A Full-Process Demonstration of the Unified DIKWP Expression-Execution Mechanism in AI-Based TCM Consultation Simulation
Yucong Duan
Benefactor: Shiming Gong
International Standardization Committee of Networked DIKWP for Artificial Intelligence Evaluation(DIKWP-SC)
World Artificial Consciousness CIC(WAC)
World Conference on Artificial Consciousness(WCAC)
(Email: duanyucong@hotmail.com)
Abstract
This paper is based on the case of traditional Chinese medicine diagnosis and treatment of “common cold-pharyngitis-tracheitis”, and is aimed at the cognitive level of middle school students to comprehensively demonstrate the “expression as execution” unified mechanism of the DIKWP semantic model in the artificial consciousness system. The DIKWP (Data-Information-Knowledge-Wisdom-Purpose) model, based on the classic DIKW (pyramid), introduces the “Purpose” layer and realizes bidirectional feedback between each level through a network semantic structure. This enables AI to possess a cognitive language shared by humans and machines, using a unified semantic representation to understand human instructions and guide its own actions. “Expression as execution” means that semantic expression not only has descriptive meaning but can also be directly used by AI as executable reasoning instructions. This paper first elaborates on the semantic executability of conceptual expressions, taking terms such as “wind-heat invading the lung” and “lung heat congestion” as examples to illustrate how they can be deconstructed into inferable semantic structures within the DIKWP framework and mapped to AI’s reasoning paths. Secondly, it analyzes how AI derives the DIKWP semantic action chain from natural language input to complete the closed-loop process of semantic understanding-reasoning-verification-anti-regulation. Then, it demonstrates the process of DIKWP semantic expression directly driving reasoning, including structured semantic transformation, causal chain graph construction, and the deduction of semantic computing nodes, supplemented by pseudocode or textual diagrams. Finally, it simulates the artificial consciousness system as a “learning doctor”, executing semantic reception, purpose generation, cognitive construction, vernacular output, and verification completion in real-time in real conversations, vividly demonstrating how the DIKWP model transforms semantic content into a computable, verifiable, and cognitively credible reasoning mechanism. Through the above full-process demonstration, we emphasize the unity of the expression and execution processes, as well as the bridging role of DIKWP semantic mathematics between language understanding and model reasoning, thereby proving that this mechanism can enhance the explainability, reliability, and cognitive credibility of AI diagnosis and treatment.
1Introduction
In the field of artificial intelligence medical diagnosis, accurately understanding and reasoning complex medical concepts is crucial. However, traditional large models mostly operate in a “black box” manner, making their internal decision-making processes difficult to explain and causing users to have concerns about trust and transparency. To address this challenge, Professor Yucong Duan’s team proposed the “Data-Information-Knowledge-Wisdom-Purpose” (DIKWP) artificial consciousness model, which extends the classic DIKW cognitive hierarchy by incorporating the Purpose layer as the highest level into the cognitive framework. The DIKWP model constructs a network-like five-layer semantic structure, enabling bidirectional interaction and iterative updating of semantic content between each layer, thereby significantly enhancing the explainability and controllability of AI decision-making processes. In other words, DIKWP provides a cognitive language shared by humans and machines, aligning the semantic computation within AI with human-understandable semantics. Under this semantic framework, every step of AI’s reasoning has a clear semantic representation that can be traced and explained, avoiding vague guesses purely based on statistical correlations. This endows the AI decision-making process with a “white box” characteristic, making the system more transparent, explainable, and aligned with the human knowledge system.
Traditional Chinese medicine (TCM) diagnosis simulation is an ideal scenario to verify this model. On the one hand, TCM diagnosis extensively uses abstract concepts (such as “wind-heat invading the lung” and “lung heat congestion”) to describe etiology and pathogenesis, which contain rich causal relationships and reasoning paths. On the other hand, the TCM diagnosis process has a distinct circular reasoning feature: physicians make preliminary judgments based on patient descriptions and then continuously verify or correct hypotheses through questioning and observation, ultimately forming a diagnosis and treatment plan. The above characteristics coincide with the closed-loop of semantic understanding-reasoning-verification in the DIKWP model. Therefore, this paper selects a common and representative TCM diagnosis and treatment case of “common cold-pharyngitis-tracheitis” for simulation. By gradually analyzing the entire process of AI from patient complaints to providing diagnostic and treatment suggestions, it demonstrates how the DIKWP semantic model achieves “expression as execution”.
