Human-intelligent interaction-oriented cognitive process visualization method, system, equipment and medium

Abstract

The invention provides a cognitive process visualization method, system and device for human-intelligence interaction and a medium, and the method comprises the steps: obtaining a natural language interaction dialogue with artificial intelligence AI, and carrying out high-dimensional semantic analysis and cognitive relationship analysis, so as to convert the natural language interaction dialogue into a structured reasoning event object; according to the reasoning event object, determining a structured cognitive chain used for revealing the thinking path and the AI effect through multi-dimensional relation analysis and stage self-adaption rules; determining a cognitive input index and a creativity index according to a structured cognitive chain, quantized node semantics, a structure and an interaction mode; and according to the cognitive input index and the creativity index, performing automatic layout and visual coding in combination with a structured cognitive chain, and generating a data-driven multi-level visual interface. According to the invention, data-driven multi-level visualization is realized, so that a teacher can realize visualization, diagnosis and intervention of a learning process in an open learning task supported by AI.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method, system, device, and medium for visualizing cognitive processes in human-intelligent interaction. Background Technology

[0002] With groundbreaking advancements in generative artificial intelligence (AI) technology, its application in education has evolved from early intelligent question answering and knowledge retrieval to becoming a cognitive partner capable of participating in complex learning processes. In scenarios such as project-based learning (PBL) and open-ended tasks, dialogue between students and AI has become a crucial medium for advancing tasks, understanding information, generating ideas, and solving problems. This interactive model breaks through the limitations of traditional learning tools, providing students with personalized and immediate support through natural language interaction. However, the in-depth application of this technology also brings new challenges. These dialogues are inherently highly unstructured, semantically leaping, and highly flexible. The key cognitive behaviors generated by students during interaction, such as reasoning processes, shifts in perspectives, and decision-making logic, are all scattered in long sequences of context in natural language form, making in-depth analysis and effective intervention of the learning process a pressing problem that current technology needs to address.

[0003] Currently, the management of student-AI interactions is mainly achieved through linear text recording and simple task progress tracking by intelligent tutoring systems (ITS) or learning management systems (LMS). The backend database fully records the linear dialogue text between students and AI, as well as related metadata. Rule-based or shallow natural language processing (NLP) techniques are used to perform preliminary processing on the logs. For example, keyword matching is used to identify whether students mention specific concepts, sentiment analysis tools are used to determine the emotional tendency of the dialogue, or behavioral indicators such as the number of questions asked, dialogue rounds, and task completion status are statistically analyzed to summarize these analysis results into simple learning reports or data dashboards for teachers to review afterward. The core objective is to track students' learning progress and participation, rather than to analyze their internal cognitive processes.

[0004] However, linear text recording completely loses the cognitive structure relationships in the dialogue, failing to present the deduction, comparison, integration, and correction between students' different ideas, and also making it difficult to reflect the activity curves of students at each stage of the task. Teachers find it difficult to accurately judge students' true thinking process in the task by reviewing the dialogue content afterward, and cannot reveal the position of the AI's response in the student's decision chain. In particular, it is difficult to identify whether students mechanically replace the AI's suggestions with their own thinking, or whether they have shown meaningful cognitive processing. This makes AI-supported learning activities significantly limited in terms of teaching controllability, learning quality assurance, and process evaluation. If teachers attempt to manually analyze this content, it requires a lot of time and is easy to overlook key nodes due to information redundancy, making it difficult to grasp students' true thinking paths.

[0005] In addition, analysis methods based on keywords and simple statistics are too superficial and cannot quantify high-dimensional cognitive indicators, such as cognitive engagement depth (constructivism, interactivity), creativity (fluency, originality), or the completeness of reasoning chains. The evaluation results are subjective and lagging.

[0006] Furthermore, since different students advance tasks at different stages, with different dialogue lengths and different paths, the aforementioned technologies can only provide scattered individual data or superficial macro progress, failing to achieve multi-level deep cognitive behavior visualization and making it difficult to dynamically locate learning risk nodes. This leaves teachers struggling to manage large-scale, asynchronous AI learning activities, severely limiting both teaching controllability and learning quality assurance. Summary of the Invention

[0007] This invention provides a method, system, device, and medium for visualizing cognitive processes in human-intelligence interaction, which addresses the shortcomings of existing linear text recording methods that cannot reveal students' true thinking paths in tasks. It achieves data-driven, multi-level visualization, enabling teachers to visualize, diagnose, and intervene in AI-supported open-ended learning tasks, thereby improving learning quality and teaching controllability.

[0008] This invention provides a method for visualizing cognitive processes in human-AI interaction, comprising: acquiring natural language interactive dialogues with artificial intelligence (AI), and performing high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogues into structured reasoning event objects; determining, based on the reasoning event objects, a structured cognitive chain for revealing thought paths and AI roles through multi-dimensional relationship analysis and stage adaptive rules; quantifying node semantics, structure, and interaction patterns based on the structured cognitive chain to determine cognitive input indicators and creativity indicators; and automatically laying out and visually encoding, based on the cognitive input indicators and creativity indicators, combined with the structured cognitive chain, to generate a data-driven multi-level visualization interface.

