Artificial intelligence-based pillow inner recommendation system combining head and neck typing and sleeping habits

Abstract

The invention discloses a pillow interior recommendation system combining head and neck typing and sleeping habits based on artificial intelligence, and particularly relates to the technical field of pillow interior recommendation, the error degree of head typing is recognized by obtaining error information of head typing and calculating a head typing error coefficient; acquiring classification deviation information of the sleep habits, and calculating classification deviation coefficients of the sleep habits to identify classification deviation degrees of the sleep habits; a matching pre-evaluation model is constructed according to the skull typing and the sleep habit, and a matching pre-evaluation index is output, so that potential abnormity can be perceived in advance before the skull typing and the sleep habit are matched; the matching abnormity degree is divided; the hidden danger perception module perceives and warns potential recommendation hidden dangers in real time by constructing a matching pre-evaluation index time sequence and calculating an index abnormal coefficient of the matching pre-evaluation index time sequence, it is ensured that pillow inner recommendation is more accurate and stable, the matching precision of personalized pillow inner recommendation is effectively improved, and the user experience is improved. Recommendation risks caused by data errors and model deviation are reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of pillow core recommendation, and more specifically, to a pillow core recommendation system based on the combination of head and neck typing and sleep habits using artificial intelligence. Background Art

[0002] With the development of artificial intelligence and intelligent health monitoring technologies, personalized sleep product recommendation has become an important means to improve users' sleep quality. A pillow core recommendation system based on the combination of AI-based head and neck typing and sleep habits aims to intelligently match the most suitable pillow core by accurately measuring the head and neck structure characteristics of users and combining long-term sleep habit data. However, this system still faces technical challenges in multiple aspects such as data collection, model construction, and personalized recommendation. Especially in the problem of matching errors caused by head typing errors and sleep habit classification biases, it may lead to inaccurate personalized matching results, thus affecting the recommendation effect. Therefore, how to intelligently and accurately perceive the head typing errors and sleep habit classification biases of the system has become an urgent technical problem to be solved. Summary of the Invention

[0003] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a pillow core recommendation system based on the combination of head and neck typing and sleep habits using artificial intelligence to solve the problems raised in the above background art.

[0004] To achieve the above object, the present invention provides the following technical solutions: A pillow core recommendation system based on the combination of head and neck typing and sleep habits using artificial intelligence includes a head typing monitoring module, a sleep habit monitoring module, a pre-matching evaluation module, a pre-matching division module, and a hidden danger perception module; The head typing monitoring module is used to obtain the error information of head typing, calculate the head typing error coefficient according to the error information of head typing, and identify the degree of head typing error; The sleep habit monitoring module is used to obtain the classification deviation information of sleep habits, calculate the sleep habit classification deviation coefficient according to the classification deviation information of sleep habits, and identify the degree of sleep habit classification deviation; The pre-matching evaluation module is used to construct a pre-matching evaluation model based on the head typing error coefficient and the sleep habit classification deviation coefficient, output a pre-matching evaluation index, and perceive potential matching abnormalities before the head typing and sleep habit matching; The pre-matching division module is used to compare the pre-matching evaluation index with a preset pre-matching evaluation index threshold to divide the potential matching abnormality degree of the head typing and sleep habits; The hidden danger perception module is used to intelligently perceive the recommendation hidden dangers of the pillow core recommendation system according to the potential matching abnormality degree of the head typing and sleep habits.