The following text will first introduce how the DIKWP model endows TCM concepts with executable semantic representations, then elaborate on the process of AI constructing semantic action chains from natural language input. Next, it will demonstrate the causal graph and algorithm demonstration of semantic expression directly driving reasoning. Finally, through a “learning doctor” dialogue example, the entire semantic closed-loop process will be concretized. Our goal is to reveal the working principle of the DIKWP model in the artificial consciousness system in an easy-to-understand manner within the framework of an academic paper, proving its effectiveness and innovativeness as a bridge between language understanding and model reasoning.
2Semantic Executability of Conceptual Expressions
Many concepts in traditional Chinese medicine (TCM) theory are not only descriptive labels but also represent a series of pathological states and causal relationships. For example, the term “wind-heat invading the lung” describes the symptoms caused by the invasion of exogenous wind-heat pathogenic factors into the lung in TCM pathogenesis. Its literal disassembly is as follows: “wind-heat” represents the exogenous pathogenic factors with heat characteristics that invade the human body, “invading” indicates the action of invasion and encroachment, and “lung” is the organ affected by the pathogenic factors. Semantically, it can be structured into a causal triplet: etiological factor = wind-heat pathogenic qi, action = invasion, object of action = lung defense system (the defensive system of the lung). According to medical interpretation, wind-heat invading the lung leads to the dysfunction of the lung’s dispersing and descending functions, resulting in symptoms such as cough, yellow phlegm, sore throat, fever, etc. Therefore, this concept itself contains a reasoning chain of etiology 
site of disease 
pathological impact. For AI, if it can represent “wind-heat invading the lung” in the form of DIKWP, it is equivalent to having an executable reasoning rule built-in: when exogenous wind-heat pathogenic factors invade the lung, specific symptoms will appear, and the treatment should focus on dispersing wind, clearing heat, and resolving phlegm.
Similarly, “lung heat congestion” describes the pathological state caused by the accumulation of excessive internal heat in the lung, which is a type of excessive internal heat. Its literal meaning can be understood as: “lung heat” refers to the excess heat in the lung organ, and “congestion” means stagnation and accumulation, extremely vigorous. Medically, it is interpreted as the internal heat of the lung leading to symptoms such as high fever, thirst, cough and wheezing, sore throat, red tongue with yellow coating, etc. (also known as “lung fire” or “lung excess heat” syndrome). This concept can be abstracted as: internal cause = internal heat accumulation in the lung, state = congestion and vigorous, consequence = lung dysfunction (unfavorable lung qi) and a series of heat-related symptoms. For AI, “lung heat congestion” means that if a patient presents symptoms such as high fever, coughing up yellow phlegm, and sore throat, and combined with the cause of exogenous pathogenic factors entering the interior and transforming into heat, it can be inferred that there is heat accumulation in the lung, and the treatment should focus on clearing lung heat and purging fire.
The above two terminological examples show that TCM concepts often contain inferable causal relationships and treatment principles. In the DIKWP semantic model, we can map these concepts to formalized knowledge units. For example, the reasoning rule of the concept “wind-heat invading the lung” can be described in pseudocode as follows:
IF there is {wind-heat pathogenic qi invading the lung}:
THEN the lung's dispersing and descending functions will be impaired -> presenting {cough, yellow phlegm, sore throat, fever, ...}
and the treatment principle = dispersing wind, clearing heat, and resolving phlegm
Another example is the rule for “lung heat congestion”:
IF there is {heat pathogenic qi accumulating in the lung}:
THEN the lung will lose its descending function -> presenting {high fever, wheezing cough, thick yellow phlegm, sore throat, ...}
and the treatment principle = clearing lung heat, purging fire, and resolving phlegm to stop cough
Through the above structured representation, AI’s “understanding” of these concepts no longer stays at the level of string matching but has the ability to execute reasoning: when the input meets the conditions, the corresponding inferences and decisions of the concepts can be triggered. This reflects a prominent advantage of the DIKWP model - the unification of descriptive semantics and executable semantics. In traditional natural language processing, phrases like “wind-heat invading the lung” are just descriptive labels, and computers need additional rules or models to utilize their meanings for reasoning. However, under the DIKWP framework, since each concept has a strictly defined semantic structure and a mathematical representation method, “wind-heat invading the lung” itself becomes a semantic unit that can be directly executed by machines. As relevant research has pointed out: “Under the DIKWP framework, each level of semantics has a strict mathematical definition and transformation function... This makes any semantic statement not only have descriptive meaning but also can be executed and verified through algorithms.” In other words, for AI, expression is the starting point of reasoning: the logical relationships contained in the form of concept expression can be directly read and calculated, thereby shortening the distance from knowledge representation to reasoning execution.