[0009] According to the present invention, a method for visualizing cognitive processes in human-AI interaction is provided. This method acquires natural language interactive dialogues with artificial intelligence (AI) and performs high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogues into structured reasoning event objects. The method includes: acquiring natural language interactive dialogues with AI and performing syntactic structure decomposition based on an attention mechanism to obtain semantic question fragment features corresponding to the student's expression and semantic response content features corresponding to the AI's response; classifying the semantic question fragment features corresponding to the student's expression and identifying the functional types of the semantic response content features corresponding to the AI's response to obtain corresponding semantic fragment categories and corresponding functional types; and classifying the semantic question fragment features corresponding to the student's expression based on their semantic question fragment features. The semantic response content features corresponding to the AI ​​response are analyzed for similarity matching, citation pattern recognition, and logical dependency relationship to determine the AI ​​impact factor. The AI ​​impact factor characterizes the degree of influence of the AI ​​response on students. Based on the semantic question fragment features, semantic fragment category, semantic response content features, and functional type corresponding to the AI ​​response, task context prediction is performed to obtain the task prediction stage. Based on the semantic question fragment features, semantic fragment category, semantic response content features, functional type, role information, timestamp, and task prediction stage corresponding to the student's expression, the corresponding inference event object is obtained, and its contextual dependency relationship with other prior inference event objects is determined, thus updating the inference event object.

[0010] According to the present invention, a method for visualizing cognitive processes for human-computer interaction is provided. The reasoning event object includes semantic question fragment features corresponding to the student's expression, semantic fragment categories, semantic response content features corresponding to the AI's response, function type, role information, timestamp, and task prediction stage. Based on the reasoning event object, a structured cognitive chain for revealing the thinking path and AI's role is determined through multi-dimensional relationship analysis and stage adaptive rules. This includes: determining the semantic similarity between the reasoning event object and other prior reasoning event objects; identifying causal relationships between the reasoning event object and other prior reasoning event objects; and determining the semantic fragment categories and task prediction stages of the reasoning event object. The system predicts context and judges topic consistency; it assesses the coherence of reasoning actions based on the semantic fragment categories and AI influence factors in the reasoning event objects; it selects corresponding reasoning rules based on the task prediction stage corresponding to the reasoning event objects, and determines the structured relationships between each reasoning event object by combining semantic similarity, causal relationship, subject consistency, and coherence of reasoning actions, forming a structured cognitive chain with directionality and hierarchy; each node in the structured cognitive chain represents the logical state of the student or AI in a corresponding round of interaction, and the connection between nodes is used to represent the reasoning transfer path from divergence to convergence, from solution exploration to solution selection, and from explanation to reflection.

[0011] According to the present invention, a method for visualizing cognitive processes for human-intelligence interaction, after determining the structured cognitive chain for revealing the thinking path and the role of AI based on the reasoning event object through multi-dimensional relationship analysis and stage adaptive rules, includes: analyzing text density, explanation depth, node expansion speed, content repetition, semantic vector similarity and chain topology changes based on the topological structure and semantic features of the structured cognitive chain, identifying learning risks, marking risk nodes and recording risk types.

[0012] According to the present invention, a method for visualizing cognitive processes for human-intelligence interaction, after determining the structured cognitive chain for revealing the thinking path and the role of AI based on the reasoning event object through multi-dimensional relationship analysis and stage adaptive rules, includes: performing semantic behavior feature and reasoning action analysis based on the structured cognitive chain to determine whether the student is in the divergent, convergent, planning, execution or reflection stage, and superimposing stage labels onto the structured cognitive chain to make the stage judgment, reasoning structure and indicator calculation mutually related.

[0013] According to the present invention, a method for visualizing cognitive processes for human-intelligence interaction quantifies the semantics, structure, and interaction patterns of nodes based on a structured cognitive chain, and determines cognitive input and creativity indicators. This includes: performing semantic depth analysis on each node based on the semantic question fragment characteristics and semantic response content characteristics in the structured cognitive chain to determine the semantic complexity of the corresponding node content; evaluating the reasoning action type based on the semantic fragment categories in the structured cognitive chain to obtain the action reasoning type; and judging the interaction structure based on the connection relationship between student nodes and AI nodes in the structured cognitive chain and the AI ​​influence factor, and determining the cognitive input indicators by combining the semantic complexity of the node content and the action reasoning type. Based on the structured cognitive chain, the number of nodes generating student ideas is determined through semantic clustering, semantic span analysis, and chain structure feature identification. Semantic clustering is performed on all idea nodes to obtain clustering results and determine category diversity. Semantic span analysis is conducted based on the structured cognitive chain to determine the semantic distance between each student idea node and the corresponding viewpoint in the AI's response, thus determining originality. Based on the chain's branching structure, the degree of chain divergence is determined, and the extension elaboration depth is determined through the logical chain length and reasoning depth of the node explanations. Creativity indicators are determined based on the number of nodes generating student ideas, category diversity, originality, and elaboration depth.

[0014] According to the present invention, a method for visualizing cognitive processes oriented towards human-intelligence interaction includes a multi-level visualization interface comprising a visual overview display and a student role-level display. Based on cognitive input indicators and creativity indicators, and combined with a structured cognitive chain, an automatic layout and visual encoding are performed to generate a data-driven multi-level visualization interface. This includes: integrating key indicators for corresponding student roles based on the structured cognitive chain, cognitive input indicators, and creativity indicators, and identifying students with learning risks; and rendering data based on the integrated key indicators for each student role and the students with learning risks to generate a visual overview display. The visualization display is presented in list or grid format. The system identifies all student roles within a class or a pre-defined range, using color coding, small icons, or mini-charts to visually represent each individual's task stage, engagement trend, and risk status. For any given student role, it generates an event sequence curve based on cognitive engagement and creativity indicators, following the chronological order of the dialogue. Alternatively, it overlays these indicators onto a structured cognitive chain. Based on the topology of the structured cognitive chain, it determines node positions, centering the main path in the view and unfolding branch paths hierarchically. These branches are then visually encoded to determine node colors, transparency, line types, and markers based on node attributes, while also highlighting risk nodes, thus generating a student role-level display.