[0005] In a preferred embodiment, the acquisition logic of the skull classification error coefficient is as follows: During the process of skull classification, extract the head feature point data of each user, and calculate the abnormal missing ratio of feature extraction of the user: , where represents the abnormal missing ratio of feature extraction of the user, represents the number of missing feature points of the user's head, represents the total number of feature points that should be extracted; according to the abnormal missing ratio of feature extraction of each user, construct an abnormal missing data set of feature extraction , where represents the th abnormal missing ratio of feature extraction of the user, , is the total number of users participating in skull classification; Calculate the distances between the abnormal missing ratios of feature extraction of all users, and construct a distance matrix : , where is the distance between the abnormal missing ratios of feature extraction of users, representing the th user and the th user's absolute difference in the abnormal missing ratio of feature extraction; Construct an abnormal topological structure of feature extraction according to the distance matrix, specifically as follows: A1. Compare the distance between the abnormal missing ratios of feature extraction of users with a preset distance threshold. If the distance between the abnormal missing ratios of feature extraction of users is less than or equal to the distance threshold, establish an edge in the topological complex; A2. For any three users, if the distances between pairwise users are all less than or equal to the distance threshold, form a triangular face; Traverse the distance matrix and repeat A1 and A2 to construct an abnormal topological structure of feature extraction; During the construction process of the abnormal topological structure of feature extraction, record the birth time and extinction time of each topological feature, and calculate the persistence: , where represents the persistence of the th topological feature, represents the birth time of the th topological feature, represents the extinction time of the th topological feature; Calculate the skull classification error coefficient: , where represents the skull classification error coefficient, , represents the total number of topological features.

[0006] In a preferred embodiment, the acquisition logic of the sleep habit classification deviation coefficient is as follows: For each user, obtain the probability distribution of the sleep habit category output by the classification model, and calculate the sleep habit classification information entropy of the th user: where represents the sleep habit classification information entropy of the th user, represents the probability of the th sleep habit category in the classification result of the th user, is the total number of sleep habit categories; According to the sleep habit classification information entropy of each user, construct the sleep habit classification information entropy set where represents the sleep habit classification information entropy of the th user, and is the total number of users participating in sleep habit classification; Calculate the deviation of the sleep habit classification information entropy between the th user and the th user: where represents the deviation of the sleep habit classification information entropy between the th user and the th user, represents the sleep habit classification information entropy of the th user; Calculate the sleep habit classification deviation coefficient: where represents the sleep habit classification deviation coefficient, .

[0007] In a preferred embodiment, construct a matching pre-evaluation model based on the skull typing error coefficient and the sleep habit classification deviation coefficient, and output the matching pre-evaluation index , and the formula on which the model is based is as follows , where in the formula represents the skull typing error coefficient, represents the sleep habit classification deviation coefficient, respectively represent the preset proportionality coefficients of the skull typing error coefficient and the sleep habit classification deviation coefficient, and are both greater than 0.

[0008] In a preferred embodiment, the pre-matching evaluation index is compared with a preset pre-matching evaluation index threshold to classify the potential matching abnormality degree between the skull type and the sleep habit, specifically as follows: If the pre-matching evaluation index is greater than the pre-matching evaluation index threshold, a high-abnormality matching signal is generated; if the pre-matching evaluation index is less than or equal to the pre-matching evaluation index threshold, a low-abnormality matching signal is generated.

[0009] In a preferred embodiment, when a high-abnormality matching signal is first generated, the pre-matching evaluation index output by the pre-matching evaluation model subsequently is acquired to construct a pre-matching evaluation index time series: , where represents the pre-matching evaluation index output by the pre-matching evaluation model at the t-th time point, t = {1, 2,..., T}, and T is a positive integer; Calculate the index abnormality coefficient: , where represents the index abnormality coefficient, represents the pre-matching evaluation index output by the pre-matching evaluation model at the T-th time point, represents the mean value of the pre-matching evaluation index time series: , represents the standard deviation of the pre-matching evaluation index time series: .

[0010] In a preferred embodiment, the index abnormality coefficient is compared with a preset index abnormality coefficient threshold to intelligently perceive the recommendation hidden danger of the pillow core recommendation system, specifically as follows: If the index abnormality coefficient is greater than the index abnormality coefficient threshold, a recommendation hidden danger abnormality signal is generated; if the index abnormality coefficient is less than or equal to the index abnormality coefficient threshold, there is no need to generate a recommendation hidden danger abnormality signal.