3From Natural Language Input to DIKWP Semantic Action Chain
In actual diagnostic conversations, the information provided by patients is often a fragmented description in natural language. AI needs to transform these raw data/information into internal semantic representations in order to utilize existing knowledge for reasoning and decision-making and form intentions and action plans. This process corresponds to the hierarchical processing of the DIKWP model: extraction from Data to Information, matching from Information to Knowledge, decision-making from Knowledge to Wisdom, and forming the final action combined with Purpose. The entire process is not completed in one go but is realized through a closed loop of “understanding-reasoning-verification-adjustment” through cyclic iteration. We will analyze this process in stages below.
3.1Semantic Understanding: Parsing Patient’s Natural Language Complaints
Suppose a patient describes to the AI doctor: “I caught a cold from wind-cold a few days ago, and today I started having a sore throat, cough, and a bit of fever.” This sentence is the original input data (D). AI first enters the semantic reception stage, processes this natural language, and transforms the useful information into a structured form. For example, it extracts key information such as “a few days ago,” “caught a cold from wind-cold” (medical history), “today,” “sore throat” (pharyngitis), “cough,” “fever,” etc. At the information level (I) of DIKWP, these contents are organized into a series of attribute-value pairs or event descriptions: the timeline of the disease (evolution over several days), the list of symptoms (pharyngitis, cough, fever), and the cause of the disease (previously exposed to wind-cold). Through bidirectional semantic mapping, AI maps the natural language proposition to the corresponding DIKWP structured representation. Research has shown that for most propositions expressed by humans in natural language, there is almost always a corresponding DIKWP semantic structure; conversely, the semantic representation of DIKWP can also be converted back to readable natural language. This means that AI can rely on DIKWP as an “intermediate language” to understand human language, and the information remains semantically consistent in this back-and-forth conversion without deviating due to free text generation. In this example, the information-level representation obtained by AI may be similar to the following:
Time: Day 4 of illness (after being exposed to cold a few days ago)
Main symptoms: Sore throat, cough
Associated symptoms: Fever (low to moderate)
Clues to etiology: Likely caused by wind-cold common cold
3.2Knowledge Matching: Activating Corresponding DIKWP Knowledge Units
After extracting the above information, AI enters the knowledge layer (K) processing, that is, trying to match the current patient’s symptom pattern with the known disease/syndrome patterns in its knowledge base. This step is similar to the “syndrome differentiation” process of human physicians. Based on the aforementioned information, AI will search the DIKWP knowledge graph to find TCM syndrome nodes that match “sore throat, cough, fever, caused by wind-cold”. There may be two possible matches: one is wind-cold invading the lung (exogenous wind-cold pathogenic factors invading the lung), and the other is wind-heat invading the lung (wind-cold common cold transforming into heat and invading the lung after several days). AI will take into account the characteristics of the symptoms: if the patient has a significant fever and severe sore throat with yellowish phlegm, it is more likely to be wind-heat invading the lung; if it is only a mild fever with chills and thin phlegm, it is inclined to be wind-cold invading the lung or residual cold not yet cleared. Assuming that AI judges that it is more in line with the “wind-heat invading the lung” pattern based on the symptoms, then the corresponding knowledge node (such as the semantic unit of “wind-heat invading the lung” defined in the previous section) will be activated. This means that AI has already summarized the specific symptoms into a more abstract pathogenesis concept at the knowledge level.
It is important to note that at this point, the AI has only made a preliminary hypothesis, which still needs to be verified. In the DIKWP framework, before the output of the knowledge layer enters the wisdom layer (W) for evaluation and decision-making, it will check whether more information is needed to support it in combination with the goal of the purpose layer (P). In simple terms, AI asks itself: “Is my diagnostic hypothesis (wind-heat invading the lung) reliable? Do I need more evidence?” This is the beginning of the “verification - anti-regulation” in the closed loop. If the model is confident that the evidence is sufficient, it may directly enter the decision-making process; otherwise, it will generate the intention to obtain more information.
3.3Verification Questioning: Forming New Intentions and Feedback to the Data Layer
In this example, AI may decide to ask questions to collect more evidence, thereby executing a feedback loop on the semantic action chain. For example, the intention (P) generated by AI can be: “Ask the patient if there is swelling in the throat and the color of the phlegm to distinguish between wind-cold and wind-heat.” This intention is issued from the purpose layer and is transformed into a natural language question through the internal semantic planning of DIKWP: “Is your throat swollen? What color is the phlegm?” Here, AI’s question is in the form of vernacular output, but it has a clear semantic motivation behind it: to verify its diagnostic hypothesis of wind-heat invading the lung (because red and swollen throat, thick yellow phlegm are more supportive of wind-heat invading the lung, while thin and white phlegm is more inclined to wind-cold).