[0015] This invention also provides a cognitive process visualization device for human-AI interaction, comprising: an object conversion module, which acquires natural language interactive dialogue with artificial intelligence (AI) and performs high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogue into structured reasoning event objects; a knowledge chain generation module, which, based on the reasoning event objects, determines a structured cognitive chain for revealing the thinking path and the role of AI through multi-dimensional relationship analysis and stage adaptive rules; an indicator quantification module, which, based on the structured cognitive chain, quantifies the semantics, structure, and interaction patterns of nodes to determine cognitive input indicators and creativity indicators; and a visualization processing module, which, based on the cognitive input indicators and creativity indicators, combines the structured cognitive chain to automatically lay out and visually encode data to generate a data-driven multi-level visualization interface.

[0016] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the cognitive process visualization method for human-intelligence interaction as described above.

[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the cognitive process visualization method for human-intelligence interaction as described above.

[0018] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the cognitive process visualization method for human-intelligence interaction as described above.

[0019] This invention provides a method, system, device, and medium for visualizing cognitive processes in human-AI interaction. By performing high-dimensional semantic parsing and cognitive relationship analysis on acquired natural language dialogues with artificial intelligence (AI), the text is converted into high-dimensional semantic vectors. This captures the deeper meaning and cognitive relationship analysis, resulting in structured reasoning event objects. Abstract dialogue content is transformed into standardized data units that can be read, processed, and analyzed by machines, laying a solid data foundation for revealing thought processes and AI effects. Furthermore, through multi-dimensional relationship analysis and stage-adaptive rules, isolated event objects are connected, establishing a structured cognitive chain for revealing thought processes and AI effects. This allows the system to move beyond analyzing fragmented dialogue points and instead understand the entire process. The flow and logical structure of thought clearly demonstrate how students' thinking evolves step by step and the role AI plays in it. This allows for the quantification of the structured cognitive chain, identifying cognitive input and creativity indicators. This shifts the assessment of the thinking process from qualitative description to quantitative analysis, providing objective and comparable data for measuring the quality and depth of students' thinking. This makes the assessment more accurate and scientific. Furthermore, the cognitive input and creativity indicators are automatically laid out and visually encoded in conjunction with the structured cognitive chain, generating a data-driven, multi-level visualization interface. This makes complex analysis results intuitive and easy to understand, significantly lowering the user's comprehension threshold. Users can quickly grasp the overall picture and details of the learning process, enabling teachers to visualize, diagnose, and intervene in AI-supported open-ended learning tasks. This improves learning quality and teaching controllability, providing clear insights for decision-making. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1 This is one of the flowcharts of the cognitive process visualization method for human-intelligence interaction provided by the present invention; Figure 2 This is the second flowchart of the cognitive process visualization method for human-intelligence interaction provided by the present invention; Figure 3This is a schematic diagram of the structure of the cognitive process visualization device for human-intelligence interaction provided by the present invention; Figure 4 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] Figure 1 This is a flowchart illustrating the cognitive process visualization method for human-intelligence interaction provided by the present invention, as shown below. Figure 1 As shown, the method includes: S11: Acquire natural language interactive dialogue with artificial intelligence (AI) and perform high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogue into structured reasoning event objects. S12, Based on the object of the reasoning event, through multi-dimensional relationship analysis and stage adaptive rules, determine the structured cognitive chain used to reveal the thinking path and the role of AI; S13. Based on the structured cognitive chain, quantify the semantics, structure and interaction patterns of nodes, and determine cognitive input indicators and creativity indicators. S14, based on cognitive input indicators and creativity indicators, combines structured cognitive chains to automatically lay out and visually encode data, generating a data-driven multi-level visualization interface.

[0024] It should be noted that the step numbers "S1N" in this manual do not represent the sequential order of the methods for visualizing cognitive processes in human-computer interaction. The following will explain in detail... Figure 2 This invention describes a method for visualizing cognitive processes in human-computer interaction.

[0025] Step S11: Obtain the natural language interaction dialogue with the artificial intelligence (AI) and perform high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interaction dialogue into structured reasoning event objects.

[0026] In this embodiment, the natural language interactive dialogue with the artificial intelligence (AI) is acquired, and high-dimensional semantic parsing and cognitive relationship analysis are performed to transform the natural language interactive dialogue into structured reasoning event objects. This includes: acquiring the natural language interactive dialogue with the AI, and performing syntactic structure decomposition based on an attention mechanism to obtain semantic question fragment features corresponding to the student's expression and semantic response content features corresponding to the AI's response; classifying the semantic question fragment features corresponding to the student's expression, and identifying the functional type of the semantic response content features corresponding to the AI's response, to obtain the corresponding semantic fragment category and corresponding functional type; and then, based on the semantic question fragment features corresponding to the student's expression and the semantic response content features corresponding to the AI's response, performing semantic response content feature identification to obtain the corresponding semantic fragment category and corresponding functional type. The AI ​​impact factor is determined by analyzing the characteristics of the response content, performing similarity matching, citation pattern recognition, and logical dependency relationship analysis. This factor characterizes the degree of influence of the AI ​​response on students. Based on the semantic question fragment features, semantic fragment categories, and the semantic response content features and functional types corresponding to the student's expression, task context prediction is performed to obtain the task prediction stage. Finally, based on the semantic question fragment features, semantic fragment categories, semantic response content features, functional types, role information, timestamps, and task prediction stage, the corresponding inference event object is obtained, and its contextual dependency relationship with other prior inference event objects is determined, thus updating the inference event object.

