential functions and potential functions. Compared with traditional static sentiment analysis methods (such as sentiment dictionary matching), this method can quantify the fluctuation intensity and stability of user emotions through potential functions, improving the depth and flexibility of sentiment feature extraction. The Aquila optimizer is used to iteratively optimize the initial population, and the best individual is selected as the system performance feature through the fitness function, solving the problem that traditional methods are difficult to efficiently extract the core features of performance data. The Aquila optimizer balances the global search and local development capabilities during population iteration and can quickly converge to the optimal solution. User behavior features, user emotion features, and system performance features are concatenated into comprehensive data features, breaking the limitation of single data source analysis in traditional methods and constructing a multi-dimensional and multi-level service quality evaluation data foundation. This method realizes the organic combination of data at the three levels of behavior, emotion, and performance, ensuring that the evaluation results are more comprehensive and accurate.

[0091] S2. Construct a service evaluation model to evaluate the service quality based on data features and optimize the Internet information service according to the evaluation results;

[0092] Specifically, constructing a service evaluation model to evaluate the service quality based on data features means mapping the data feature Z through a non-linear function:

[0093]

[0094] where is the mapping result, Q is the weight matrix, and b is the bias vector, which are obtained by fitting historical data;

[0095] Define a performance evaluation function based on queuing theory for the mapping result :

[0096]

[0097] Through the performance evaluation function and the mapping result, a service evaluation model P is constructed:

[0098]

[0099] where and are weight coefficients, which are determined through experiments or optimization, and ;

[0100] Optimize Q, b, and by using the gradient descent method, and calculate the scoring error of the service evaluation model as the optimization objective for iterative optimization. Stop the iteration when the scoring error converges, and output the optimal Q, b, and and apply them to the non-linear function and the service evaluation model;

[0101] Input the data feature Z into the trained service evaluation model to obtain the service quality score. Compare the service quality score with the set scoring threshold. If the service quality score is higher than the scoring threshold, the service quality is good; otherwise, it is poor.

[0102] The non-linear function maps the comprehensive features, effectively extracting the non-linear associations and potential patterns in the data features. Compared with traditional linear mapping methods, the non-linear function can capture the higher-order relationships between complex features more accurately, thus providing more accurate and comprehensive input data for subsequent performance evaluation and service quality assessment. The performance evaluation function combines the core indicators in queuing theory, which not only reflects the utilization rate of system resources but also quantifies the response efficiency of the service. Compared with traditional single-index evaluation methods, this function can measure service performance from multiple perspectives, significantly improving the scientificity and comprehensiveness of the evaluation model. In addition, the dynamics of the performance evaluation function enable it to adapt to changes in data input in real time, improving the robustness of service evaluation. The service evaluation model constructs a comprehensive evaluation framework with both feature extraction and performance analysis functions by combining the non-linear mapping results and the performance evaluation function. The weight coefficients and can dynamically balance the influence of data characteristics and system performance, ensuring that the model can generate reasonable service quality scores in different scenarios. Compared with static parameter settings, the dynamic optimization process of the gradient descent method can adapt to real-time changing data input, improving the prediction accuracy and learning ability of the model. Comparing the service quality score with the set threshold not only clarifies the criteria for the quality of service but also provides a direct guiding basis for further optimization and improvement. This step realizes the operability of the evaluation results, making the optimization of service quality have a clear direction.

[0103] Further, optimizing Internet information services according to the evaluation results means that if the service quality is poor, a genetic population is formed based on system performance data, a CNN model is constructed, the input of the CNN model is set as the system performance data, and the output is the corresponding user behavior data and user feedback data. The CNN model is trained, and the service quality score is calculated using the user behavior data and user feedback data corresponding to the system performance data in the genetic population. The service quality score is used as the genetic fitness. The roulette wheel method is used to select genetic individuals from the genetic population for crossover and mutation, and the iteration is repeated until the genetic fitness converges. The system performance data with the highest fitness in the iterated genetic population is output as the final system performance data, and the performance parameters in the final system performance data are extracted as the optimization solution.

[0104] By constructing a genetic population, the initial system performance data is diversified into a set of candidate solutions. This diversity provides a larger solution space for subsequent optimization, effectively avoiding the generation of local optimal solutions. At the same time, the setting of the genetic population is directly related to the service quality score, providing a clear goal orientation for the subsequent evolution process. By establishing the mapping relationship between system performance data and user behavior data and user feedback data through the CNN model, it can more intuitively reflect the impact of system performance on user experience. Compared with traditional linear or rule-based mapping methods, the CNN model uses its powerful modeling ability for nonlinear relationships to significantly improve the accuracy and adaptability of the mapping. The service quality score calculated by the CNN model is directly used as the genetic fitness, making the optimization process of the genetic algorithm highly consistent with the service quality goal. Compared with the abstract fitness definition in traditional genetic algorithms, this scoring method is more in line with the actual application requirements, ensuring the rationality and scientificity of the optimization direction. The roulette wheel method maintains the diversity of the population while retaining high-fitness individuals through a probability selection mechanism, avoiding optimization failure caused by premature convergence. Its introduction in the genetic algorithm significantly improves the global search ability of the algorithm. Through iterative optimization and fitness convergence judgment, the algorithm can gradually approach the optimal solution in dynamic adjustment. When the change in fitness stops, the optimal individual in the population is the final optimization result. This dynamic optimization mechanism overcomes the limitation of static optimization methods that cannot adapt to real-time changing data.