The patient’s response belongs to the new natural language input, and the closed loop returns to the semantic reception stage. For example, the patient replies: “The phlegm is a bit yellow, and my throat is very swollen, and it hurts to swallow.” AI parses it again: extracting information such as “yellow phlegm” and “obvious sore throat.” After categorization at the information level, these new pieces of information further confirm the previous hypothesis - at the knowledge level, “wind-heat invading the lung” is reinforced. However, if the patient’s response does not match the characteristics of wind-heat, AI will adjust the knowledge matching and may return to considering other diagnoses such as wind-cold. In this continuous cycle, AI achieves a “questioning while diagnosing” similar to human doctors: in each round, it updates its internal semantic state (data and information layers) based on the response, thereby correcting the knowledge judgment and the next intention. This closed-loop semantic reasoning is the core mechanism of the DIKWP artificial consciousness model, enabling the system to have self-regulation capabilities.
3.4Decision Making: Entering the Wisdom Layer and Drawing Conclusions
After several rounds of dialogue, AI has obtained enough information to confirm the diagnosis. For example, in this case, through detailed inquiry of symptoms, AI is convinced that the patient first caught a cold from wind-cold, then transformed into heat entering the interior, and currently belongs to wind-heat invading the lung, with a trend of heat evil conducting inward and congesting the lung. At this point, the wisdom layer (W) of the DIKWP model begins to play a role: based on the diagnosis from the knowledge layer and past experience, it makes decisions on the next actions. This step is equivalent to the doctor determining the treatment method and specific plan based on the diagnosis. In our simulation, the wisdom layer may deduce the logic of “If not cleared of heat in time, it may develop into lung heat congestion (tracheitis/pulmonary infection), so the treatment should not only disperse wind and clear heat but also clear lung fire to prevent the condition from worsening.” This decision-making process is also based on the semantic graph: AI has generated a causal chain deduction graph from exogenous wind-cold 
wind-heat invading the lung 
lung heat congestion within its internal system, clarifying the development of the disease and the timing of intervention. Through the deduction of this causal chain graph, the wisdom layer ensures that AI’s decisions have a logically consistent basis and serve the goals preset in the purpose layer (such as “control inflammation and prevent worsening of the condition”).
3.5Vernacular Output: Explaining Conclusions and Giving Suggestions
Finally, AI translates the decisions from the wisdom layer and the considered intentions into a response to the patient. While maintaining medical accuracy, the expression is as much as possible in everyday language (vernacular output) that the patient can understand. For example, the AI doctor may summarize to the patient as follows: “Judging from your symptoms, you initially caught a cold from wind-cold, but it has now turned into wind-heat invading the lung, causing sore throat and fever. In simple terms, the common cold did not fully improve and has evolved into signs of pharyngitis and bronchitis. Don’t worry too much. In our view, this is called ‘wind-heat invading the lung,’ which means that the exogenous heat evil has affected the respiratory system. Next, we need to eliminate the heat evil in the body and make the lung function smoothly to prevent it from further developing into a more serious lung infection.” In this passage, the term “wind-heat invading the lung” is explained in a popular way, while AI also implies the direction of treatment (clearing heat, promoting lung function). If this is a popular science diagnosis aimed at middle school students or the general public, AI may continue to give specific measure suggestions, such as traditional Chinese medicine or dietary therapy: “It is recommended that you take traditional Chinese medicine like Yin Qiao San to clear heat and detoxify, drink more water, and rest more. If the symptoms worsen, seek medical attention in time.” Through such output, AI completes the entire closed loop from patient complaints to explaining the condition and proposing suggestions. Internally, each step of its reasoning is traceable: from the initial data extraction, knowledge matching, to questioning and verification, causal deduction, all supported by DIKWP semantic nodes. This transparent reasoning chain not only makes AI’s behavior understandable to users but also facilitates developers in tracking the basis of AI’s decision-making, thereby assessing its credibility.
4The Reasoning Process Driven by DIKWP Semantic Expression
This section further demonstrates how DIKWP semantic expressions directly drive the internal reasoning of AI. We will structure the reasoning process of the case in the previous section in the form of causal chain graphs and semantic computing nodes, and provide pseudocode examples to highlight the mechanism of “expression as execution.”
4.1Causal Chain Graph: Semantic Evolution from Exogenous to Endogenous Disturbance
Based on the case situation, AI constructs the following etiology and pathogenesis causal chain graph:
Exogenous wind-cold -> Surface syndrome (initial common cold) -> Wind-heat invading the lung -> Pharyngitis (inflammation of the lung system) -> Lung heat congestion -> Tracheitis (deep into the lung, bronchitis)
The graph uses arrows to indicate the order of development and causal relationships. Each node corresponds to a semantic description:
·Exogenous wind-cold: Invasion of external cold pathogenic factors (initial cause of disease).
