Life semantic tensor map modeling method

Abstract

The invention provides a life semantic tensor map modeling method, which is used for embedding Chinese traditional'five-element '(gift, meaning, kernel, debit and dao) philosophy value into a DIKWP five-layer semantic architecture (data-information-knowledge-intelligence-intention) to realize unified modeling of physiological and psychological multi-source data. A high-dimensional semantic representation space is formed by constructing five semantic subspaces and a tensor relation, and multi-level reasoning and closed-loop feedback are supported. The method covers data acquisition, semantic coding, map construction, inference analysis and intention decision, can integrate physiological signals such as heart rate and psychological data such as emotion and questionnaires, reveals deep association, provides physical and mental integrated semantic interpretation for disease diagnosis and generates personalized regulation suggestions. Oriental philosophy and an AI model are fused, the cognitive and decision-making ability of medical AI for complex physical and mental problems is improved, innovativeness is high, and the application value is high.

Description

Technical Field

[0001] This invention relates to the fields of artificial intelligence, knowledge graphs, and medical informatics, and particularly to a data modeling method that integrates the traditional Chinese "Five Elements" philosophical values ​​with the DIKWP five-dimensional semantic architecture, specifically a life semantic tensor graph modeling method. This method is used for high-level semantic representation and reasoning of multimodal data of living organisms, and can be applied to scenarios such as physical and mental health status analysis and disease diagnosis. Background Technology

[0002] With the development of technologies such as wearable sensors, biomonitoring, and affective computing, it is possible to collect a large amount of multimodal vital data on individuals, including physiological signals (such as heart rate, blood pressure, and blood sugar), emotional characteristics of language text or speech, and psychological questionnaire scores. This data contains important information about an individual's physiological and psychological state. However, current technologies often analyze physiological and psychological indicators separately, lacking a unified semantic understanding of the relationship between the two. For example, in traditional medical diagnosis, physiological data is analyzed by medical models, while psychological emotions are assessed through psychological scales; the two lack integration and fail to comprehensively reflect the patient's overall health status, encompassing both mind and body.

[0003] Existing AI semantic modeling frameworks are mostly based on the DIKW model, a four-level cognitive architecture of Data-Information-Knowledge-Wisdom. The DIKWP model, proposed in recent years, adds a Purpose layer to DIKW, enabling the semantic reasoning process to possess goal-oriented and closed-loop feedback characteristics. The DIKWP model emphasizes the guiding role of high-level intent on low-level cognitive processes, which is of great significance in AI cognitive and decision-making systems. However, existing DIKWP models mainly focus on general cognitive processes and lack targeted design for semantic fusion of multi-source data in specific application areas (such as healthcare). Furthermore, current knowledge graph and semantic network technologies can integrate multi-source heterogeneous data under a unified framework, for example, by fusing health data through ontology and fuzzy logic to achieve semantic interoperability. However, these methods typically remain at the data and information level of integration, lacking the introduction of higher-level semantics (such as wisdom and intent layers) and humanistic values.

[0004] Traditional Chinese philosophy contains rich thoughts on the unity of body and mind and the unity of knowledge and action. For example, Confucian and Taoist thoughts have distilled several core value concepts: propriety, righteousness, benevolence, virtue, and the Way. These values are regarded as different levels of a person's conduct and cognitive improvement: propriety represents external norms and order, righteousness represents internal criteria of right and wrong, benevolence embodies the emotions of compassion and love and interpersonal care, virtue refers to the cultivated moral character, and the Way refers to the highest-level cosmic law or ultimate goal. In traditional health preservation and medical ethics, these concepts are used to emphasize the balance between physiology and psychology and the impact of moral emotions on health. However, in modern artificial intelligence and medical data analysis, there has been no effective attempt to integrate these Eastern philosophical values into technical models. Most existing medical AIs are based on the Western biomedical paradigm and lack a semantic description of the patient's psychological-physiological whole, resulting in the inability to often give meaningful explanations and intervention suggestions for physiological symptoms caused by psychological factors (or vice versa).