[0105] S3. Apply the optimization result and store it in the database;

[0106] Specifically, applying the optimization result and storing it in the database means comparing the fitness of the final system performance data with the fitness of the current system performance data. If the fitness of the final system performance data is higher than that of the current system performance data, the optimization solution is applied; otherwise, optimization is performed again.

[0107] After the optimization solution is applied, an application record is generated and stored in the database synchronously with the system performance data.

[0108] By comparing the fitness of the final system performance data with that of the current system performance data, it is ensured that only better optimization solutions will be applied. This dynamic evaluation mechanism solves the problem of the lack of a scientific evaluation standard in traditional optimization methods, avoiding resource waste and service quality degradation caused by blind application. When the fitness of the final system performance data does not exceed that of the current system performance data, the system will re-optimize. This recursive optimization method ensures the continuity and accuracy of the optimization process, avoiding the service quality bottleneck problem caused by insufficient single optimization. After each application of the optimization solution, the generated application records are synchronously stored in the database, providing a reliable basis for subsequent optimization evaluation and historical data traceability. Compared with the problem of the lack of a recording mechanism in traditional optimization methods, this step significantly improves the traceability and transparency of the optimization process.

[0109] S4. Self-learn and optimize the service evaluation model through reinforcement learning;

[0110] Specifically, self-learning and optimizing the service evaluation model through reinforcement learning means regularly evaluating the accuracy of the service evaluation model. When the accuracy is lower than the preset standard, the reinforcement learning algorithm is used to optimize the service evaluation model until the accuracy is higher than the preset standard.

[0111] This embodiment also provides an Internet information service quality evaluation system based on artificial intelligence, including:

[0112] A data processing module, configured to collect user behavior data, user feedback data, and system performance data for preprocessing and extract data features;

[0113] An evaluation and optimization module, configured to build a service evaluation model and perform service quality evaluation based on the extracted data features, and at the same time optimize the service quality according to the evaluation results;

[0114] A record storage module, configured to record and store optimization solutions and system performance data;

[0115] An adaptive learning module, configured to evaluate the accuracy of the service evaluation model and optimize the service evaluation model through the reinforcement learning algorithm.

[0116] This embodiment also provides a computer device applicable to the case of the Internet information service quality evaluation method based on artificial intelligence, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the Internet information service quality evaluation method based on artificial intelligence as proposed in the above embodiment.

[0117] The computer device can be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be achieved through WIFI, carrier network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the outer shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0118] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for evaluating the quality of Internet information services based on artificial intelligence as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0119] In summary, the present invention combines the features extracted from the user behavior data, user feedback data, and system performance data respectively to form data features, greatly improving the accuracy and relevance of feature extraction, ensuring the authenticity of the data for service quality evaluation, and improving the accuracy of service quality evaluation by constructing a service quality evaluation model, effectively solving the problem of service quality optimization.

[0120] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. An artificial intelligence-based Internet information service quality assessment method, characterized in that: include, Collect user behavior data, user feedback data, and system performance data through the network for preprocessing and extracting data features; Build a service evaluation model to evaluate service quality based on data characteristics, and optimize Internet information services based on the evaluation results; Apply the optimization results and store them in the database; Self-learning optimization of service evaluation model through reinforcement learning.

2. The method for evaluating the quality of Internet information services based on artificial intelligence as claimed in claim 1, characterized in that: The preprocessing of collecting user behavior data, user feedback data and system performance data through the network refers to collecting user behavior data, user feedback data and system performance data through the network, supplementing missing values ​​using the K nearest neighbor interpolation method, and using the Kalman filter to denoise the collected data and then normalize it.