·Surface syndrome: The pathogenic factors are on the surface, with symptoms such as chills and fever (initial state).
·Wind-heat invading the lung (wind-heat invading the lung): The pathogenic factors transform into heat and invade the lung defense, causing symptoms such as sore throat, cough, and other lung-related symptoms.
·Pharyngitis: A Western medical term, corresponding to the inflammation of the throat, which matches the lung-related symptoms described in traditional Chinese medicine.
·Lung heat congestion: Internal heat congestion in the lung, with symptoms such as high fever, wheezing cough, and worsening sore throat.
·Tracheitis: A Western medical term, inflammation of the bronchial mucosa, often caused by infection or internal heat.
It can be seen that this causal chain corresponds the TCM syndromes (wind-cold, wind-heat invading the lung, lung heat congestion) with the Western medical disease names (common cold, pharyngitis, tracheitis), forming different expressions of the same disease process. In the DIKWP model, knowledge from different modalities and systems can be incorporated into the same semantic network. Each node in the graph is actually a semantic computing node, containing several inputs, outputs, and internal reasoning functions. For example, the internal logic of the “wind-heat invading the lung” node can be represented as:
Input conditions: Presence of exogenous pathogenic factors and symptoms such as sore throat, cough, and yellow phlegm.
Output: Confirm the syndrome as wind-heat invading the lung; generate corresponding treatment suggestions (dispelling wind, clearing heat, and relieving cough).
Subsequent: If not treated in time, it may develop into the next node ``lung heat congestion.''
The expression of this node itself acts as a reasoning step: when the input conditions are met, the node will output the corresponding conclusion and shift the focus to the subsequent node. For AI, such semantic nodes are like executable code or functions. In the DIKWP knowledge graph, nodes are connected by causal relationships, allowing AI to traverse the graph for reasoning. In comparison, without such a clear semantic graph, traditional AI may only be able to guess the next state through probabilistic associations. However, DIKWP provides AI with a clear logical path: from the invasion of exogenous pathogenic factors to internal transmission, each step has a medical basis and semantic explanation. This logical path can be automatically executed within AI through graph traversal algorithms or rule engines. For example, when AI locates the current node as “wind-heat invading the lung” in the knowledge graph and detects that the condition “symptoms persist/worsen” is met, it will follow the “leads to” edge to the “lung heat congestion” node, predict the possible situation in the next stage, and adjust the current decision accordingly (such as giving lung-clearing and fire-purging medicine in advance to prevent the condition from worsening).
4.2Semantic Computing Node Deduction: Pseudocode Illustration
To more clearly illustrate the reasoning driven by semantic expressions, we use pseudocode to demonstrate the deduction process within AI. The following pseudocode is based on the DIKWP semantic network for reasoning and decision-making in this case:
# Obtain the relevant node objects from the knowledge graph
windCold = KnowledgeGraph.getNode("Exogenous wind-cold")windHeat = KnowledgeGraph.getNode("Wind-heat invading the lung")lungHeat = KnowledgeGraph.getNode("Lung heat congestion")# Initial state: Hypothesis of the patient's etiology and pathogenesiscurrent_node = windHeat # Assume the current syndrome is wind-heat invading the lung# Execute the verification phaseif patient.symptoms.contains(["sore throat","yellow phlegm","fever"]) and patient.history.contains("exposed to wind-cold"): current_node.confirmed = Trueelse: current_node.confirmed = False # Adjust the hypothesis based on symptoms (omitted)# If the current node is confirmed, execute the corresponding reasoning and decision-makingif current_node == windHeat and current_node.confirmed: diagnosis = "Wind-heat invading the lung" treatment = ["dispelling wind, clearing heat","promoting lung function and resolving phlegm"] # List of treatment principles advice = "Take Yin Qiao San and other heat-clearing medicines, drink more water and rest" # Determine whether to predict the next stage if patient.feverHigh or patient.symptoms.contains("worsening sore throat"): next_node = lungHeat # Take preemptive measures to avoid entering lung heat congestion treatment.append("clearing lung fire") advice += ", and take lung-clearing medicine if necessary to prevent worsening of the condition"
The above pseudocode demonstrates how AI uses semantic nodes to perform conditional checks and trigger actions: when the conditions of the “wind-heat invading the lung” knowledge node are met in the patient, AI confirms the node as valid, and the information attached to the node (diagnosis name, treatment principles, and suggestions, etc.) is extracted as the basis for decision-making. At the same time, AI checks the severity of symptoms and follows the links in the knowledge graph to predict possible developments (
), thereby taking timely intervention measures. The entire logic essentially originates from the semantic definitions and their interrelationships in the knowledge graph. Therefore, knowledge representation (expression) and reasoning execution are closely coupled here: there is no explicit call to an external reasoning engine, and the graph itself carries the reasoning process.