[0005] Thus, how to perform unified semantic representation and reasoning on multi-source life data, especially to achieve collaborative interpretation of physiological-psychological data by combining the dimension of humanistic philosophy, is a technical problem in the current field of intelligent medicine. If the data of the living body can be fused with traditional Chinese "five-element" philosophical values at the semantic level and combined with a hierarchical cognitive architecture such as DIKWP, it is expected to achieve a more comprehensive and profound understanding of an individual's health status and provide more targeted diagnosis and intervention plans. This is exactly the problem to be solved by this invention. Summary of the Invention

[0006] The main purpose of this invention is to provide a method for modeling a life semantic tensor map, which embeds traditional Chinese "five-element" philosophical values (propriety, righteousness, benevolence, virtue, and the Way) into the five-dimensional semantic structure of DIKWP (data-information-knowledge-wisdom-intention) for unified semantic representation, reasoning, and feedback on multi-dimensional data of the living body. Through this method, the semantic gap between physiological data and psychological data can be bridged, and an interpretable high-dimensional semantic representation space can be constructed, so as to achieve unified interpretation of physical and mental data in disease diagnosis and personalized adjustment feedback.

[0007] The technical problem to be solved by this invention is: how to fuse multi-source heterogeneous life data (physiology and psychology) and introduce the "five-element" philosophical value system at the semantic level to construct a semantic model with multi-level reasoning and closed-loop feedback capabilities. To solve this problem, this invention proposes a technical solution that combines five philosophical value dimensions with the DIKWP hierarchical semantic architecture in the form of a tensor map to perform high-level semantic annotation and deep association on the data, achieving a comprehensive description and causal reasoning of an individual's state.

[0008] To achieve the above object, this invention provides a method for modeling a life semantic tensor map, the overall architecture of which is as Figure 1As shown below. This method mainly includes the following steps:

[0009] Multi-source data collection and preprocessing: Obtain multi-dimensional data about the living body, including but not limited to physiological sensing data (such as heart rate, blood pressure, electroencephalogram, etc.), language / text emotion data (such as speech emotion analysis, emotional polarity of diary texts), and scores of psychological questionnaires or scales (such as self-rating depression scale, anxiety scale, etc.). Preprocess the data from different sources, including noise filtering, feature extraction, and time synchronization, etc., to convert it into a standardized data representation. For example, extract statistical features from time series signals such as heart rate, and convert text into an emotional score vector, etc.

[0010] Semantic encoding and five-element subspace mapping: At the "data" layer of the DIKWP architecture, assign preliminary semantic labels to the preprocessed data and map it to the five semantic subspaces of the "five-element" philosophy. Specifically, define five semantic subspaces corresponding to the five value dimensions of propriety, righteousness, benevolence, virtue, and the way. Each preprocessed data is mapped to one or more relevant subspaces according to its attributes and meanings: For example, physiological sensing data mainly reflects the objective state and regularity of the body and can be mapped to the propriety value range (indicating the degree of compliance with physiological norms and health standards); language emotion data reflects emotions and interpersonal care and can be mapped to the benevolence value range (indicating individual psychological emotions and empathy ability); the results of psychological questionnaires involve values and subjective feelings and can be partially mapped to the righteousness value range (internal right and wrong judgment) and the virtue value range (individual conduct / behavior consistency), and at the same time, those indicators about the sense of meaning and purpose in life can be mapped to the way value range (pointing to the ultimate goal). Through this mapping, the original data is converted into a symbolic or vector representation with "five-element" philosophical semantic tags. For example, a heart rate data record can be expressed as {propriety: deviating from the benchmark by 0.1}, and a diary text expressing negative emotions can be expressed as {benevolence: negative 0.8, the way: confused 0.5}, etc.