[0011] The technical effects and advantages of the present invention: 1. The present invention obtains the error information of head typing in real time through the head typing monitoring module, calculates the head typing error coefficient based on the error information of head typing, and identifies the error degree of head typing; at the same time, obtains the classification deviation information of sleep habits, and calculates the sleep habit classification deviation coefficient according to the classification deviation information of sleep habits to identify the classification deviation degree of sleep habits; next, the matching pre-evaluation module constructs a matching pre-evaluation model based on both, outputs the matching pre-evaluation index, and realizes the early perception of potential abnormalities before the matching of head typing and sleep habits; the matching pre-division module compares the evaluation index with a preset threshold to finely divide the matching abnormal degree; the hidden danger perception module constructs a time series of the matching pre-evaluation index and calculates the index abnormal coefficient of the time series of the matching pre-evaluation index to perceive and warn of potential recommendation hidden dangers in real time, ensuring that the pillow core recommendation is more accurate and stable, effectively improving the matching accuracy of personalized pillow core recommendation, reducing the recommendation risk caused by data errors and model deviations, and moreover, through an intelligent and dynamically adaptive hidden danger perception and warning mechanism, it can provide timely warnings for optimizing the matching strategy, improve the user's sleep experience and health protection, and realize the deep integration of intelligent health monitoring and personalized product recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings; Figure 1 It is a flowchart of the system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0014] Embodiment: The present invention provides a pillow core recommendation system based on the combination of head and neck typing and sleep habits as shown in Figure 1 including a head typing monitoring module, a sleep habit monitoring module, a matching pre-evaluation module, a matching pre-division module, and a hidden danger perception module; The head typing monitoring module is used to obtain the error information of head typing, calculate the head typing error coefficient based on the error information of head typing, and identify the error degree of head typing; The sleep habit monitoring module is used to obtain the classification deviation information of sleep habits, calculate the sleep habit classification deviation coefficient according to the classification deviation information of sleep habits, and identify the classification deviation degree of sleep habits; A matching pre-evaluation module, which is used to construct a matching pre-evaluation model based on the skull typing error coefficient and the sleep habit classification deviation coefficient, output a matching pre-evaluation index, and perceive potential matching anomalies before the skull typing and sleep habit matching; A matching pre-division module, which is used to compare the matching pre-evaluation index with a preset matching pre-evaluation index threshold to divide the potential matching anomaly degree between the skull typing and the sleep habit; A hidden danger perception module, which is used to intelligently perceive the recommended hidden dangers of the pillow core recommendation system according to the potential matching anomaly degree between the skull typing and the sleep habit; In the present invention, the skull typing error coefficient is used to measure the abnormal error degree existing in the process of skull typing by the pillow core recommendation system. A larger skull typing error coefficient indicates that there are greater abnormal hidden dangers in the process of skull typing, which may lead to a decrease in matching accuracy and affect the accuracy of pillow core recommendation; a smaller skull typing error coefficient indicates that the skull typing result is more accurate, which helps to improve the reliability of the recommendation effect. Based on the skull typing error coefficient, perceiving potential matching anomalies before the skull typing and sleep habit matching helps to identify matching deviations in advance, optimize the matching strategy, improve the accuracy of personalized recommendation, and enhance the sleep comfort and experience of users.