It is worth noting that in the DIKWP model, each level of semantic processing has corresponding mathematical support. For example, the knowledge level often uses formal logic and graph theory to represent concepts and their relationships; the wisdom level may apply decision trees or evaluation functions to select the best plan; the purpose level ensures that actions are aligned with the ultimate goal through utility functions or objective functions. Such strict formalization ensures the accuracy and consistency of semantic reasoning. For our case, since a structured TCM knowledge graph and logical rules are used, AI’s reasoning of the condition is similar to a deductive proof process, with each step being justified. This process can be seen as performing “semantic mathematics” operations, and also as AI engaging in “conscious” thinking. In other words, the DIKWP semantic model blurs the boundary between knowledge representation and reasoning process, making “expression as reasoning” possible.
Through the above demonstration, it can be seen that DIKWP semantic expressions directly drive AI’s reasoning process: TCM concepts are formalized into executable nodes and rules, forming a causal chain graph; AI performs computational deduction along the graph, just like the human brain thinking along a logical chain. This mechanism greatly increases the transparency and reliability of AI reasoning because we can review and understand each step of AI’s decision-making at the semantic level. For readers in the education field (such as middle school students), it also presents an intuitive picture of how AI “thinks”: AI does not draw conclusions out of thin air but reasons step by step along the knowledge network, just like writing detailed steps when solving a problem. This not only helps to understand the basis of AI diagnosis but also demonstrates an interpretable paradigm of the internal working principle of artificial intelligence.
5Artificial Consciousness “Learning Doctor” Dialogue Simulation
To make the above mechanism more vivid, this section simulates the real-time operation of an artificial consciousness system in a diagnostic scenario through a dialogue between an AI doctor and a patient. We will use the cycle of semantic reception - intention generation - cognitive construction - vernacular output - verification and completion as a clue to analyze the processes occurring internally in AI round by round. This approach helps readers intuitively feel how the DIKWP model enables AI to exhibit “intelligent behavior” similar to that of human doctors.
Scenario setting: The patient is a middle school student who has recently caught a cold, with initially mild symptoms that have gradually worsened. The AI doctor communicates with the patient through a chat interface, providing TCM diagnostic and treatment suggestions. The dialogue language is mainly in vernacular Chinese to ensure that middle school students can easily understand. We will note the internal semantic processing of AI in parentheses after each segment of the dialogue.
Patient: Doctor, I’ve had a cold for several days now. My throat is very sore, I keep coughing, and I have a bit of a fever. What should I do? (AI semantic reception: Identifies key information such as “cold for several days,” “sore throat (pharyngitis),” “cough,” and “fever”; initially extracts the disease course and symptom data)
AI Doctor: Is your sore throat severe? Have you coughed up any phlegm? If so, what color is it? (AI intention generation: Since the patient mentioned sore throat, cough, and fever, AI associates it with possible syndromes such as wind-cold common cold or wind-heat invading the lung, but needs to distinguish between them, so it generates the intention to ask about the color of phlegm and the degree of pain; cognitive construction: Based on the knowledge of wind-cold/wind-heat syndrome differentiation, designs questions to obtain the key information of phlegm nature; vernacular output: Transforms the intention into a question sentence that the patient can easily understand)
Patient: There’s not much phlegm, but occasionally I cough up a bit of yellow. My throat is particularly sore, and it feels swollen. (AI semantic reception: Captures new information such as “yellow phlegm” and “obvious sore throat and swelling”; updates the patient’s symptom information and matches it with existing knowledge)
AI Doctor: It seems that your cold has worsened somewhat and now involves wind-heat invading the throat and lung. To put it simply, it’s an inflammatory condition of the wind-heat type that has caused your sore throat, cough, and fever. In this case, we need to clear heat and detoxify, as well as disperse wind and promote lung function. Have you had a dry throat recently, with little relief from drinking water? (AI cognitive construction: Confirms the syndrome of “wind-heat invading the lung” based on the new information, and the activated knowledge node outputs diagnostic judgment and treatment suggestions; at the same time, AI speculates that the patient may have other accompanying symptoms such as dry throat and decides to ask again to verify the completeness of symptoms; vernacular output: First explains the diagnostic result, describes the meaning of “wind-heat invading the lung” in popular language, and then takes the opportunity to ask about other symptoms to verify and complete)
Patient: Yes, my throat has been particularly dry these past two days, and I always feel very thirsty. (AI semantic reception: Obtains the symptom of “dry throat and thirst,” which matches the wind-heat syndrome; further confirms the diagnosis)