[0011] Tensor Graph Construction: The semantically labeled data is fused and synchronized within the DIKWP architecture, and a semantic tensor graph is constructed hierarchically, starting from the information layer. At the information layer, association analysis is performed on the symbols of each subspace mapped from the data layer to form preliminary information units. For example, simultaneously collected high heart rate (abnormality in the ritual dimension) and high anxiety (abnormality in the benevolence dimension) are associated with the information: "Current stress level is elevated." At the knowledge layer, domain knowledge and historical data are introduced to perform pattern recognition and causal reasoning on the associations in the information layer. For example, the knowledge layer can identify patterns such as "Long-term stress (benevolence) may lead to fluctuations in physiological indicators (ritual)" and represent them as knowledge graph relationship nodes across subspaces. Here, each piece of knowledge can be represented as a tensor entity or relationship involving multiple dimensions, such as the association strength of <stress, heart rate> in the (benevolence, ritual) subspace. Simultaneously, the relationships within or between different philosophical subspaces are characterized in tensor form: that is, a high-dimensional array or matrix is ​​used to represent the connections between the five value dimensions at different levels. Figure 2 As shown, the semantic tensor graph comprises a matrix structure formed by unfolding five subspace dimensions along the DIKWP hierarchy. Each unit node represents a semantic entity or relation at a certain level and value dimension. The multi-level structure of the tensor characterizes the interactions between different levels and value dimensions. For example, at the knowledge layer, the tensor slices between the etiquette and benevolence dimensions reflect the knowledge association between physiology and emotion.

[0012] Multi-layered semantic reasoning: After completing the knowledge layer modeling, the wisdom layer performs comprehensive reasoning on the tensor graph to extract higher-level insights. The wisdom layer combines interdisciplinary professional knowledge and ethical principles to evaluate and integrate the knowledge layer output, forming a global judgment on the individual's state. For example, the wisdom layer might comprehensively analyze: "Repeated abnormal physiological indicators accompanied by long-term negative emotions indicate psychosomatic imbalance." This judgment combines knowledge from the dimensions of propriety and benevolence, and introduces considerations from the dimension of virtue (such as whether the patient consistently follows medical advice and maintains good lifestyle habits, which falls under the category of virtue). The wisdom layer reasoning can use a rule-based reasoning engine or a deep learning model to infer from the tensor graph, obtaining weighted conclusions. Because this invention introduces an intention layer (the dimension of Tao), the reasoning process of the wisdom layer also references the ultimate goal to ensure that the reasoning results are consistent with the individual's health goals. For example, if the system's ultimate goal (Tao) is to improve the patient's overall psychosomatic health, then the wisdom layer will focus on identifying the key factors hindering this goal.

[0013] Intent-Level Decision-Making and Feedback: At the highest intent level, this invention transforms reasoning results into specific decisions and feedback measures, achieving closed-loop regulation. The intent level, combined with the value orientation of the Dao dimension (i.e., the ultimate goal of pursuing harmony between body and mind), generates strategies based on the conclusions of the wisdom level, outputting feedback information containing diagnostic explanations and regulatory suggestions. For example, the system might generate a conclusion like this: "The patient's recent blood pressure elevation (Li) is mainly due to anxiety (Ren) caused by excessive work pressure, leading to insufficient sleep and decreased medication adherence (De). It is recommended to adjust the dosage of antihypertensive medication, implement psychological stress reduction interventions (such as mindfulness training), and establish a regular work-rest schedule (Li) to alleviate stress and fundamentally improve health." This feedback includes both a medical explanation of the physiological abnormality and points out the underlying psychological causes and corresponding intervention plans, reflecting a comprehensive consideration of multiple levels of Li-Ren-De-Dao. This feedback will act on the patient and clinicians, guiding subsequent treatment and lifestyle adjustments. Simultaneously, the system enters the next monitoring-analysis cycle, achieving closed-loop regulation: the decisions at the intent level influence new data collection and interpretation biases (e.g., a greater focus on indicators of patient psychological stress), thereby continuously optimizing the model's semantic understanding of individuals. This intention-driven two-way feedback mechanism ensures that the model can dynamically adapt to changes in individual states, improving the timeliness and effectiveness of diagnosis and intervention.