[0015] The acquisition logic of the skull typing error coefficient is as follows: During the process of skull typing, extract the head feature point data of each user and calculate the abnormal missing ratio of feature extraction of the user: , where represents the abnormal missing ratio of feature extraction of the user, represents the number of missing feature points of the user's head, represents the total number of feature points that should be extracted; according to the abnormal missing ratio of feature extraction of each user, construct an abnormal missing data set of feature extraction , where represents the th user's abnormal missing ratio of feature extraction, , is the total number of users participating in skull typing; Calculate the distance between the abnormal missing ratios of feature extraction of all users and construct a distance matrix : , where is the distance between the abnormal missing ratios of feature extraction of users, representing the th user and the th user's absolute difference in the abnormal missing ratio of feature extraction; Construct a feature extraction abnormal topological structure according to the distance matrix, specifically as follows: A1. Compare the distance between the abnormal missing ratios of user feature extraction with a preset distance threshold. If the distance between the abnormal missing ratios of user feature extraction is less than or equal to the distance threshold, establish an edge in the topological complex. A2. For any three users, if the distances between every two users are all less than or equal to the distance threshold, form a triangular face. Traverse the distance matrix and repeat A1 and A2 to construct the topological structure of feature extraction anomalies. During the construction of the topological structure of feature extraction anomalies, record the birth time and death time of each topological feature and calculate the persistence: , where represents the persistence of the th topological feature, represents the birth time of the th topological feature, represents the death time of the th topological feature; The topological features mentioned above include but are not limited to connected components, loops, holes, etc.; A connected component is a group of users connected by a specific distance threshold in the topological structure. In graph theory, each connected component is a maximum connected subgraph; a connected component can reflect the similarity of users in terms of the abnormal missing ratio of feature extraction. More connected components usually indicate the inconsistency and dispersion of feature data, which may lead to a large error in skull typing. A loop is a cycle or closed path in the topological structure, usually indicating a periodic or repeating pattern existing in the data. The existence of a loop means that there may be repeated missing or abnormal patterns in the feature data; A hole is a "void" in the topological structure, that is, an area without data point connections, indicating obvious missing or faults in the data. The existence of a hole indicates a blank area or information missing in the data, which may lead to the omission of important features during the skull typing process and further affect the accuracy of the system; It should be noted that the birth time represents the scale at which the topological feature first appears, and the death time represents the time when the topological feature disappears after increasing the scale; Calculate the skull typing error coefficient: , where represents the skull typing error coefficient, , represents the total number of topological features; In the process of skull classification, it is first necessary to obtain the head feature point extraction data of each user and analyze the integrity of this data. During the data collection process, due to factors such as equipment errors, occlusion, sampling angles, etc., some feature points may not be successfully extracted. To measure these abnormal missing situations, it is necessary to calculate the abnormal missing ratio of feature extraction for each user, which reflects the degree of data missing in the feature extraction process. Based on the abnormal missing ratios of feature extraction for all users, a data set containing the abnormal missing situations of feature extraction for all users is constructed for further analysis of the stability and consistency of feature extraction. Subsequently, by calculating the distances between the abnormal missing ratios of feature extraction for all users, a distance matrix is formed. Each element in this matrix represents the absolute difference in the abnormal missing ratio of feature extraction between two users. This matrix is used to describe the relative differences in the quality of feature extraction between users and provides data support for subsequent topological analysis. After obtaining the distance matrix, it is necessary to construct an abnormal topological structure of feature extraction to analyze the topological features of feature data at different scales. Specifically, first, the distance between the abnormal missing ratios of feature extraction for users is compared with a preset distance threshold. If the distance between two users is less than or equal to this threshold, a connecting edge is established in the topological complex, thus forming a basic topological relationship. Further, if the pairwise distances between any three users are all less than or equal to this threshold, a triangular face is formed in the topological complex to reflect the strong correlation among the three. By traversing the entire distance matrix and continuously adding connecting edges and triangular faces in the above manner, a complete abnormal topological structure of feature extraction is finally formed, thereby reflecting the overall pattern of abnormal missing of feature extraction among different users. During the construction of the abnormal topological structure of feature extraction, it is necessary to record the birth time and death time of each topological feature to characterize the persistence of the topological feature. Among them, the birth time represents the scale at which this topological feature first appears, and the death time represents the time when this topological feature disappears after the scale increases to a certain critical value. Based on this information, the persistence of each topological feature can be calculated, that is, the stability of this topological feature in the data structure. The higher the persistence of a topological feature, the more likely it is to represent the long-term existence of the abnormal missing pattern of feature extraction, and thus the more likely it is to affect the accuracy of skull classification. Finally, by comprehensively calculating the persistence of all topological features, the skull classification error coefficient is obtained. This error coefficient is used to measure the overall stability and abnormality degree of feature extraction during skull classification. A larger error coefficient means that there are more instabilities or abnormal missing in the feature extraction process, which may lead to inaccurate skull classification results; while a smaller error coefficient indicates that the feature extraction is relatively stable and the reliability of skull classification is higher.

