Intelligent recommendation system based on product picture visual presentation

Abstract

The invention belongs to the technical field of product recommendation management, and particularly relates to an intelligent recommendation system based on product picture visual presentation, which comprises an intelligent recommendation unit, an operation management analysis unit and a management maintenance end, the intelligent recommendation unit comprises a product request image uploading module, an image visual processing module, an image matching and screening module and a product information extraction and recommendation module; through mutual cooperation of the image visual processing module, the image matching and screening module and the product information extraction and recommendation module, accurate screening and recommendation of products are realized, and whether a transportation management alarm signal is generated or not is judged through operation management analysis. And when the operation management alarm signal is generated, system operation supervision is enhanced, and reasonable improvement measures are taken in a targeted manner, so that the subsequent safe, stable, efficient and accurate operation of the recommendation system is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of product recommendation management, and in particular to an intelligent recommendation system based on visual presentation of product pictures. Background Art

[0002] In the field of e-commerce, product recommendation systems have become an important means to improve user experience and sales conversion rates. Traditional recommendation systems mainly rely on data such as user browsing history, purchase history, and text descriptions for recommendations; However, due to insufficient utilization of the visual features of product images, it is impossible to make product recommendations through visual image presentation, resulting in the need to improve recommendation accuracy and user experience. In addition, it is impossible to comprehensively evaluate the operating status of the recommendation system based on multi-directional analysis and issue timely warnings. It is difficult to strengthen system operation supervision in a timely manner and make targeted and reasonable improvement measures, which is not conducive to ensuring the subsequent safe, stable, efficient and accurate operation of the recommendation system.

[0003] In view of the above technical defects, a solution is now proposed. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent recommendation system based on the visual presentation of product pictures, which solves the problem that the prior art cannot make product recommendations through visual presentation of images, resulting in the need to improve the recommendation accuracy and user experience, and cannot comprehensively evaluate the operating status of the recommendation system based on multi-directional analysis and provide timely warnings, which is not conducive to ensuring the subsequent safe, stable, efficient and accurate operation of the recommendation system.

[0005] To achieve the above object, the present invention provides the following technical solutions: The intelligent recommendation system based on the visual presentation of product images includes an intelligent recommendation unit, an operation management and analysis unit, and a management and maintenance terminal, wherein the intelligent recommendation unit includes a product request image uploading module, an image visual processing module, an image matching and screening module, and a product information extraction and recommendation module; a user uploads a product request image through the product request image uploading module, and the uploaded product request image is sent to the image visual processing module; The image visual processing module improves the image quality through preprocessing steps including denoising, contrast enhancement and image size adjustment, and uses a deep learning algorithm to extract features from the preprocessed image to extract key features including color, texture and shape in the image, and sends the preprocessed product request image and corresponding extracted information to the image matching screening module; The image matching and screening module matches the product request image uploaded by the user with multiple candidate product images, and uses the pre-trained image matching neural network to calculate the matching degree between the product request image and each candidate product image. Based on the matching result, the candidate product image sequence that best matches the target product is screened out, and the candidate product sequence is sent to the product information extraction and recommendation module; The product information extraction and recommendation module further processes the screened candidate product image sequence, extracts the corresponding product description information, including product name, price and specifications, and pushes the extracted product description information to the corresponding user's smart terminal device; the operation management analysis unit determines whether to generate an operation management alarm signal through operation management analysis, and sends it to the management and maintenance end when the operation management alarm signal is generated.

[0006] Furthermore, the operation management analysis process of the operation management analysis unit is as follows: A detection period with a number of days set to Y1 is obtained when the number of days reaches Y1. The recommended characteristic value, congestion characteristic value and risk characteristic value are compared with the preset recommended characteristic threshold, the preset congestion characteristic threshold and the preset risk characteristic threshold respectively. If the recommended characteristic value exceeds the preset recommended characteristic threshold and the congestion characteristic value and the risk characteristic value do not exceed the preset congestion characteristic threshold and the preset risk characteristic threshold, then no operation management alarm signal is generated. In other cases, an operation management alarm signal is generated.