[0014] Through the above steps, this invention constructs a unified high-dimensional semantic representation space that integrates the Five Elements Philosophy and the DIKWP architecture. In this space, any living organism's data corresponds to a semantic vector or node with five value-dimensional coordinates, and different data and their derived information are linked into a network through tensor relationships. This semantic tensor graph possesses both the interpretable structure of a knowledge graph and contains multidimensional relational information of tensors, enabling the system to perform complex pattern recognition and reasoning. For example, when a patient exhibits new symptom data, the system can map it to the semantic space and quickly find related past psychological states, thereby determining whether the symptom is caused by psychological factors or purely physiological reasons. This significantly improves the ability to identify psychological-physiological interactions, contributing to more accurate disease diagnosis and customized treatment plans.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] Unified Semantic Interpretation of Mind-Body Data Fusion: This invention represents physiological indicators and psychological states within the same semantic framework through a "five-element" semantic subspace. This enables the explanation of cross-domain phenomena such as "the influence of emotional changes on physiological indicators," achieving a unified understanding of mind-body interaction. This fills the gap in existing medical AI regarding the fusion of psychological and physiological data.

[0017] Multi-layered Reasoning and Interpretability: The hierarchical semantic model based on the DIKWP architecture enables the system to reason at multiple levels, including data, information, knowledge, wisdom, and intent. From specific data correlations to high-level health goals, each layer has a clear semantic meaning, making the decision-making process transparent and interpretable. In particular, by incorporating Chinese philosophical values, many reasoning results can be explained using popular humanistic concepts, such as "health imbalance caused by failure to follow proper etiquette," improving the user's understanding and acceptance of professional conclusions.

[0018] Closed-loop feedback and personalized adjustment: By incorporating the intent layer, a closed loop is formed from data collection to decision feedback. The system can dynamically adjust its focus and processing of lower-level data based on higher-level objectives. For example, when the goal is to reduce patients' psychological stress, the system will proactively increase the weight of the emotional subspace data and provide psychological intervention suggestions. This adaptive feedback mechanism improves the model's ability to cope with individual differences and state changes, making interventions more personalized and effective.

[0019] An innovative semantic tensor graph structure: This invention introduces the philosophical value dimension into semantic tensor representation, forming a novel tensor knowledge graph structure. Compared to the binary representation of traditional knowledge graphs that only have node-edge relationships, tensor graphs can represent multiple relationships (such as associations involving propriety, benevolence, and virtue simultaneously), better capturing high-dimensional interactions in complex systems. This structure has the potential to be extended to other fields requiring multimodal and multi-value comprehensive judgments, possessing broad application prospects and patent exclusivity.

[0020] In summary, the life semantic tensor graph modeling method provided by this invention can realize holographic semantic modeling of living organism data, and is particularly suitable for auxiliary diagnosis and decision support of complex diseases with both physiological and psychological factors in the medical field, demonstrating significant practical value and innovation. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall architecture of the life semantic tensor graph modeling method of the present invention.

[0022] Figure 2 This is a schematic diagram of the tensor relationship between the five-element semantic subspace constructed in this invention and the DIKWP hierarchy.

[0023] Figure 3 This is a flowchart of the life semantic tensor graph modeling method of the present invention.

[0024] Figure 4 This is a schematic diagram illustrating the integrated application of the present invention in a medical artificial intelligence system. Detailed Implementation

[0025] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings. The overall structure of the present invention can be found in [reference needed]. Figure 1 The system mainly comprises a data acquisition module, a semantic mapping module, a tensor graph construction module, and an inference and feedback module. These modules can be implemented by hardware and software units on a computing device, such as sensor interfaces, electrical signal processors, knowledge graph databases, inference engines, and user interfaces. This embodiment uses a medical AI system for chronic disease management as an example to illustrate the workflow and effects of the method of the present invention.