[0016] The sleep habit monitoring module is used to obtain the classification deviation information of sleep habits, calculate the sleep habit classification deviation coefficient according to the classification deviation information of sleep habits, and identify the classification deviation degree of sleep habits; The sleep habit classification deviation coefficient in the present invention is used to measure the degree of deviation of the pillow core recommendation system in classifying the user's sleep habits. The size of this coefficient directly affects the system's accurate characterization of the user's sleep characteristics and the reliability of the recommendation. A larger sleep habit classification deviation coefficient indicates that the system has a higher uncertainty when classifying the user's sleep habits, resulting in poor stability of the classification results, thereby reducing the accuracy of the matching, and even causing the incompatibility of the pillow core recommendation. On the contrary, a smaller sleep habit classification deviation coefficient indicates that the system has a higher consistency when classifying sleep habits, the classification model can better identify the user's long-term sleep habits, and the credibility of the data is strong, thereby improving the accuracy of pillow core recommendations and user comfort. Based on this coefficient, potential matching anomalies are perceived before the head typing is matched with the sleep habits, so that the system can adjust or optimize the error before matching, improve the effectiveness of personalized recommendations, and ensure that the final recommended pillow core is more in line with the user's real needs and improve the user's sleep quality and experience.

[0017] The logic for obtaining the sleep habit classification deviation coefficient is as follows: For each user, the probability distribution of the sleep habit category output by the classification model is calculated. The information entropy of the sleep habits of each user is: ,in Indicates The classified information entropy of the sleeping habits of each user, Indicates The sleep habit categories are The probability of a user's classification result, , is the total number of sleep habit categories; According to the sleep habit classification information entropy of each user, a sleep habit classification information entropy set is constructed ,in Indicates The classified information entropy of the sleeping habits of each user, , Total number of users categorized for participating sleep habits; Calculate the User and Deviation of the information entropy of the sleep habits classification of each user: ,in Indicates User and The deviation of the information entropy of the sleep habit classification of each user, Indicates The information entropy of the sleep habits of each user; Calculate the coefficient of bias for sleep habit classification: ,in represents the classification deviation coefficient of sleep habits, ; It should be noted that the above formulas are all dimensionless and only take their numerical values for calculation. Common methods for removing dimensions include Min-Max normalization, Z-Score standardization, etc., which will not be elaborated here; First, for each user, obtain the probability distribution of the sleep habit categories output by the classification model. These probability distributions reflect the probability values of each user in various sleep habit categories. Then, use these probability values to calculate the classification information entropy of each user's sleep habits. Information entropy is used to measure the uncertainty of classification. For users with higher information entropy, it means that the system is more uncertain about the classification of this user, and the classification deviation may be larger. Once the classification information entropy of each user's sleep habits is calculated, an information entropy set can be constructed, which contains the classification information entropy of all users. This step provides a data basis for subsequent classification deviation calculation. Next, calculate the deviation between the classification information entropy of each user and that of other users. This calculation helps to quantify the degree of classification deviation between different users, that is, the difference between the classification information entropy of a certain user and that of other users. If the difference between the two is large, it means that the classification deviation between these two users is large, which may indicate that the classification model performs inconsistently on these users. Finally, based on the calculated classification information entropy deviation between users, an overall classification deviation coefficient of sleep habits can be obtained. This coefficient reflects the overall level of classification deviation in the entire user group. A higher deviation coefficient indicates that there are significant differences in the classification results of the model among different users, while a lower deviation coefficient indicates that the classification results of the model are more consistent, with higher stability and reliability. This process, from obtaining the probability distribution of each user's sleep habit classification, to calculating the classification information entropy of each user, to measuring the classification deviation, and finally obtaining the classification deviation coefficient of sleep habits, provides a quantitative indicator for analyzing and optimizing the sleep habit classification model.