3. The method for evaluating the quality of Internet information services based on artificial intelligence as claimed in claim 2, characterized in that: The extracting data features comprises extracting user behavior features, user emotion features and system performance features respectively according to the pre-processed user behavior data, user feedback data and system performance data; Extracting user behavior features refers to constructing a behavior feature matrix X from the collected user behavior data: ； Where T is the time series, and D is the data type in the user behavior data; Use recurrent neural network RNN ​​to build a behavior feature extraction model, set the model input as the behavior feature matrix, and the output as the user behavior feature. Train the behavior feature extraction model based on the training data, and input the behavior feature matrix X into the trained behavior feature extraction model to obtain the user behavior feature. Extracting user emotion features refers to using a pre-trained Word2Vec model to convert user feedback data into word vectors y, combining all word vectors y to form a feature vector Y of the user feedback data, and constructing a wave function based on the feature vector Y. : ； in is the mean vector of word vectors, is the standard deviation; According to the wave function Define the superpotential function : ； Where q is the adjustment coefficient, To adjust the frequency, is the wave function The derivative of Through the superpotential function Finally, the potential function is obtained : ； in is the superpotential function The derivative of Build a bidirectional LSTM network and train it. Input the forward and reverse directions of the bidirectional LSTM network to obtain the forward hidden state and the reverse hidden state, and then combine the forward hidden state and the reverse hidden state to obtain the bidirectional hidden state as the user emotion feature; Extracting system performance characteristics refers to representing the collected system performance data as a time series matrix S: ； Where T is the time series, and I is the data type of system performance data; Define the time window size as t*I, randomly generate a dimension vector s from the time series matrix based on the time window as the initial individual, form the initial individuals into an initial population, and calculate the fitness of the initial individuals in the initial population: ； in is the fitness of the i-th initial individual, The classification error of the i-th individual using the KNN classifier, is the weight factor, is the i-th initial individual, t is the length of the time window; Use Aquila optimizer to optimize the initial population, and select the initial individuals for update iteration according to fitness. Stop iteration when fitness changes converge, and use the individuals with the highest fitness in the population after stopping iteration as the system performance feature. The user behavior characteristics, user emotion characteristics and system performance characteristics are spliced ​​together to form data feature Z.

4. The method for evaluating the quality of Internet information services based on artificial intelligence as claimed in claim 3, characterized in that: The construction of the service evaluation model to evaluate the service quality based on data features refers to mapping the data feature Z through a nonlinear function: ； in is the mapping result, Q is the weight matrix, and b is the bias vector; Define performance evaluation function for mapping results based on queuing theory : ； Through the performance evaluation function The service evaluation model P is constructed by combining the mapping results: ； in and is the weight coefficient; The gradient descent method is used to optimize Q, b, as well as , and calculate the scoring error of the service evaluation model as the optimization target for iterative optimization. When the scoring error converges, stop the iteration and output the optimal Q, b, as well as Application to nonlinear functions and service evaluation models; The data feature Z is input into the trained service evaluation model to obtain the service quality score, and the service quality score is compared with the set score threshold. If the service quality score is higher than the score threshold, the service quality is good, otherwise the service quality is poor.

5. The method for evaluating the quality of Internet information services based on artificial intelligence as claimed in claim 4, characterized in that: The optimizing the Internet information service according to the evaluation result means that if the service quality is poor, a genetic population is formed based on the system performance data, a CNN model is constructed, the input of the CNN model is set as the system performance data, and the output is the corresponding user behavior data and user feedback data, the CNN model is trained, and the user behavior data and user feedback data corresponding to the system performance data in the genetic population are obtained through the CNN model to calculate the service quality score, the service quality score is used as the genetic fitness, a roulette method is used to select genetic individuals from the genetic population for crossover mutation, and iterations are repeated until the genetic fitness converges, the system performance data with the highest fitness in the iterated genetic population is output as the final system performance data, and the performance parameters in the final system performance data are extracted as the optimization scheme.

6. The method for evaluating the quality of Internet information services based on artificial intelligence as claimed in claim 5, characterized in that: The said applying the optimization result and storing it in the database means comparing the fitness of the final system performance data with the fitness of the current system performance data, if the fitness of the final system performance data is higher than the fitness of the current system performance data, then applying the optimization scheme, otherwise re-optimizing; After the optimization plan is applied, the application records generated and the system performance data are synchronously stored in the database.

7. The method for evaluating the quality of Internet information services based on artificial intelligence as claimed in claim 6, characterized in that: The self-learning optimization of the service evaluation model through reinforcement learning refers to periodically evaluating the accuracy of the service evaluation model, and when the accuracy is lower than a preset standard, optimizing the service evaluation model using a reinforcement learning algorithm until the accuracy is higher than the preset standard.

8. An artificial intelligence-based Internet information service quality assessment system, based on the artificial intelligence-based Internet information service quality assessment method according to any one of claims 1 to 7, characterized in that: include, Data processing module, used to collect user behavior data, user feedback data and system performance data for preprocessing and extracting data features; Evaluation and optimization module, which is used to build a service evaluation model and evaluate the service quality based on the extracted data features, and optimize the service quality according to the evaluation results; A record storage module is used to record and store optimization plans and system performance data; The adaptive learning module is used to evaluate the accuracy of the service evaluation model and optimize the service evaluation model through the reinforcement learning algorithm.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the Internet information service quality assessment method based on artificial intelligence described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for evaluating the quality of Internet information services based on artificial intelligence described in any one of claims 1 to 7 are implemented.

Images