[0027] It should be added that before acquiring the natural language interaction dialogue with the artificial intelligence (AI), the following steps are required: acquiring the raw interaction text, which includes the interaction content between the student and the AI ​​in the current dialogue round; performing basic cleaning and formatting on the raw interaction text, such as removing irrelevant format symbols, standardizing encoding, and ensuring that the speaker and timestamp information of each round of dialogue are accurate, in order to prepare for subsequent context-aware analysis.

[0028] In addition, the process involves acquiring natural language interactive dialogues with artificial intelligence (AI) and performing syntactic structure decomposition based on an attention mechanism to obtain semantic question fragment features corresponding to the student's expression and semantic response content features corresponding to the AI's response. This includes: acquiring natural language interactive dialogues with AI and extracting features to obtain question text features and response text features; based on the question text features and response text features, using an attention mechanism to capture long-distance semantic dependencies in the natural language interactive dialogue to obtain corresponding context vectors, and combining these with the corresponding question text features and response text features to obtain corresponding semantic question fragment features and semantic response content features.

[0029] Furthermore, semantic fragment categories include concepts, facts, inferences, evaluations, comparisons, hypotheses, and reflections. By classifying the features of semantic question fragments, the content scattered across different rounds can be reconnected in the semantic space, thus forming a consistent and continuous semantic trajectory. Additionally, the functional types of semantic response content features are identified, including explanatory supplements, alternative ideas, detailed suggestions, or solution examples, to distinguish the role of student-generated reasoning from AI-generated content.

[0030] Furthermore, it is necessary to calculate the AI ​​impact factor based on the degree of similarity between the student's expression and the AI ​​content, the citation pattern, and the logical dependency relationship, so as to determine whether a certain statement belongs to self-generated, AI-based elaboration, AI-dependent restatement, or AI-led thought transfer, thereby distinguishing the student's actual thinking contribution from the reasoning path triggered by AI.

[0031] Specifically, based on the semantic question fragment features corresponding to the student's expression and the semantic response content features corresponding to the AI's response, similarity matching, citation pattern recognition, and logical dependency analysis are performed to determine the AI ​​influence factor. This includes: performing similarity matching based on the semantic question fragment features corresponding to the student's expression and the semantic response content features corresponding to the AI's response to obtain semantic similarity; using natural language processing technology to identify explicit citation signals based on the semantic question fragments and performing logical dependency analysis to determine whether the viewpoints of the semantic question fragment features are extensions, rebuttals, or complete topic shifts based on the AI ​​viewpoints; and using a preset algorithm or model based on semantic similarity, citation signals, and logical dependencies to determine the AI ​​influence factor of the corresponding semantic question fragment features, so as to determine whether the statement belongs to self-generated, AI-based exposition, AI-dependent restatement, or AI-driven thought transfer based on the AI ​​influence factor.

[0032] It should be added that the task prediction stage includes information exploration, divergent thinking, solution convergence, and planning implementation. By analyzing the topics that appear in the dialogue and the distribution of cognitive behavior types such as exploration, planning, and reflection, the current task stage of the dialogue can be predicted, thus providing a macro background for understanding student behavior.

[0033] In addition, by standardizing semantic question fragment features, semantic fragment categories, semantic response content features, function types, role information, timestamps, and task prediction stages into a unified template for reasoning event objects, natural language text of any length can be represented in a structured manner, laying the foundation for subsequent chain construction.

[0034] Step S12: Based on the reasoning event object, determine the structured cognitive chain used to reveal the thinking path and the role of AI through multi-dimensional relationship analysis and stage adaptive rules.

[0035] In this embodiment, the inference event object includes the semantic question fragment features corresponding to the student's expression, the semantic fragment category, the semantic response content features corresponding to the AI's response, the function type, role information, timestamp, and task prediction stage. Based on the inference event object, a structured cognitive chain for revealing the thinking path and the AI's role is determined through multi-dimensional relationship analysis and stage adaptive rules. This includes: determining the semantic similarity between the inference event object and other prior inference event objects; identifying causal relationships based on the inference event object and other prior inference event objects; and predicting the theme based on the semantic fragment category and task context of the inference event object. Consistency assessment; evaluation of reasoning action coherence based on semantic fragment categories and AI influence factors in the reasoning event objects; selection of corresponding reasoning rules based on the task prediction stage corresponding to the reasoning event objects, and determination of the structured relationships between each reasoning event object by combining semantic similarity, causal relationship, subject consistency and reasoning action coherence, forming a structured cognitive chain with directionality and hierarchy; where each node in the structured cognitive chain represents the logical state of the student or AI in a corresponding round of interaction, and the connection between nodes is used to represent the reasoning transfer path from divergence to convergence, from solution exploration to solution selection, and from interpretation to reflection.

[0036] It should be added that structured cognitive chains can be constructed using graph-based reasoning models to transform linear dialogues into a visualized reasoning process. Specifically, by analyzing the semantic similarity, causal relationships, thematic consistency, and coherence of reasoning actions between events, the structured relationships between nodes can be inferred, forming a reasoning network with directionality and hierarchy.

[0037] Furthermore, in this structured cognitive chain, each node represents the logical state exhibited by the student or AI in a particular round of expression, while the connections between nodes are used to present the reasoning transfer path from divergence to convergence, from solution exploration to solution selection, and from explanation to reflection. Simultaneously, this structured cognitive chain not only demonstrates the progression of viewpoints but also identifies patterns such as leaps, backtracking, and rapid iterations in students' thinking, revealing how artificial intelligence plays a role in the chain—whether it provides initial stimuli, triggers conceptual changes, or serves as a source of evidence or evaluation criterion. This comprehensively reconstructs the direction of knowledge flow, the depth of reasoning, and the structural hierarchy of students in the task, revealing the process by which students form complex thought structures through supplementation, correction, branching, and integration, facilitating teacher review and intervention.