[0018] The matching pre-evaluation module is used to construct a matching pre-evaluation model based on the skull typing error coefficient and the classification deviation coefficient of sleep habits, and output a matching pre-evaluation index to perceive potential matching abnormalities before the skull typing and sleep habit matching; Construct a matching pre-evaluation model based on the skull typing error coefficient and the classification deviation coefficient of sleep habits, and output a matching pre-evaluation index , and the formula on which the model is based is as follows , where represents the skull typing error coefficient, represents the classification deviation coefficient of sleep habits, respectively represent the preset proportional coefficients of the skull typing error coefficient and the classification deviation coefficient of sleep habits, and are both greater than 0; It should be noted that the above formulas are all dimensionless and only take their numerical values for calculation. Common methods for removing dimensions include Min-Max normalization, Z-Score standardization, etc., which will not be elaborated here; It is set according to the actual situation. For example, the expert weighting method is adopted, that is, experts in related fields are invited to determine the preset proportional coefficients of various indicators through professional opinion surveys and comprehensive evaluations. For example, it can be 0.5, 0.5; It can be seen from the above calculation expressions that the larger the skull typing error coefficient and the larger the sleep habit classification deviation coefficient, the larger the matching pre-evaluation index, indicating that the skull typing error coefficient and the sleep habit classification deviation coefficient are relatively large, and there is a relatively high risk of potential matching anomalies. At this time, the matching process may face greater inconsistencies or mismatches. On the contrary, the smaller the skull typing error coefficient and the smaller the sleep habit classification deviation coefficient, the smaller the matching pre-evaluation index, indicating that the deviation between the two is smaller, the matching process is more stable, and the risk is lower; The matching pre-division module is used to compare the matching pre-evaluation index with the preset matching pre-evaluation index threshold to divide the potential matching anomaly degree of skull typing and sleep habits, specifically as follows: If the matching pre-evaluation index is greater than the matching pre-evaluation index threshold, it indicates that the potential matching anomaly degree of skull typing and sleep habits is relatively high, and there are relatively large matching problems. At this time, the matching effect is not ideal, and a high-abnormality matching signal is generated; if the matching pre-evaluation index is less than or equal to the matching pre-evaluation index threshold, it indicates that the potential matching anomaly degree of skull typing and sleep habits is relatively low, the matching effect of the system is relatively stable, and the potential matching problems are relatively small. In this case, the matching process can continue, and the matching success rate is relatively high, and a low-abnormality matching signal is generated; The hidden danger perception module is used to intelligently perceive the recommended hidden dangers of the pillow core recommendation system according to the potential matching anomaly degree of skull typing and sleep habits; When a high-abnormality matching signal is generated for the first time, obtain the matching pre-evaluation index output by the matching pre-evaluation model subsequently to construct a matching pre-evaluation index time series: , where represents the matching pre-evaluation index output by the matching pre-evaluation model at the t-th time point, t = {1, 2,..., T}, and T is a positive integer; Calculate the index anomaly coefficient: , where represents the index anomaly coefficient, represents the matching pre-evaluation index output by the matching pre-evaluation model at the T-th time point, represents the mean value of the matching pre-evaluation index time series: , Indicates the standard deviation of the time series of the pre-match evaluation index: ; It should be noted that the above formulas are all dimensionless and only take their numerical values for calculation. Common methods for removing dimensions include Min-Max normalization, Z-Score standardization, etc., which will not be elaborated here; Compare the index anomaly coefficient with the preset index anomaly coefficient threshold to intelligently perceive the potential risks in the pillow core recommendation system, as follows: If the index anomaly coefficient is greater than the index anomaly coefficient threshold, it indicates that the current pre-match evaluation index deviates significantly from the normal range, suggesting that there may be long-term abnormal risks in the pillow core recommendation system, leading to its recommendation results becoming increasingly far from the actual needs of users, thus affecting the user's sleep quality and generating a recommendation risk abnormal signal; if the index anomaly coefficient is less than or equal to the index anomaly coefficient threshold, it indicates that the current pre-match evaluation index is within the normal range, suggesting that the matching results of the pillow core recommendation system are relatively stable, no obvious system risks are found, and there is no need to generate a recommendation risk abnormal signal; The present invention obtains the error information of the head type classification in real time through the head type classification monitoring module, calculates the head type classification error coefficient based on the error information of the head type classification to identify the degree of error in the head type classification; at the same time, obtains the classification deviation information of sleep habits and calculates the sleep habit classification deviation coefficient based on the classification deviation information of sleep habits to identify the degree of classification deviation of sleep habits; next, the pre-match evaluation module constructs a pre-match evaluation model based on the two, outputs the pre-match evaluation index, and realizes the early perception of potential anomalies before the matching of the head type and sleep habits; the pre-match division module then compares the evaluation index with the preset threshold to finely divide the degree of matching anomalies; the risk perception module constructs a time series of the pre-match evaluation index and calculates the index anomaly coefficient of the time series of the pre-match evaluation index to real-time perceive and warn of potential recommendation risks, ensuring that the pillow core recommendation is more accurate and stable, effectively improving the matching accuracy of personalized pillow core recommendations, reducing the recommendation risks caused by data errors and model deviations, and moreover, through an intelligent and dynamically adaptive risk perception and warning mechanism, it can provide timely warnings for optimizing the matching strategy, improve the user's sleep experience and health protection, and realize the deep integration of intelligent health monitoring and personalized product recommendations.