[0007] Further, the operation management analysis unit includes a recommendation performance analysis module, a request congestion analysis module, and a risk assessment analysis module; The recommendation performance analysis module analyzes the recommendation performance of the intelligent recommendation unit during the detection period, obtains the recommendation characteristic value through analysis, and sends the recommendation characteristic value to the operation management analysis unit; the request congestion analysis module analyzes the operation congestion status of the intelligent recommendation unit during the detection period, obtains the congestion characteristic value through analysis, and sends the congestion characteristic value to the operation management analysis unit; the risk assessment analysis module analyzes the system operation risk, generates the risk characteristic value through analysis, and sends the risk characteristic value to the operation management analysis unit.

[0008] Furthermore, the specific analysis process of the recommended performance analysis module is as follows: The total number of product recommendation requests made by the intelligent recommendation unit during the detection period is obtained and the ratio of the number of product recommendation requests to the set standard request number is marked as the recommendation request value, and the number of orders placed according to the recommended product information during the detection period is calculated by the ratio of the total number of product recommendation requests to obtain the recommendation effective value; The time when the user completes the image upload through the product request image upload module is collected and marked as the upload time, and the time when the corresponding user's smart terminal device receives the recommendation information is marked as the recommendation time, and the interval between the recommendation time and the upload time is marked as the recommendation effect time; All thrust efficiency durations within the detection period are obtained and their average is calculated to obtain the thrust efficiency detection value, and the proportion of thrust efficiency durations that do not exceed the preset thrust efficiency duration threshold within the detection period is marked as the thrust efficiency optimization value; the recommended feature value is obtained by numerically calculating the recommended request value, the recommended effective value, the thrust efficiency detection value and the thrust efficiency optimization value.

[0009] Furthermore, the specific analysis process of the request congestion analysis module is as follows: The real-time number of product recommendation requests being processed is collected and marked as a request processing check value, and the request processing check value is compared with a preset request processing check threshold. If the request processing check value exceeds the preset request processing check threshold, it is determined that the current state is in a request partial congestion state; when it is determined that the state is in a request partial congestion state, the timing starts and the time is counted until the request partial congestion state ends, thereby obtaining the request partial congestion duration; Compare the request bias congestion duration with the preset request bias congestion duration threshold; if the request bias congestion duration exceeds the preset request bias congestion duration threshold, mark the corresponding request bias congestion duration as the request abnormal congestion duration, and mark the number of request abnormal congestion durations in the detection period as the abnormal congestion detection value; The congestion duration of all requests during the detection period is obtained and summed up to obtain the total congestion time value, and the average of all request detection values ​​during the detection period is calculated to obtain the request table value. The congestion characteristic value is obtained by numerically calculating the total congestion time value, the congestion detection value and the request table value.

[0010] Furthermore, the specific analysis process of the risk assessment analysis module is as follows: The total number of network attacks on the system during the detection period is obtained and marked as the attack risk value, and the total number of system crashes during the detection period is collected and marked as the crash risk value, and the duration of each system crash is marked as the crash time calculation value. All crash time calculation values ​​of system crashes during the detection period are summed up to obtain the crash time detection value; the risk characteristic value is obtained by numerically calculating the attack risk value, the crash risk value and the crash time detection value.

[0011] Furthermore, the operation management and analysis unit also includes a sustainability assessment module. When no operation management alarm signal is generated, the sustainability assessment module analyzes the sustainability status of the system, generates a sustainability qualification signal or a sustainability abnormality signal through the analysis, and sends the sustainability qualification signal or the sustainability abnormality signal to the management and maintenance end via the operation management and analysis unit.