[0038] In addition, the inference rules are used to: emphasize the idea generation path and semantic span in the divergence phase; strengthen logical evaluation and solution selection clues in the convergence phase; and focus on metacognitive expression and corrective behavior in the reflection phase. The system thereby achieves adaptive processing of different types of data.

[0039] In an optional embodiment, after determining the structured cognitive chain used to reveal the thinking path and the role of AI based on the reasoning event object through multi-dimensional relationship analysis and stage adaptive rules, the process includes: analyzing text density, explanation depth, node expansion speed, content repetition, semantic vector similarity and chain topology changes based on the topological structure and semantic features of the structured cognitive chain, identifying learning risks, marking risk nodes and recording risk types.

[0040] It should be noted that, based on the structured cognitive chain, further analysis of text density, explanation depth, node expansion speed, content repetition, semantic vector similarity, and chain topological changes is used to determine whether students exhibit shallow responses, repetitive input, expression patterns without elaboration, or broken reasoning chains, in order to identify potential risk states in the learning process. When situations such as a sudden drop in expression length, no semantic expansion of several consecutive nodes, excessively close semantic distance to the AI ​​response, or no expansion of the chain path over a long period occur, risk nodes are automatically marked and risk types are recorded.

[0041] In addition, when identifying risks, if the text density or expression length drops sharply, students may lose interest or encounter difficulties; if the content repetition or the semantic distance with AI is too close, students may not have conducted in-depth analysis; if the node expansion speed or chain topology changes, students may experience mental stagnation or reasoning breakdown. When the above signals are triggered, risk nodes will be automatically marked and risk types will be recorded.

[0042] In another optional embodiment, after determining the structured cognitive chain for revealing the thinking path and the role of AI based on the reasoning event object through multi-dimensional relationship analysis and stage adaptive rules, the process includes: performing semantic behavior feature and reasoning action analysis based on the structured cognitive chain to determine whether the student is in the divergent, convergent, planning, execution or reflection stage, and superimposing stage labels onto the structured cognitive chain to make the stage judgment, reasoning structure and indicator calculation mutually related.

[0043] It should be noted that during the stage analysis process, by analyzing semantic behavioral features (such as idea generation, comparison, evaluation, reflection, etc.) and reasoning actions, it is determined whether the student is in the divergent, convergent, planning, execution, or reflection stage. Stage labels are then superimposed on the structured cognitive chain, linking stage judgment with reasoning structure and indicator calculation. These structured labels help identify whether students are performing mismatched cognitive operations at incorrect stages, such as premature evaluation in the divergent stage or lack of alternative comparison in the convergent stage, thus providing a basis for teaching intervention.

[0044] Furthermore, if idea generation is frequent, it is considered to be in the divergent stage; if comparison and evaluation are frequent, it is considered to be in the convergent stage.

[0045] Step S13: Based on the structured cognitive chain, quantify the semantics, structure and interaction patterns of the nodes, and determine the cognitive input indicators and creativity indicators.

[0046] Specifically, based on the structured cognitive chain, the semantics, structure, and interaction patterns of nodes are quantified to determine cognitive input and creativity indicators. This includes: conducting in-depth semantic analysis of each node based on the semantic question fragment characteristics and semantic response content characteristics in the structured cognitive chain to determine the semantic complexity of the corresponding node content; evaluating the reasoning action type based on the semantic fragment categories in the structured cognitive chain to obtain the action reasoning type; judging the interaction structure based on the connection relationship and AI influence factor between student nodes and AI nodes in the structured cognitive chain, and determining the cognitive input indicators by combining the semantic complexity of the node content and the action reasoning type; and based on the structured cognitive chain, through... Through semantic clustering, semantic span analysis, and chain structure feature recognition, the number of nodes generating student ideas is determined; semantic clustering is performed on all idea nodes to obtain clustering results and determine category diversity; based on the structured cognitive chain, semantic span analysis is performed to determine the semantic distance between each student idea node and the corresponding viewpoint in the AI's response, thus obtaining originality; based on the structured cognitive chain and the chain branch structure, the degree of chain divergence is determined, and the extension elaboration depth is determined through the logical chain length and reasoning depth explained by the nodes; based on the number of nodes generating student ideas, category diversity, originality, and elaboration depth, creativity indicators are determined.

[0047] It should be noted that, based on a structured cognitive chain, students' cognitive input and creative performance are quantified in multiple dimensions. Regarding cognitive input indicators, the level of cognitive processing is assessed by considering the semantic depth of the content at each node, the type of reasoning action, and the structure of the student's interaction with the AI. Changes in input during the learning process are presented in a time-series format, thereby identifying whether students have shifted from passive reception to active response, whether they have engaged in autonomous construction behavior, and whether they have entered a state of deep collaborative reasoning with the AI, thus revealing their learning initiative at different stages.

[0048] For creativity indicators, semantic clustering, semantic span analysis, and chain structure features are used to identify the quantity, category diversity, novelty, and elaboration depth of student-generated ideas. The uniqueness of different viewpoints is estimated by using distance in the semantic space, the degree of divergence is judged based on the branch structure of the chain, and the level of elaboration is judged by the length of the logical chain explained by the node and the depth of reasoning. The above indicators can be generated as independent curves or presented in a visual overlay in the chain structure, enabling teachers to observe students' creative thinking process from three dimensions: structure, time, and semantics.