[0019] The above formulas are all dimensionless and only take their numerical values for calculation. The formula is obtained by collecting a large amount of data and performing software simulation to obtain a formula closest to the real situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0020] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that contains one or more sets of available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0021] It should be understood that in various embodiments of the present application, the order numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0022] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs, etc., which can store program codes.

[0023] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

Claims

1. A pillow core recommendation system based on the combination of artificial intelligence-based head and neck typing and sleep habits, characterized in that: It includes a cranial classification monitoring module, a sleep habit monitoring module, a pre-matching evaluation module, a pre-matching division module, and a hidden danger perception module; The cranial classification monitoring module is used to obtain the error information of cranial classification, calculate the cranial classification error coefficient according to the error information of cranial classification, and identify the error degree of cranial classification; The sleep habit monitoring module is used to obtain the classification deviation information of sleep habits, calculate the sleep habit classification deviation coefficient according to the classification deviation information of sleep habits, and identify the classification deviation degree of sleep habits; The pre-matching evaluation module is used to construct a pre-matching evaluation model based on the cranial classification error coefficient and the sleep habit classification deviation coefficient, output a pre-matching evaluation index, and perceive potential matching anomalies before the cranial classification and sleep habits are matched; The pre-matching division module is used to compare the pre-matching evaluation index with a preset pre-matching evaluation index threshold to divide the potential matching anomaly degree of the cranial classification and sleep habits; The hidden danger perception module is used to intelligently perceive the recommended hidden dangers of the pillow core recommendation system according to the potential matching anomaly degree of the cranial classification and sleep habits.

2. The pillow core recommendation system based on the combination of artificial intelligence head and neck typing and sleep habits according to claim 1, characterized in that: The acquisition logic of the cranial classification error coefficient is as follows: During the skull typing process, extract the data of the head feature points of each user and calculate the abnormal missing ratio of feature extraction for the user: , where represents the abnormal missing ratio of feature extraction for the user, represents the number of missing feature points on the user's head, represents the total number of feature points that should be extracted; construct a set of abnormal missing data for feature extraction according to the abnormal missing ratio of feature extraction for each user , where represents the abnormal missing ratio of feature extraction for the th user, , is the total number of users participating in skull typing; Calculate the distances between the abnormal missing ratios of all user feature extractions and construct a distance matrix : , where is the distance between the abnormal missing ratios of user feature extractions, representing the and user absolute difference in the abnormal missing ratios of feature extractions; Construct a feature extraction abnormal topological structure according to the distance matrix, specifically as follows: A1. Compare the distance between the abnormal missing ratios of user feature extraction with a preset distance threshold. If the distance between the abnormal missing ratios of user feature extraction is less than or equal to the distance threshold, establish an edge in the topological complex; A2. For any three users, if the distance between any two users is less than or equal to the distance threshold, form a triangular surface; Traverse the distance matrix and repeat A1 and A2 to construct the feature extraction abnormal topological structure; During the construction of the abnormal topology of feature extraction, record the birth time and death time of each topological feature, and calculate the persistence: , where represents the persistence of the -th topological feature, represents the birth time of the -th topological feature, represents the death time of the -th topological feature; Calculate the cranial classification error coefficient: , where represents the cranial classification error coefficient, , represents the total number of topological features.