[0012] Furthermore, the specific analysis process of the sustainability assessment module includes: The number of registered users and the number of high-level users are obtained and marked as the user total inspection value and the high-level user inspection value respectively, and the system return amount and the operation investment amount are collected and the ratio of the two is marked as the profit and loss detection value; and the number of negative reviews for the system during the detection period is collected and marked as the negative review detection value; The sustainability assessment value is obtained by numerically calculating the user's total inspection value, advanced household inspection value, profit and loss detection value, and bad review detection value, and the sustainability assessment value is numerically compared with the preset sustainability assessment threshold. If the sustainability assessment value exceeds the preset sustainability assessment threshold, a sustainability qualification signal is generated; if the sustainability assessment value does not exceed the preset sustainability assessment threshold, a sustainability abnormality signal is generated.

[0013] Furthermore, the sustainability evaluation module is communicated with the user detection and grading module, and the user detection and grading module obtains all registered users, marks the interval between the registration time of the corresponding user and the current time as the registration time, and marks the interval between the time when the corresponding user last sent a product recommendation request and the current time as the recommendation interval value, and calculates the ratio of the total number of product recommendation requests sent by the corresponding user in the historical stage to its registration time to obtain the request performance value; The user detection analysis value is obtained by numerically calculating the registration duration, recommendation interval value and request performance value, and the user detection analysis value is numerically compared with the preset user detection analysis threshold. If the user detection analysis value exceeds the preset user detection analysis threshold, the corresponding user is marked as a high-level user.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. In the present invention, through advanced image processing and artificial intelligence algorithms, accurate screening and recommendation of products are achieved, and through operation management analysis, it is determined whether to generate an operation management alarm signal. When the operation management alarm signal is generated, the system operation supervision is strengthened and reasonable improvement measures are made in a targeted manner to ensure the subsequent safe, stable, efficient and accurate operation of the recommendation system; 2. In the present invention, the sustainability evaluation module analyzes the sustainability status of the system operation when no operation management alarm signal is generated, and generates a sustainability qualified signal or a sustainability abnormal signal through analysis. When a sustainability abnormal signal is generated, the management and maintenance end issues a corresponding warning, which is beneficial for the administrator to adjust the subsequent operation management plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings; Figure 1 This is a system block diagram of Embodiment 1 of the present invention; Figure 2 This is a system block diagram of Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Embodiment 1: Figure 1 As shown, the intelligent recommendation system based on product image visual presentation proposed by the present invention includes an intelligent recommendation unit, an operation management and analysis unit and a management and maintenance terminal, wherein the intelligent recommendation unit includes a product request image uploading module, an image visual processing module, an image matching and screening module and a product information extraction and recommendation module; the user uploads the product request image through the product request image uploading module, and sends the uploaded product request image to the image visual processing module; The image visual processing module improves the image quality through preprocessing steps, including denoising, contrast enhancement, and image resizing. It also uses deep learning algorithms to extract features from preprocessed images, such as convolutional neural networks (CNN) and other models, to extract key features such as color, texture, and shape from the image. It also sends the preprocessed product request image and the corresponding extracted information to the image matching and screening module. The image matching and screening module matches the product request image uploaded by the user with multiple candidate product images, and uses the pre-trained image matching neural network to calculate the matching degree between the product request image and each candidate product image. Based on the matching result, the candidate product image sequence that best matches the target product is screened out, and the candidate product sequence is sent to the product information extraction and recommendation module; The product information extraction and recommendation module further processes the screened candidate product image sequence to extract the corresponding product description information, including product name, price and specifications, and pushes the extracted product description information to the corresponding user's smart terminal device to achieve accurate product recommendation; through advanced image processing and artificial intelligence algorithms, accurate screening and recommendation of products are achieved.