[0026] (1) Multi-source data acquisition and preprocessing: The system first acquires various data from the patient through the data acquisition module. Taking a middle-aged patient with hypertension and depressive tendencies as an example: wearable devices continuously monitor physiological indicators such as heart rate and blood pressure, and an application on a smartphone collects the patient's daily voice and text information (for sentiment analysis). The patient fills out a mental health questionnaire (such as the PHQ-9 depression assessment form) weekly. All data undergoes format conversion and feature extraction through the preprocessing module. For example, the heart rate signal is filtered to obtain features such as the mean resting heart rate and heart rate variability; the voice and text are processed by natural language processing to obtain emotion scores (probability of positive / negative emotions); and the questionnaire answers are quantified into scores for each psychological dimension. Preprocessing also includes time alignment of different data to ensure that physiological and psychological states within the same time period can be correlated for analysis.

[0027] (2) Semantic Encoding and Value Subspace Mapping: The preprocessed data is sent to the semantic mapping module for semantic annotation in the data layer of the DIKWP architecture. The system has a built-in "five-element value mapping rule base", which maps the data to the corresponding philosophical value semantic subspace according to the data type and context. For example, in this case, the patient's physiological indicators (heart rate, blood pressure) are marked as entities under the "Li" dimension, indicating the degree of deviation of their physical state from medical norms (such as the standard blood pressure range); their language emotion analysis results (such as detecting negative emotions) are marked as entities under the "Ren" dimension, indicating the patient's current emotional / psychological state; the questionnaire score involves multiple dimensions, among which the behavior reflecting the patient's compliance with medical advice and regular life is marked as the "De" dimension (virtue / behavioral habits), the items reflecting pain and despair are marked as the "Dao" dimension (lack of purpose / meaning), and the items related to moral confusion can be marked as the "Yi" dimension (value judgment). During the mapping process, rules from the knowledge base are used to transform quantitative data into qualitative semantics, such as: "Blood pressure 140 / 90, higher than normal" (Li: slightly high), "Emotional score slightly low" (Ren: negative), "Questionnaire shows strong sense of hopelessness in life" (Dao: lack of purpose), etc. This step outputs a set of basic units with semantic labels, called semantic data units. Each semantic data unit can be represented in the form of a triple: (data content, value label, quantity / level), for example: (blood pressure, Li, slightly high), (emotional state, Ren, negative 0.8), etc.

[0028] (3) Constructing a Semantic Tensor Graph: Next, the tensor graph construction module integrates all semantic data units into a unified semantic tensor graph. For example... Figure 2 As shown, the system maintains a semantic subspace for each of the five value dimensions, and semantic nodes are organized according to the DIKWP hierarchy within each subspace. Specifically, at the information layer, the system pairs or associates semantic units from the data layer to generate more meaningful information nodes. For example, in this case, both "high blood pressure (Li)" and "negative emotions (Ren)" appear simultaneously. The system combines this with the time factor to associate them into a single message: "Negative emotions may lead to high blood pressure." This can be represented as a cross-dimensional information node Info1 = {Li: Blood Pressure↑, Ren: Emotion↓}. Similarly, if the questionnaire reveals that the patient has recently been frequently staying up late and not taking medication on time (a Virtue dimension question), which is also associated with abnormal physiological indicators, then the information node Info2 = {Li: Blood Pressure↑, Virtue: Poor Compliance} is generated, representing the association between physiological deterioration and unhealthy lifestyle habits. In the information layer, nodes from different value subspaces form cross-dimensional information relationships through association (see [link to relevant documentation]). Figure 2 (As shown by the dashed lines in the middle information layer). These information nodes are essentially two-dimensional slices in the semantic tensor graph, encompassing combinations of multiple value dimensions at the information layer.