[0049] In one optional embodiment, after determining the cognitive input index and creativity index by quantifying the semantics, structure and interaction patterns of nodes according to the structured cognitive chain, the method includes: generating an event sequence curve according to the time sequence of the dialogue based on the cognitive input index and creativity index, or overlaying the cognitive input index and creativity index onto the structured cognitive chain.

[0050] Step S14: Based on cognitive input indicators and creativity indicators, and combined with the structured cognitive chain, automatic layout and visual encoding are performed to generate a data-driven multi-level visualization interface.

[0051] In this embodiment, the multi-level visualization interface includes a visual overview display and a student role level display. Based on cognitive input and creativity indicators, and combined with a structured cognitive chain, automatic layout and visual encoding are performed to generate a data-driven multi-level visualization interface. This includes: integrating key indicators for corresponding student roles based on the structured cognitive chain, cognitive input, and creativity indicators, and identifying students with learning risks; rendering data based on the integrated key indicators for each student role and the students with learning risks to generate a visual overview display; wherein the visualization display shows the entire class or a preset range of students in a list or grid format. The system generates student roles and uses color coding, small icons, or mini-charts to visually represent each person's task stage, engagement trend, and risk status. For any student role, it generates an event sequence curve based on cognitive engagement and creativity indicators, following the chronological order of the dialogue. Alternatively, it overlays cognitive engagement and creativity indicators onto a structured cognitive chain. Based on the topology of the structured cognitive chain, it determines the node positions, placing the main path at the center of the view and branching paths hierarchically. These branches are then visually encoded to determine node color, transparency, line type, and markers based on node attributes, highlighting risk nodes and generating a student role level display.

[0052] It should be noted that, in order to enable the complex dialogue structure and indicator system to be quickly understood in the teaching scenario, a multi-level visualization presentation system was constructed. In the visualization overview display, each student's task stage, cognitive chain length, activity level, input trend, creativity performance, and risk distribution are integrated into an overview panel, enabling teachers to quickly identify students who are stagnant, repeatedly rely on AI, lack reasoning depth, or are abnormally active. In the student role-level display, the reasoning chain is presented in a graph structure, and the semantic content, source of reasoning, node attributes, AI influence factors, and related indicators of each node are visually superimposed, allowing teachers to sequentially view the student's thinking process from divergent to convergent.

[0053] Specifically, based on the topology of the structured cognitive chain, node positions are determined, ensuring the main path is centered in the view and branch paths unfold hierarchically. Visual encoding is then performed to determine node color, transparency, line type, and markers based on node attributes, highlighting risk nodes and generating student role-level displays. This includes: using graph layout algorithms, such as force-directed layout and hierarchical layout, to analyze the topology of the structured cognitive chain to calculate the optimal position of each node, ensuring the main path is centered in the view and branching viewpoints unfold hierarchically around the main path, forming a mind map; based on the attributes of the nodes and edges of the structured cognitive chain, visual encoding rules are used to automatically color nodes according to AI impact factors, adjust node size according to Elaboration depth, select edge type according to logical relationship type, and adjust node transparency according to time proximity or task stage; based on risk status, corresponding nodes are highlighted using special colors, flashing, or magnification effects; information on risk nodes is synchronously summarized in the risk statistics area, generating corresponding indicator curves, and dynamically linked to nodes in the structured cognitive chain; clicking on a risk node will highlight the relevant curve area.

[0054] Furthermore, based on the overall status of the current task and the specific characteristics of the student data, the system automatically determines the combination layout and presentation of the structured cognitive chain, indicator curves, and risk list. For example, if the system detects that a student is in a divergent stage, it may place the indicator curve representing flexibility in a more prominent position; if multiple risk nodes are detected, it may automatically enlarge the risk list area. The entire presentation process is driven entirely by backend logic, achieving a seamless and automated transformation from raw data to final insights without relying on any manual screening or intervention.

[0055] It is evident that the above mechanism enables real-time modeling of the learning process from multiple dimensions, constructing a complete, coherent, and interpretable representation of the learning process from five aspects: semantic depth, behavioral structure, reasoning path, stage transition, and risk state. This transforms the thought processes in project-based learning and open-ended tasks from invisible to visible, and from difficult to assess to diagnosable, providing a solid technical foundation for teachers to implement process-oriented and personalized guidance in large-scale scenarios.

[0056] In summary, this invention, through high-dimensional semantic parsing and cognitive relationship analysis of the acquired natural language interactive dialogue with artificial intelligence (AI), transforms the text into high-dimensional semantic vectors, captures its deeper meaning and cognitive relationship analysis, and obtains structured reasoning event objects. This transforms abstract dialogue content into standardized data units that can be read, processed, and analyzed by machines, laying a solid data foundation for revealing thought processes and AI effects. Furthermore, through multi-dimensional relationship analysis and stage-adaptive rules, isolated event objects are connected, establishing a structured cognitive chain for revealing thought processes and AI effects. This allows the system to move beyond analyzing scattered dialogue points and instead understand the entire flow and logic of thought processes. The framework clearly demonstrates how students' thinking evolves step by step and the role AI plays in it. It then quantifies the structured cognitive chain, identifying cognitive input and creativity indicators. This shifts the assessment of the thinking process from qualitative description to quantitative analysis, providing objective and comparable data for measuring the quality and depth of students' thinking. This makes the assessment more accurate and scientific. Furthermore, the cognitive input and creativity indicators are automatically laid out and visually encoded in conjunction with the structured cognitive chain, generating a data-driven, multi-level visualization interface. This makes complex analysis results intuitive and easy to understand, significantly lowering the barrier to understanding for teachers. Teachers can quickly grasp the overall picture and details of the learning process, enabling them to visualize, diagnose, and intervene in AI-supported open-ended learning tasks. This improves learning quality and teaching controllability, providing clear insights for decision-making.