[0018] The operation management and analysis unit determines whether to generate an operation management alarm signal through operation management analysis. When an operation management alarm signal is generated, it is sent to the management and maintenance end. When the management and maintenance end receives the operation management alarm signal, it issues a corresponding warning to remind the administrator to strengthen the system operation supervision and make targeted and reasonable improvement measures to ensure the subsequent safe, stable, efficient and accurate operation of the recommended system. The operation management and analysis process of the operation management and analysis unit is as follows: The number of days is set to be a detection period of Y1, preferably, Y1 is thirty days; when the number of days reaches Y1, the recommended characteristic value TX, the congestion characteristic value QP and the risk characteristic value SX are obtained, and the recommended characteristic value TX, the congestion characteristic value and the risk characteristic value SX are numerically compared with the preset recommended characteristic threshold, the preset congestion characteristic threshold and the preset risk characteristic threshold respectively; If the recommended characteristic value TX exceeds the preset recommended characteristic threshold and the congestion characteristic value QP and the risk characteristic value SX do not exceed the preset congestion characteristic threshold and the preset risk characteristic threshold, it indicates that the system operation is generally good, and no operation management alarm signal is generated. Other situations indicate that the system operation is generally poor, and subsequent supervision needs to be strengthened and corresponding improvement measures need to be taken, then an operation management alarm signal is generated.

[0019] It should be noted that the operation management analysis unit includes a recommendation performance analysis module, a request congestion analysis module and a risk assessment analysis module; the recommendation performance analysis module analyzes the recommendation performance of the intelligent recommendation unit during the detection period, obtains the recommendation feature value TX through analysis, and sends the recommendation feature value TX to the operation management analysis unit, which can not only reasonably evaluate the product recommendation performance during the detection period, but also provide data support for the operation management analysis process to ensure the comprehensiveness of the analysis and the accuracy of the analysis results; the specific analysis process of the recommendation performance analysis module is as follows: The total number of product recommendation requests made by the intelligent recommendation unit during the detection period is obtained and the ratio of the number of product recommendation requests to the set standard number of requests is marked as the recommendation request value, and the number of times the user places an order according to the recommended product information during the detection period is calculated by the ratio of the total number of product recommendation requests to obtain the recommendation effective value; The time when the user completes the image upload through the product request image upload module is collected and marked as the upload time, and the time when the corresponding user's smart terminal device receives the recommendation information is marked as the recommendation time, and the interval between the recommendation time and the upload time is marked as the recommendation effect time; All the thrust efficiency durations in the detection period are obtained and their average is calculated to obtain the thrust efficiency detection value, and the thrust efficiency duration is numerically compared with the preset thrust efficiency duration threshold, and the number of thrust efficiency durations that do not exceed the preset thrust efficiency duration threshold in the detection period is marked as the thrust efficiency excellent detection value; The recommended request value RX, recommended effective value NX, push efficiency detection value FM and push efficiency optimization detection value WY are numerically calculated using the formula TX=eq×RX+tu×NX+up / FM+ng×WY to obtain the recommended characteristic value TX; wherein eq, tu, up, ng are preset proportional coefficients whose values ​​are greater than zero, and the larger the numerical value of the recommended characteristic value TX, the better the overall performance of the product recommendation during the detection period.

[0020] The request congestion analysis module analyzes the operation congestion status of the intelligent recommendation unit during the detection period, obtains the congestion characteristic value QP through analysis, and sends the congestion characteristic value QP to the operation management analysis unit; the specific analysis process of the request congestion analysis module is as follows: The real-time number of product recommendation requests being processed is collected and marked as a request processing check value, and the request processing check value is compared with a preset request processing check threshold. If the request processing check value exceeds the preset request processing check threshold, it indicates that the request processing status at the corresponding moment is relatively busy, and it is determined that the current state is in a request partial congestion state; when it is determined that the state is in a request partial congestion state, the timing starts, and the time is counted until the request partial congestion state ends, thereby obtaining the request partial congestion duration; Compare the request bias congestion duration with the preset request bias congestion duration threshold; if the request bias congestion duration exceeds the preset request bias congestion duration threshold, mark the corresponding request bias congestion duration as the request abnormal congestion duration, and mark the number of request abnormal congestion durations in the detection period as the abnormal congestion detection value; The congestion duration of all requests during the detection period is obtained and summed up to obtain the total congestion time value, and the average of all request detection values ​​during the detection period is calculated to obtain the request table value. The total congestion time value HS, the abnormal congestion detection value GY and the request table value XW are numerically calculated using the formula QP=su×HS+tp×GY+eu×XW to obtain the congestion characteristic value QP; wherein su, tp, eu are preset weight coefficients with values ​​greater than zero, and the larger the value of the congestion characteristic value QP, the greater the system operation pressure during the detection period, and the more it is necessary to strengthen system optimization supervision and improve its processing capacity in the subsequent period.