[0029] Furthermore, at the knowledge layer, the system incorporates a medical knowledge base and historical data for deeper reasoning. The knowledge base includes existing medical theories (such as the causal knowledge that "stress raises blood pressure," corresponding to the relationship between benevolence and propriety) and statistical correlations (such as the pattern derived from follow-up data that "failure to take medication as prescribed leads to worsening of the condition," corresponding to the relationship between virtue and propriety). At the knowledge layer, the tensor graph combines this general knowledge with the current patient's information nodes to form weighted knowledge nodes. For example, fusing Info1 with medical knowledge yields knowledge node K1: "The patient's recent increase in blood pressure is very likely caused by long-term psychological stress," where "very likely" is a weight or confidence level. Fusing Info2 with knowledge yields knowledge node K2: "The patient's poor blood pressure control is closely related to poor adherence." Nodes in the knowledge layer typically involve combinations of three or more dimensions (equivalent to third-order or higher relationships in tensors). For example, K1 involves a ternary relationship of benevolence-propriety-the Way (emotional, physiological, and health goals): because stress (benevolence) remains unabated, physiological (propriety) indicators are difficult to control, threatening overall health goals (the Way); K2 involves a relationship of propriety-virtue-the Way: lifestyle habits (virtue) affect physiology (propriety), thus deviating from health goals (the Way). These higher-order relationships are represented using tensor structures. In terms of data structure implementation, multidimensional arrays or associative lists can be used for storage: for example, a 5-dimensional array T[d,i,k,w,p], with indices corresponding to propriety, righteousness, benevolence, virtue, and the Way, where a value of 1 is assigned at certain combination indices in the knowledge layer to indicate the existence of that multidimensional relationship. If a graph database is used, relationships are recorded using hyperedges or tuples. Through this tensor-based processing, the model can accommodate and represent complex multi-factor relationships.

[0030] (4) Multi-level semantic reasoning: The semantic reasoning module performs reasoning deduction on the constructed tensor graph. First, it integrates all knowledge layer nodes at the wisdom layer to form a comprehensive assessment of the patient's condition. In this example, the wisdom layer summarizes K1, K2, and other potentially related knowledge nodes, and then combines ethical and humanistic principles for assessment. The wisdom layer considers that the patient is in a state of high blood pressure (a problem in the benevolence dimension) and that blood pressure has not been effectively controlled for a long time (a problem in the propriety dimension), while there are also compliance and lifestyle problems (a problem in the virtue dimension). Based on this, it gives an overall judgment: "The patient's mind and body are out of balance, and the current treatment plan needs to be adjusted." This judgment process may use fuzzy reasoning algorithms or rule-based expert systems to synthesize factors of different dimensions into a high-level conclusion according to certain weights. For example, the wisdom layer rules or models can be set as follows: strong correlation between benevolence and propriety dimensions + problems in the virtue dimension => conclusion of mind and body imbalance, and this conclusion points to the goal that needs to be adjusted (the Tao dimension). Since this invention integrates the intention layer, it will also check whether the comprehensive assessment conforms to the system goal, namely, improving the patient's health, which is the "Tao". If it does not meet the requirements, the gap is identified, for example, "The main reasons for the current state deviating from the health goal are psychological stress and bad habits."

[0031] It should be noted that the reasoning of the intelligent layer not only relies on preset rules but can also utilize machine learning models. For example, a deep neural network can be trained to encode tensor graphs into vectors for classification or regression, outputting health risk scores, etc. In this case, the method of this invention provides rich feature dimensions (propriety, righteousness, benevolence, virtue, and the Way), enabling the model to learn complex patterns of mind-body interaction. Regardless of the reasoning mechanism used, the output of the intelligent layer is a highly abstract and semantically clear description of the patient's current condition, in a human-understandable form. For example, in this case, it can be expressed as: "Due to long-term high stress (benevolence) and poor compliance (virtue), the patient's hypertension (propriety) is not well controlled, and their overall health (the Way) is at risk."

[0032] (5) Intent-Level Decision-Making and Feedback: At the intent level, based on the aforementioned intelligent assessment, the system formulates specific intervention decisions and provides feedback to users and medical personnel. For example... Figure 3 and Figure 4 As shown, the intent layer is equivalent to the intent unit of the entire model, clearly defining the goals pursued by the system (e.g., controlling blood pressure, improving mental health, and enhancing quality of life). For factors that deviate from the goals, the system generates personalized intervention plans. Continuing the previous example, the intent layer decision module might output the following action suggestions:

[0033] Medical adjustments (etiquette level): It is recommended that the doctor increase the dosage of the patient's antihypertensive medication or change the medication to directly control physiological indicators. It is also recommended to arrange a comprehensive physical examination to obtain more data to support the diagnosis.