AI Doctor: Well, your symptoms basically fit what we call the “wind-heat invading the lung” syndrome. This means that exogenous wind-heat pathogenic factors have invaded the lung, causing the current symptoms of sore throat, cough, and fever. If we don’t clear the lung heat in time, it may spread downward to the trachea, turning into bronchitis, which is a more severe condition of lung heat congestion. So, we need to reduce fever and clear heat as soon as possible. I suggest you take traditional Chinese medicine like Yin Qiao San to clear heat and detoxify, and reduce inflammation and fever. At the same time, drink more warm water, get plenty of rest, and follow a light diet, which will help with recovery. (AI wisdom decision-making and output: Integrating all the information, AI forms a complete treatment plan in the wisdom layer, including warnings about possible developments and current treatment suggestions; the intention layer ensures that the explanation and suggestions are easy to accept and in the patient’s best interest. The final vernacular output is divided into three parts: summarizing the etiology and pathogenesis with TCM concepts and explaining their meanings again; reminding the patient of the possible worsening of the condition to enhance their attention; and giving specific feasible suggestions, including the name of the traditional Chinese medicine and lifestyle adjustment measures. The entire response is professional yet not abstruse, with educational value)
Patient: Okay, thank you, Doctor! I understand, and I will follow your advice on taking medicine and resting. (AI semantic reception: Receives the patient’s feedback of “understand,” infers that the patient has understood and accepted the advice; the task is essentially completed)
AI Doctor: You’re welcome. If there is no improvement in two or three days, or if more severe symptoms appear, such as persistent high fever or chest pain, be sure to let me know or go to the hospital for a check-up. I hope you recover soon! (AI intention generation: The final step of verification and completion before ending the dialogue, reminding the patient to follow up to verify the effectiveness of the treatment and supplement subsequent measures; vernacular output: Informs the patient in a caring tone when to seek further medical attention, demonstrating the AI’s human touch and sense of responsibility)
Through the above dialogue, it can be seen that the behavior pattern of the AI doctor is very similar to that of a human doctor: it continuously adjusts its judgment based on the patient’s narrative, asks targeted questions, and ultimately provides explanations and suggestions. This similarity stems from the semantic closed-loop mechanism endowed to AI by the DIKWP model. We can correspond to each of the previously mentioned steps:
·Semantic Reception: Each time the patient speaks, AI quickly extracts key points of symptoms and medical history, transforming natural language into standardized semantic representations (such as a list of symptoms and their attributes).
·Intention Generation: Based on the current semantic state, AI decides what to do next. For example, when information is insufficient, the intention is to “ask more symptoms”; when the diagnosis is clear, the intention is to “explain and suggest treatment”; and when concluding, the intention is to “inform follow-up observation matters.”
·Cognitive Construction: Before executing the intention, AI uses knowledge and reasoning to build specific content. For example, when designing questions, it selects the most effective ones; when explaining the condition, it draws on relevant knowledge (including the correspondence between TCM concepts and Western medical terms); and when making suggestions, it considers whether the patient can understand and follow them. This step involves calling upon the knowledge stored in the DIKWP model and integrating it, reflecting AI’s real-time “conscious” activity.
·Vernacular Output: AI transforms its internal semantic content into natural language, striving to be professional yet accessible. This demonstrates the DIKWP model’s ability for bidirectional mapping - AI can not only turn human speech into its own semantic representation but also convert its reasoning results into language that people can understand. The AI’s multiple uses of plain language to explain medical terms in the dialogue are a testament to this capability.
·Verification and Completion: AI always pays attention to whether the dialogue has achieved the intended purpose. If the patient’s response does not match expectations, AI reassesses; even after the patient indicates understanding and acceptance of the advice, AI still performs a final check (informing follow-up conditions) to consolidate the diagnostic and treatment effect. This verification and supplementation at the end of the dialogue reflect AI’s sense of responsibility and “self-reflection” ability - just as human doctors instruct patients on precautions, AI ensures that its diagnosis and suggestions are verified and followed up in the coming period. This process is similar to the metacognitive cycle in the DIKWP model, where the entire process is reviewed and completed after the main task is finished to ensure the thorough completion of the closed loop.
Through this dialogue simulation, we can intuitively see how the DIKWP artificial consciousness model enables AI to work like a “learning doctor”: the learning aspect is reflected in its ability to learn new information about the patient during interactions and continuously update its cognition; the doctor aspect is reflected in its use of medical knowledge and reasoning capabilities to make professional judgments and communicate in a humanized manner. Throughout the process, “expression” and “execution” are unified: AI’s understanding of medical concepts directly guides its reasoning execution, and the results of reasoning are communicated back to humans through language, forming a perfect closed loop. This unified mechanism makes AI’s diagnostic process transparent and verifiable - each step is justified and traceable. For readers such as middle school students, they can not only learn medical knowledge from the dialogue but also understand AI’s internal thinking logic, greatly enhancing their trust and acceptance of AI.