[0021] The risk assessment and analysis module analyzes the system operation risk, generates a risk characteristic value SX through analysis, and sends the risk characteristic value SX to the operation management analysis unit; the specific analysis process of the risk assessment and analysis module is as follows: The total number of network attacks on the system during the detection period is obtained and marked as the attack risk value, and the total number of system crashes during the detection period is collected and marked as the crash risk value, and the duration of each system crash is marked as the crash time calculation value, and all crash time calculation values ​​of the system crashes during the detection period are summed up to obtain the crash time detection value; The attack risk value LR, the collapse risk value WP and the collapse detection value SF are numerically calculated using the formula SX=tg×LR+ns×WP+fw×SF to obtain the risk characteristic value SX; wherein tg, ns, and fw are preset weight coefficients whose values ​​are greater than zero, and the larger the value of the risk characteristic value SX, the greater the overall system operation risk during the detection period, and the less conducive it is to ensuring the safe and stable operation of the system.

[0022] Embodiment 2: Figure 2 As shown, the difference between this embodiment and the first embodiment is that the operation management analysis unit further includes a sustainability evaluation module. When the operation management alarm signal is not generated, the sustainability evaluation module analyzes the sustainability status of the system operation, generates a sustainability qualified signal or a sustainability abnormal signal through analysis, and sends the sustainability qualified signal or the sustainability abnormal signal to the management and maintenance end through the operation management analysis unit. When the management and maintenance end receives the sustainability abnormal signal, it issues a corresponding warning, which is beneficial for the administrator to adjust the subsequent operation management plan to ensure the operation efficiency. The specific analysis process of the sustainability evaluation module is as follows: The number of registered users and the number of high-level users are obtained and marked as the user total inspection value and the high-level user inspection value respectively, and the system return amount and the operation investment amount are collected and the ratio of the two is marked as the profit and loss detection value; and all evaluation information of users on the system during the detection period is collected, and the number of negative reviews is obtained and marked as the negative review detection value; The user total inspection value TN, the advanced household inspection value SK, the profit and loss detection value LP and the bad review detection value MF are numerically calculated by the formula QM=(mu×TN+re×SK+wt×LP) / (ru×MF+1) to obtain the sustainability evaluation value QM; and mu, re, wt, ru are preset proportional coefficients with values ​​greater than zero, and the smaller the value of the sustainability evaluation value QM, the less sustainable the system operation is, and the worse its subsequent development prospects are overall; The sustainability assessment value QM is numerically compared with the preset sustainability assessment threshold. If the sustainability assessment value QM exceeds the preset sustainability assessment threshold, it indicates that the subsequent development prospects of the system are generally good, and a sustainability qualified signal is generated; if the sustainability assessment value QM does not exceed the preset sustainability assessment threshold, it indicates that the subsequent development prospects of the system are generally poor, and a sustainability abnormal signal is generated.

[0023] Furthermore, the sustainability evaluation module is communicated with the user detection and grading module, and the user detection and grading module obtains all registered users, marks the interval between the registration time of the corresponding user and the current time as the registration time, and marks the interval between the time when the corresponding user last sent a product recommendation request and the current time as the recommendation interval value, and calculates the request performance value by calculating the ratio of the total number of product recommendation requests sent by the corresponding user in the historical stage to its registration time; The user detection analysis value YX is obtained by numerically calculating the registration duration ZF, the recommended interval value HP and the request performance value QY through the formula YX=(es×ZF+sw×QY) / (bt×HP+1); wherein es, sw and bt are preset proportional coefficients with values ​​greater than zero, and the larger the value of the user detection analysis value YX is, the stronger the corresponding user's dependence on the system is; The user detection analysis value YX is numerically compared with the preset user detection analysis threshold. If the user detection analysis value YX exceeds the preset user detection analysis threshold, it indicates that the corresponding user has a strong dependence on the system, and the corresponding user is marked as a high-level user; and the marking information of all users is sent to the sustainability evaluation module to provide data support for the analysis process of the sustainability evaluation module, thereby ensuring the accuracy of its analysis results.