[0057] The following describes the cognitive process visualization device for human-intelligent interaction provided by the present invention. The cognitive process visualization device for human-intelligent interaction described below can be referred to in correspondence with the cognitive process visualization method for human-intelligent interaction described above.

[0058] Figure 3 A schematic diagram of a cognitive process visualization device for human-computer interaction is shown. The device includes: The object conversion module 31 acquires the natural language interactive dialogue with artificial intelligence (AI) and performs high-dimensional semantic parsing and cognitive relationship analysis to convert the natural language interactive dialogue into structured reasoning event objects. The knowledge chain generation module 32 determines a structured cognitive chain for revealing the thinking path and the role of AI based on the reasoning event object through multi-dimensional relationship analysis and stage adaptive rules. The indicator quantification module 33 quantifies the semantics, structure and interaction patterns of nodes based on the structured cognitive chain, and determines cognitive input indicators and creativity indicators. The visualization processing module 34 automatically lays out and visually encodes data-driven multi-level visualization interfaces based on cognitive input indicators and creativity indicators, combined with structured cognitive chains.

[0059] It should be noted that the specific principles of the embodiments of the present invention are the same as those of the method embodiments described above. For details, please refer to the method embodiments above. More detailed explanations will not be repeated here.

[0060] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include a processor 410, a communications interface 420, a memory 430, and a communication bus 440. The processor 410, communications interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a cognitive process visualization method for human-computer interaction. This method includes: acquiring natural language interactive dialogue with artificial intelligence (AI), and performing high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogue into structured reasoning event objects; determining a structured cognitive chain for revealing thought paths and AI roles based on the reasoning event objects through multi-dimensional relationship analysis and stage adaptive rules; quantifying node semantics, structure, and interaction patterns based on the structured cognitive chain to determine cognitive input indicators and creativity indicators; and automatically laying out and visually encoding the structured cognitive chain based on the cognitive input indicators and creativity indicators to generate a data-driven multi-level visualization interface.

[0061] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0062] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the cognitive process visualization method for human-intelligence interaction provided by the above methods. This method includes: acquiring natural language interactive dialogue with artificial intelligence (AI) and performing high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogue into structured reasoning event objects; determining a structured cognitive chain for revealing the thinking path and the role of AI based on the reasoning event objects through multi-dimensional relationship analysis and stage adaptive rules; quantifying the semantics, structure, and interaction patterns of nodes based on the structured cognitive chain to determine cognitive input indicators and creativity indicators; and automatically laying out and visually encoding the structured cognitive chain based on the cognitive input indicators and creativity indicators to generate a data-driven multi-level visualization interface.

[0063] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements a method for visualizing cognitive processes for human-AI interaction provided by the methods described above. This method includes: acquiring natural language interactive dialogue with artificial intelligence (AI), and performing high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogue into structured reasoning event objects; determining a structured cognitive chain for revealing thought paths and AI roles based on the reasoning event objects through multi-dimensional relationship analysis and stage adaptive rules; quantifying node semantics, structure, and interaction patterns based on the structured cognitive chain to determine cognitive input indicators and creativity indicators; and automatically laying out and visually encoding the structured cognitive chain based on the cognitive input indicators and creativity indicators to generate a data-driven multi-level visualization interface.

[0064] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0065] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for visualizing cognitive processes in human-computer interaction, characterized in that, include: Acquire natural language interactive dialogues with artificial intelligence (AI), and perform high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogues into structured reasoning event objects; Based on the inference event object, a structured cognitive chain for revealing the thinking path and the role of AI is determined through multi-dimensional relationship analysis and stage adaptive rules; Based on the structured cognitive chain, the semantics, structure, and interaction patterns of nodes are quantified to determine cognitive input indicators and creativity indicators. Based on the cognitive input indicators and creativity indicators, and combined with the structured cognitive chain, automatic layout and visual encoding are performed to generate a data-driven, multi-level visualization interface.

2. The method for visualizing cognitive processes oriented towards human-computer interaction according to claim 1, characterized in that, Acquire natural language interactive dialogues with artificial intelligence (AI), and perform high-dimensional semantic parsing and cognitive relationship analysis to transform the natural language interactive dialogues into structured reasoning event objects, including: Acquire natural language interactive dialogues with artificial intelligence (AI), and perform syntactic structure decomposition based on attention mechanism to obtain semantic question fragment features corresponding to student expression and semantic response content features corresponding to AI response; The semantic question fragment features corresponding to the student's expression are classified, and the semantic response content features corresponding to the AI's response are functionally identified to obtain the corresponding semantic fragment category and corresponding functional type; Based on the semantic question fragment features corresponding to the student's expression and the semantic response content features corresponding to the AI's response, similarity matching, citation pattern recognition, and logical dependency analysis are performed to determine the AI ​​impact factor; wherein, the AI ​​impact factor is used to characterize the degree of influence of the AI ​​response on the student; Based on the semantic question fragment features corresponding to the student's expression, the semantic fragment category, the semantic response content features corresponding to the AI ​​response, and the function type, task context prediction is performed to obtain the task prediction stage. Based on the semantic question fragment features corresponding to the student's expression, the semantic fragment category, the semantic response content features corresponding to the AI ​​response, the function type, role information, timestamp, and task prediction stage, a corresponding inference event object is obtained, and the contextual dependency relationship with other prior inference event objects is determined, and the inference event object is updated.