6Conclusion
This paper focuses on the case of “common cold-pharyngitis-tracheitis” and provides a detailed demonstration of the full-process application of the DIKWP semantic model in TCM AI diagnosis simulation. At the conceptual level, we have explained how semantic executability transforms abstract TCM terms into AI-reasonable units, achieving a leap from descriptive knowledge to operational knowledge. At the process level, we have shown how AI, through the DIKWP model, maps natural language input into a semantic action chain and iterates continuously in a closed loop of understanding-reasoning-verification until a reliable diagnosis and treatment plan is reached. Through causal chain graphs and pseudocode examples, it can be seen that semantic expressions themselves drive the logical calculations within AI, truly achieving “expression as reasoning, expression as execution.” The final dialogue simulation integrates all of this into a specific context, allowing us to glimpse how an AI doctor driven by artificial consciousness works: AI, through the DIKWP model, achieves a deep integration of human language and medical knowledge, demonstrating reasoning and communication abilities similar to those of humans.
The DIKWP model serves as a bridge, establishing a two-way connection between language understanding and model reasoning. It introduces a clear cognitive hierarchy and mathematically defined semantics into AI systems, ensuring that each step of AI’s decision-making is traceable. In our case, this means that AI can clearly explain “why ask this question,” “why draw this conclusion,” and “why make this suggestion.” Such explainability is crucial for building user trust in AI doctors. As reported, the DIKWP model makes AI systems more transparent, explainable, and aligned with human values and knowledge systems. In our simulation, we also saw that the suggestions given by AI were no different from those of human doctors and could explain the reasons behind them - this is precisely because the DIKWP semantic closed loop ensures cognitive consistency and correctness.
Moreover, the “expression and execution unity” of the DIKWP model provides a general paradigm that is not only applicable to TCM diagnosis but also to other fields where AI needs to perform complex reasoning, such as engineering fault diagnosis, legal consultation, and educational Q&A. When AI can “think” in human semantics and make its thinking process understandable to humans, artificial intelligence will no longer be a cold and mysterious black box but will become an intelligent entity capable of equal communication. This marks an important step for AI towards the direction of white-box and artificial consciousness.
In summary, through this special report, we have clearly demonstrated how DIKWP semantic content becomes the cornerstone of a computable, verifiable, and cognitively credible mechanism in AI systems. The full-process demonstration with “expression as execution” as the clue shows that when AI has a rigorous semantic model like DIKWP, it can better understand our language, reason and act according to established knowledge and purposes, and provide feedback for verification. In the near future, with the further development of semantic artificial consciousness models such as DIKWP, we have reason to expect that AI in more fields will exhibit performances similar to that of a “learning doctor” - possessing professional knowledge and interacting with people in a trustworthy manner to provide higher-level intelligent services for humanity.
7References
[1]Yucong Duan et al. Research Report on DIKWP Artificial Consciousness Model. February 2025.
[2]China Rongmei Industry Network. Professor Yucong Duan: The DIKWP Artificial Consciousness Model Leads the Future of AI, with 114 Patents Awaiting Industrial Implementation. Phoenix Regional News, March 29, 2025.
[3]Yu Zhang. What Does Wind-Heat Invading the Lung Mean. 39 Health Network, June 11, 2024.
[4]Huiping Su. Lung Heat Excess Syndrome (Also Known as Lung Heat Congestion Syndrome). China Medical Information Inquiry Platform, [Publication Date Unknown].
[5]Xin Liu. The Commercial Valuation of the DIKWP Artificial Consciousness Model is Based on 
Million, Leading AI Governance Out of the “Black Box” and into the “White Box” Era. Scientific Research Chitchat, November 23, 2024.
[6]Yucong Duan. Research on the Core Role of the DIKWP Model in Human-Machine Bidirectional Cognitive Language. International Artificial Intelligence DIKWP Assessment Standards Committee, April 2025.
[7]Yucong Duan, Shi Ming Gong. From Semantic Space to Concept Space: Breakthroughs and Applications of the DIKWP Model in Artificial Consciousness. Science Net, 2024.
[8]National Administration of Traditional Chinese Medicine Terminology Standardization and Promotion Project. Traditional Chinese Medicine Clinical Diagnostic and Therapeutic Terms (Revised Edition) Part 2: Syndromes. China Medical Information Inquiry Platform.