3. The pillow core recommendation system based on the combination of artificial intelligence-based head and neck typing and sleep habits according to claim 1, wherein: The acquisition logic of the sleep habit classification deviation coefficient is as follows: For each user to obtain the probability distribution of the sleep habit categories output by the classification model, calculate the -th user's sleep habit classification information entropy: , where represents the -th user's sleep habit classification information entropy, represents the -th sleep habit category's probability in the classification result of the -th user, , is the total number of sleep habit categories; Construct a set of sleep habit classification information entropy based on the sleep habit classification information entropy of each user , where represents the sleep habit classification information entropy of the th user, , and is the total number of users participating in sleep habit classification; Calculate the th user and the th user's deviation of the entropy of sleep habit classification: , where represents the th user and the th user's deviation of the entropy of sleep habit classification, represents the th user's entropy of sleep habit classification; Calculate the classification deviation coefficient of sleep habits: , where represents the classification deviation coefficient of sleep habits, .

4. The pillow core recommendation system based on the combination of artificial intelligence-based head and neck typing and sleep habits according to claim 1, characterized in that: Construct a matching pre-assessment model based on the cranial classification error coefficient and the sleep habit classification deviation coefficient, and output the matching pre-assessment index , the formula on which the model is based is as follows , where represents the cranial classification error coefficient, represents the sleep habit classification deviation coefficient, respectively represent the preset proportionality coefficients of the cranial classification error coefficient and the sleep habit classification deviation coefficient, and are both greater than 0.

5. The pillow core recommendation system based on the combination of artificial intelligence-based head and neck typing and sleep habits according to claim 4, characterized in that: Compare the pre-matching evaluation index with a preset pre-matching evaluation index threshold to divide the potential matching anomaly degree of the cranial classification and sleep habits, specifically as follows: If the pre-matching evaluation index is greater than the pre-matching evaluation index threshold, generate a high-abnormality matching signal; if the pre-matching evaluation index is less than or equal to the pre-matching evaluation index threshold, generate a low-abnormality matching signal.

6. The pillow core recommendation system based on the combination of artificial intelligence-based head and neck typing and sleep habits according to claim 5, characterized in that: When the high anomaly matching signal is first generated, obtain the matching pre-evaluation index subsequently output by the matching pre-evaluation model to construct a time series of matching pre-evaluation indices: , where represents the matching pre-evaluation index output by the matching pre-evaluation model at the t-th time point, t = {1, 2,..., T}, and T is a positive integer; Calculate the exponential anomaly coefficient: , where represents the exponential anomaly coefficient, represents the pre-matching evaluation index output by the pre-matching evaluation model at the T-th time point, represents the mean of the time series of the pre-matching evaluation index: , represents the standard deviation of the time series of the pre-matching evaluation index: .

7. The pillow core recommendation system based on the combination of artificial intelligence-based head and neck typing and sleep habits according to claim 6, characterized in that: Compare the index abnormality coefficient with a preset index abnormality coefficient threshold to intelligently perceive the recommended hidden dangers of the pillow core recommendation system, specifically as follows: If the index abnormality coefficient is greater than the index abnormality coefficient threshold, generate a recommended hidden danger abnormality signal; if the index abnormality coefficient is less than or equal to the index abnormality coefficient threshold, there is no need to generate a recommended hidden danger abnormality signal.

Images