3. The method for visualizing cognitive processes for human-computer interaction according to claim 1, characterized in that, The inference event objects include the semantic question fragment features corresponding to the student's expression, the semantic fragment category, the semantic response content features corresponding to the AI's response, the function type, role information, timestamp, and task prediction stage. Based on the inference event objects, a structured cognitive chain for revealing the thinking path and the role of AI is determined through multi-dimensional relationship analysis and stage adaptive rules, including: Based on the inference event object, determine the semantic similarity with other prior inference event objects; Based on the inference event object and other prior inference event objects, causal relationships are identified; Based on the semantic fragment category and task context prediction of the inference event object, a topic consistency judgment is made; Based on the semantic fragment categories and AI influence factors in the inference event object, the coherence of the inference action is evaluated; Based on the task prediction stage corresponding to the inference event object, the corresponding inference rule is selected, and the structured relationship between each inference event object is determined by combining semantic similarity, causal relationship, subject consistency and inference action coherence, forming a structured cognitive chain with directionality and hierarchy; wherein, each node in the structured cognitive chain represents the logical state of the student or AI in a corresponding round of interaction, and the connection between nodes is used to represent the reasoning transfer path from divergence to convergence, from solution exploration to solution selection, and from explanation to reflection.

4. The method for visualizing cognitive processes for human-computer interaction according to claim 3, characterized in that, After determining the structured cognitive chain used to reveal the thought process and the role of AI based on the inference event object through multi-dimensional relationship analysis and stage adaptive rules, the following are included: Based on the topological structure and semantic features of the structured cognitive chain, we analyze text density, interpretation depth, node expansion speed, content repetition, semantic vector similarity, and chain topological changes to identify learning risks, mark risk nodes, and record risk types.

5. The method for visualizing cognitive processes for human-computer interaction according to claim 3, characterized in that, After determining the structured cognitive chain used to reveal the thought process and the role of AI based on the inference event object through multi-dimensional relationship analysis and stage adaptive rules, the following are included: Based on the structured cognitive chain, semantic behavioral features and reasoning actions are analyzed to determine whether the student is in the divergent, convergent, planning, execution, or reflection stage. Stage labels are then superimposed on the structured cognitive chain to link stage judgment with reasoning structure and indicator calculation.

6. The method for visualizing cognitive processes for human-computer interaction according to claim 1, characterized in that, Based on the structured cognitive chain, the semantics, structure, and interaction patterns of nodes are quantified to determine cognitive input indicators and creativity indicators, including: Based on the semantic question fragment features and semantic response content features in the structured cognitive chain, semantic depth analysis is performed on each node to determine the semantic complexity of the corresponding node content; The action reasoning type is obtained by evaluating the semantic fragment categories in the structured cognitive chain. Based on the connection relationship between student nodes and AI nodes in the structured cognitive chain and the AI ​​influence factor, the interaction structure is judged, and the cognitive input index is determined by combining the semantic complexity of the node content and the action reasoning type. Based on the structured cognitive chain, the number of nodes that generate ideas for students is determined through semantic clustering, semantic span analysis, and chain structure feature identification. Semantic clustering is performed on all idea nodes to obtain clustering results and determine category diversity; Based on the structured cognitive chain, semantic span analysis is performed to determine the semantic distance between each student's thought node and the corresponding viewpoint in the AI's answer, thus obtaining originality; Based on the structured cognitive chain, the degree of chain divergence is determined according to the chain branch structure, and the extension elaboration depth is determined by the logical chain length and reasoning depth explained by the nodes. Creativity metrics are determined based on the number of nodes generating student ideas, the category diversity, the originality, and the elaboration depth.

7. The method for visualizing cognitive processes for human-computer interaction according to claim 1, characterized in that, The multi-level visualization interface includes a visual overview display and a student role level display; based on the cognitive input index and creativity index, and combined with the structured cognitive chain, automatic layout and visual encoding are performed to generate a data-driven multi-level visualization interface, including: Based on the structured cognitive chain, the cognitive input indicators, and the creativity indicators, the key indicators for the corresponding student roles are integrated, and students with learning risks are identified. Based on the key indicators of each student role after integration and the students with learning risks, data rendering is performed to generate a visual overview display. The visualization display shows the student roles of the whole class or a preset range in the form of a list or grid, and uses color coding, small icons or mini charts to intuitively present each person's task stage, input trend and risk status. For any student role, an event sequence curve is generated according to the cognitive input index and the creativity index, in the order of the dialogue time, or the cognitive input index and the creativity index are superimposed on the structured cognitive chain. Based on the topology of the structured cognitive chain, the node positions are determined so that the main path is located in the center of the view, the branch paths are expanded hierarchically, and visual encoding is performed to determine the node color, transparency, line type and markings according to the node attributes, as well as to highlight risk nodes and generate a student role level display.

8. A visualization device for cognitive processes oriented towards human-computer interaction, characterized in that, include: The object conversion module acquires the natural language interactive dialogue with artificial intelligence (AI) and performs high-dimensional semantic parsing and cognitive relationship analysis to convert the natural language interactive dialogue into structured reasoning event objects. The knowledge chain generation module, based on the reasoning event object, determines a structured cognitive chain for revealing the thinking path and the role of AI through multi-dimensional relationship analysis and stage adaptive rules. The indicator quantification module quantifies the semantics, structure, and interaction patterns of nodes based on the structured cognitive chain, and determines cognitive input indicators and creativity indicators. The visualization processing module automatically lays out and visually encodes the cognitive input indicators and creativity indicators in conjunction with the structured cognitive chain to generate a data-driven, multi-level visualization interface.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the cognitive process visualization method for human-computer interaction as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the cognitive process visualization method for human-intelligence interaction as described in any one of claims 1 to 7.

Images