[0034] Psychological intervention (human level): It is recommended that patients receive psychological counseling or participate in stress reduction training, such as mindfulness meditation courses or stress management counseling, to reduce persistent mental stress.

[0035] Behavioral improvement (Germany level): Develop daily routines and medication reminder plans for patients, and use mobile applications to encourage patients to maintain a regular lifestyle and take medication on time. For example, set alarm reminders and check-in mechanisms to improve adherence.

[0036] Education and Communication (Meaning Level): Explain the current situation to the patient and their family, emphasizing the importance of managing both mental and physical health. This is essentially value-based guidance, helping the patient establish correct understanding (e.g., "emotional regulation is part of treatment").

[0037] Goal Resetting (Taoist Level): Collaborate with the patient to develop a realistic and achievable overall health goal, such as "stabilizing blood pressure within the XX range and reducing anxiety score by X points within three months." By motivating through these goals, the patient's initiative to improve their physical and mental well-being is enhanced, demonstrating the guiding role of "Tao."

[0038] The above suggestions will be provided to patients and their doctors through a user interface, forming a clear diagnostic report and action plan. The report will include a semantic explanation section (e.g., "Analysis of the main causes of elevated blood pressure"), enabling doctors and patients to understand the rationale behind the suggestions and increasing their willingness to adopt them. This humanized explanatory power is attributed to the invention's integration of traditional philosophical semantics, making AI decision-making no longer a black box of numerical values, but rather a statement with certain moral and purposeful implications. After feedback implementation, the system continues to monitor subsequent changes in the patient's data. If the measures are effective, the tensor graph will be updated accordingly; for example, blood pressure returning to normal will be mapped to "the ritual dimension returning to normal," and improvement in psychological scores will be mapped to "improvement in the benevolence dimension," with the weight of the associated nodes decreasing. The system adjusts the model accordingly and enters the next cycle. If there are still areas where the goal has not been achieved, the system will analyze the reasons again and continuously optimize the intervention strategy.

[0039] It should be noted that the method of this invention can be implemented in a computer system using a combination of software and hardware. The semantic tensor graph can be constructed using an object-oriented database or triplet storage, and the quinary subspace and DIKWP hierarchy can be distinguished by adding labels or dimension fields. The reasoning module can implement advanced logical deduction using a knowledge reasoning engine (such as the rule-based ExpertSystem), or implement pattern recognition and prediction using a trained deep learning model (such as a graph neural network, tensor decomposition model, etc.). The introduction of the five-dimensional philosophical value can be seen as adding a new set of relation types or feature dimensions to the traditional knowledge graph. Developers can add "quinary" labels to nodes and relations based on existing open-source knowledge graph frameworks, thereby utilizing existing reasoning algorithms to process the semantic graph of this invention. In addition, machine learning tools such as TensorFlow can be used to input the five-dimensional semantic representation into the model for training, allowing the model to autonomously learn the correlations between the dimensions. These implementation methods are all within the scope of the concept of this invention and do not affect the implementation of its core idea.

[0040] Application Scenarios Expansion: Although this embodiment focuses on healthcare, the methodology of this invention is also applicable to semantic modeling of living organisms in other fields. For example, in the field of smart elderly care, it integrates the physiological indicators and emotional states of the elderly to predict health risks; in the field of human-computer interaction, it combines users' physiological signals (such as eye movements and heartbeats) and emotional feedback to improve the robot's understanding of user intentions; and even in the field of education, it achieves personalized teaching intervention through semantic analysis of physiological feedback (fatigue) and psychological states (interest, frustration). These scenarios all require the integration of multi-source data and multi-layered meanings. The five-element semantic tensor model of this invention provides a universal methodology, integrating Eastern philosophical wisdom into the semantic framework of artificial intelligence, and offering a new approach to building "human-centered" intelligent systems.