[0024] The working principle of the present invention is as follows: when in use, a product request image is uploaded through a product request image upload module, an image vision processing module pre-processes and extracts features of the product request image, an image matching and screening module calculates the degree of match between the product request image and each candidate product image, and screens out a sequence of candidate product images that best matches the target product according to the matching result, a product information extraction and recommendation module further processes the screened sequence of candidate product images, extracts the description information of the corresponding product and pushes it to the smart terminal device of the corresponding user, thereby achieving accurate screening and recommendation of the product, and performing operation management analysis through an operation management analysis unit to determine whether to generate an operation management alarm signal, and when an operation management alarm signal is generated, strengthening system operation supervision and making targeted and reasonable improvement measures, thereby ensuring the subsequent safe, stable, efficient and accurate operation of the recommendation system.

[0025] The above formulas are all dimensionless and numerical calculations. The formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions. The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the contents of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that technicians in the relevant technical field can understand and use the present invention well. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. An intelligent recommendation system based on the visual presentation of product images, characterized by: It includes an intelligent recommendation unit, an operation management and analysis unit, and a management and maintenance terminal, wherein the intelligent recommendation unit includes a product request image upload module, an image visual processing module, an image matching and screening module, and a product information extraction and recommendation module; the user uploads the product request image through the product request image upload module, and sends the uploaded product request image to the image visual processing module; The image visual processing module improves the image quality through preprocessing steps, and uses a deep learning algorithm to extract features from the preprocessed image to extract the key features of the image, and sends the preprocessed product request image and corresponding extraction information to the image matching and screening module; The image matching and screening module matches the product request image uploaded by the user with multiple candidate product images, and uses the pre-trained image matching neural network to calculate the matching degree between the product request image and each candidate product image. Based on the matching result, the candidate product image sequence that best matches the target product is screened out, and the candidate product sequence is sent to the product information extraction and recommendation module; The product information extraction and recommendation module further processes the screened candidate product image sequence, extracts the corresponding product description information, and pushes the extracted product description information to the corresponding user's smart terminal device; the operation management analysis unit determines whether to generate an operation management alarm signal through operation management analysis, and sends it to the management and maintenance end when the operation management alarm signal is generated.

2. The intelligent recommendation system based on product image visual presentation according to claim 1, characterized in that: The operation management analysis process of the operation management analysis unit is as follows: The detection period is set to Y1 days. When the number of days reaches Y1, the recommended characteristic value, congestion characteristic value and risk characteristic value are obtained. If the recommended characteristic value exceeds the preset recommended characteristic threshold and the congestion characteristic value and the risk characteristic value do not exceed the preset congestion characteristic threshold and the preset risk characteristic threshold, then no operation management alarm signal is generated. In other cases, an operation management alarm signal is generated.

3. The intelligent recommendation system based on product image visual presentation according to claim 2, characterized in that: The operation management analysis unit includes a recommendation performance analysis module, a request congestion analysis module, and a risk assessment analysis module; The recommendation performance analysis module analyzes the recommendation performance of the intelligent recommendation unit during the detection period, obtains the recommendation feature value through analysis, and sends the recommendation feature value to the operation management analysis unit; Request the congestion analysis module to analyze the operation congestion status of the intelligent recommendation unit during the detection period, obtain the congestion characteristic value through the analysis, and send the congestion characteristic value to the operation management analysis unit; The risk assessment and analysis module analyzes the system operation risk, generates risk characteristic values ​​through analysis, and sends the risk characteristic values ​​to the operation management and analysis unit.