Claims

1. A method of life semantic tensor graph modeling (embedding the DIKWP semantic structure of the "five-element philosophy"), characterized in that, The method comprises the following steps: Collecting multi-source data of a living body, including physiological sensing data, language emotion data and psychological questionnaire data; Preprocessing and feature extraction of the data; Semantic encoding of the preprocessed data in the data layer of the DIKWP architecture, mapping the data to five semantic subspaces of Li, Yi, Ren, De and Dao based on the traditional Chinese "five-element" philosophy, constructing a semantic tensor graph, and performing multi-dimensional correlation analysis on the mapped semantic data in the information layer to generate information nodes, and fusing domain knowledge in the knowledge layer to form high-order semantic relationships; Inference fusion of the tensor graph in the wisdom layer to obtain a comprehensive assessment of the state of the living body; finally, generating a decision feedback in the intention layer to diagnose and explain the living body and provide adjustment suggestions, and feeding back the feedback to the data collection and interpretation process to form a closed loop.

2. The method of claim 1, wherein: The five semantic subspaces correspond to Li, Yi, Ren, De and Dao in the Chinese philosophical value system, respectively, wherein the Li subspace is used to represent the semantics of the objective behavior and physiological state of the living body relative to the standard, the Yi subspace is used to represent the semantics of the relevant ethical judgment and standard compliance of the living body, the Ren subspace is used to represent the semantics of the emotion and psychological state of the living body, the De subspace is used to represent the semantics of the behavior habit and character cultivation of the living body, and the Dao subspace is used to represent the semantics of the overall purpose, intention and ultimate value pursuit of the living body.

3. The method of claim 1, wherein: The semantic tensor graph uses a tensor or high-dimensional graph structure to represent the relationships between different subspaces and different levels, and its construction includes: in the information layer, different subspace data nodes at the same time or with causal correlation are generated into information nodes spanning multiple value dimensions through correlation rules; in the knowledge layer, the information nodes are promoted to higher-level knowledge nodes using a pre-constructed medical knowledge base and statistical model, and the knowledge nodes contain the combined relationship of two or more value dimensions; the nodes and relationships in the tensor graph have corresponding value dimension labels and level labels, thereby forming a five-dimensional (Li, Yi, Ren, De, Dao) by five-layer (data, information, knowledge, wisdom, intention) semantic matrix representation.

4. The method of claim 1, wherein: The multi-level semantic reasoning includes rule-based reasoning and / or machine learning model-based reasoning; in the reasoning process of the wisdom layer, the multiple value dimension relationships in the knowledge layer are comprehensively evaluated in combination with pre-set semantic rules and machine learning models to identify the patterns and abnormalities of the physical and mental state of the living body, wherein the target constraints of the intention layer are introduced to ensure that the reasoning result is consistent with the expected health goal; the output of the wisdom layer reasoning is the semantic description of the current state of the living body and the identification result of the priority problem.

5. The method of claim 1, wherein: The intention layer decision and feedback include generating an intervention scheme according to the evaluation result of the wisdom layer, the intervention scheme including physiological adjustment measures, psychological counseling measures and behavior / habit adjustment suggestions, and presenting the intervention scheme to the living body itself or its manager through a user interface; at the same time, the value dimension information targeted by the feedback scheme is input into the data collection or interpretation process of the next cycle to dynamically update the semantic tensor graph, thereby realizing closed-loop feedback regulation.

6. The method of any one of claims 1-5, wherein: The method is executed by an artificial intelligence system on a computing device, the system comprising a sensor data acquisition unit, a semantic processing unit and a decision support unit; wherein the semantic processing unit realizes storage and reasoning of the semantic tensor graph by using a knowledge graph database and a reasoning engine, and the decision support unit delivers feedback suggestions generated by the intention layer to the user end through an application program interface, thereby realizing health management services for the living body.

Images