4. The intelligent recommendation system based on product image visual presentation according to claim 3 is characterized in that: The specific analysis process of the recommended performance analysis module is as follows: The total number of product recommendation requests made by the intelligent recommendation unit during the detection period is obtained and the ratio of the number of product recommendation requests to the set standard request number is marked as the recommendation request value, and the number of orders placed according to the recommended product information during the detection period is calculated by the ratio of the total number of product recommendation requests to obtain the recommendation effective value; The time when the user completes the image upload through the product request image upload module is collected and marked as the upload time, and the time when the corresponding user's smart terminal device receives the recommendation information is marked as the recommendation time, and the interval between the recommendation time and the upload time is marked as the recommendation effect time; All thrust efficiency durations within the detection period are obtained and their average is calculated to obtain the thrust efficiency detection value, and the proportion of thrust efficiency durations that do not exceed the preset thrust efficiency duration threshold within the detection period is marked as the thrust efficiency optimization value; the recommended feature value is obtained by numerically calculating the recommended request value, the recommended effective value, the thrust efficiency detection value and the thrust efficiency optimization value.

5. The intelligent recommendation system based on product image visual presentation according to claim 3 is characterized in that: The specific analysis process of the request congestion analysis module is as follows: The real-time number of product recommendation requests being processed is collected and marked as the request processing check value. If the request processing check value exceeds the preset request processing check threshold, it is determined that the current state is in a request congestion state and the request congestion duration is obtained; If the requested biased jam duration exceeds the preset requested biased jam duration threshold, the corresponding requested biased jam duration is marked as the requested abnormal jam duration, and the number of requested abnormal jam durations during the detection period is marked as the abnormal jam detection value; The congestion duration of all requests during the detection period is obtained and summed up to obtain the total congestion time value, and the average of all request detection values ​​during the detection period is calculated to obtain the request table value. The congestion characteristic value is obtained by numerically calculating the total congestion time value, the congestion detection value and the request table value.

6. The intelligent recommendation system based on product image visual presentation according to claim 3, characterized in that: The specific analysis process of the risk assessment analysis module is as follows: The total number of network attacks on the system during the detection period is obtained and marked as the attack risk value, and the total number of system crashes during the detection period is collected and marked as the crash risk value, and the duration of each system crash is marked as the crash time calculation value. All crash time calculation values ​​of system crashes during the detection period are summed up to obtain the crash time detection value; the risk characteristic value is obtained by numerically calculating the attack risk value, the crash risk value and the crash time detection value.

7. The intelligent recommendation system based on product image visual presentation according to claim 1, characterized in that: The operation management and analysis unit also includes a sustainability assessment module. When no operation management alarm signal is generated, the sustainability assessment module analyzes the system's application sustainability status, generates a sustainability qualification signal or a sustainability abnormality signal through the analysis, and sends the sustainability qualification signal or the sustainability abnormality signal to the management and maintenance end through the operation management and analysis unit.

8. The intelligent recommendation system based on product image visual presentation according to claim 7, characterized in that: The specific analysis process of the sustainability assessment module includes: The sustainability assessment value is obtained by numerically calculating the user's total inspection value, advanced household inspection value, profit and loss detection value, and bad review detection value. If the sustainability assessment value exceeds the preset sustainability assessment threshold, a sustainability qualification signal is generated; if the sustainability assessment value does not exceed the preset sustainability assessment threshold, a sustainability abnormality signal is generated.

9. The intelligent recommendation system based on product image visual presentation according to claim 8, characterized in that: The sustainability assessment module is communicated with the user detection and grading module. The user detection and grading module obtains all registered users, and numerically calculates the registration duration, recommended interval value and request performance value of the corresponding user to obtain a user detection analysis value. If the user detection analysis value exceeds the preset user detection analysis threshold, the corresponding user is marked as a high-level user.

Images