Mental improvement algorithm based on human-computer interaction

Abstract

The invention discloses a mental improvement algorithm based on human-computer interaction, and the algorithm comprises the following steps: S1, collecting the behavior data, cognitive ability indexes and learning preferences of a user through a human-computer interaction interface, and constructing an initial cognitive feature vector containing the attention concentration degree and cognitive style classification; and S2, based on a cognitive challenge module, dynamically generating a multi-dimensional interaction task according to the current cognitive level of the user. According to the method, fusion of cross-domain task complexity and a neuroscience incentive mechanism is dynamically adjusted through the fuzzy logic controller, and accurate balance between cognitive challenge and individual adaptive learning is achieved. Compared with a traditional mental training algorithm, a semantic network analysis technology is introduced into cross-domain knowledge association task generation, and the problems of task staticization, excitation simplification and path linearization in a traditional algorithm are effectively solved in combination with a three-dimensional cognitive map constructed by a map attention network and a deep reinforcement learning optimization strategy.

Description

Technical Field

[0001] This invention relates to the field of algorithm technology, and in particular to a mental enhancement algorithm based on human-computer interaction. Background Technology

[0002] Traditional mental training systems typically employ a static task generation mechanism based on a rule base, where there is a linear correspondence between the cognitive challenge level and the neural activation strategy.

[0003] However, cross-domain knowledge association relies on a pre-defined two-dimensional task tree structure, which results in insufficient dynamic matching ability between task complexity and the user's real-time cognitive state. It is difficult to accurately adjust cognitive load based on EEG feedback or behavioral data. Moreover, existing neural incentive mechanisms mostly adopt fixed threshold reward models, which cannot adapt to the nonlinear cognitive fluctuations generated by individuals during cross-disciplinary knowledge transfer, resulting in training fatigue or idle cognitive resources.

[0004] Traditional 3D cognitive mapping relies excessively on supervised learning frameworks, leading to the solidification of semantic relationships between knowledge nodes and a lack of autonomous discovery of implicit logical relationships across domains. Although existing research has attempted to optimize task paths through reinforcement learning, its action space design is limited by single-dimensional knowledge progression relationships, failing to address the imbalance in the allocation of multimodal cognitive resources and easily generating cognitive conflicts when dealing with the fusion of unstructured knowledge. These shortcomings result in significant bottlenecks in the adaptive learning efficiency, neural activation effectiveness, and cross-domain knowledge integration dimensions of existing systems. To address these issues, we introduce a mind enhancement algorithm based on human-computer interaction. Summary of the Invention

[0005] This invention discloses a mind enhancement algorithm based on human-computer interaction, aiming to solve the technical problems in the background art.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A mindset enhancement algorithm based on human-computer interaction includes the following steps:

[0008] S1. Collect user behavior data, cognitive ability indicators and learning preferences through the human-computer interaction interface, and construct an initial cognitive feature vector including attention concentration and cognitive style classification.

[0009] S2. Based on the cognitive challenge module, multi-dimensional interactive tasks are dynamically generated according to the user's current cognitive level. The tasks include logical reasoning, abstract modeling, and cross-domain knowledge association tasks. The task difficulty coefficient is adjusted by a fuzzy logic controller to activate the user's higher-order thinking.

[0010] S3. Through the feedback and incentive module, the user's task completion rate and cognitive growth trajectory are analyzed in real time. Combined with the neuroscience-driven dopamine reward simulation mechanism, personalized incentive signals and cognitive diagnostic reports including cognitive load threshold and knowledge transfer efficiency are generated.

[0011] S4. Based on the personalized learning module, cognitive graph modeling technology is used to analyze the characteristics of users' thinking patterns, and an adaptive learning path that matches the user's cognitive development stage is generated through knowledge transfer reinforcement strategies.

[0012] S5. Quantify the improvement effect of user mindset through multi-dimensional evaluation indicators, including metacognitive ability, critical thinking index and knowledge integration efficiency, and iteratively optimize algorithm parameters.

[0013] In a preferred embodiment, in step S2:

[0014] The logical complexity of the multi-dimensional interactive task is divided into L1 to L5 levels, where L1 is a single-step reasoning task and L5 is a cross-domain knowledge association task, and the association threshold is ≥0.7.

[0015] The adjustment rule for the task difficulty coefficient is as follows:

[0016]

[0017] Where D is the difficulty level, P h To score performance on historical tasks, P c The current cognitive level is rated, and k is a moderating factor.

[0018] In a preferred embodiment, in step S2:

[0019] The cross-domain knowledge association task is achieved through semantic network analysis technology, which forces the association of knowledge points from at least two different knowledge domains, and the association path length does not exceed 3 layers of nodes.

[0020] The abstract modeling task includes at least one of mathematical formula derivation and physical principle visualization modeling.

[0021] In a preferred embodiment, in step S3:

[0022] The dopamine reward simulation mechanism generates phased incentive pulses based on the time point when the user breaks through the cognitive bottleneck. The time point is predicted by a hidden Markov model with a prediction error of ≤5%.

[0023] The cognitive diagnostic report includes 12 dimensions of indicators, among which field-independent / dependent cognitive style classification and cross-domain knowledge transfer efficiency are mandatory indicators.

[0024] In a preferred embodiment, in step S4:

[0025] The cognitive graph modeling technology constructs a three-dimensional graph through graph attention networks. The three-dimensional dimensions include knowledge domains, cognitive ability levels, and thinking pattern characteristics.

[0026] The knowledge transfer enhancement strategy is as follows: when a user completes a task in a specific domain, a cross-domain training task with a correlation degree ≥ 0.7 is forcibly inserted.

[0027] In a preferred embodiment, the node weights of the 3D graph are calculated using a convolutional neural network, and the node weight update formula is as follows:

[0028]

[0029] Where σ is the activation function, α ij h represents the node association attention coefficient. j Represent the hidden vectors for historical tasks.

[0030] In a preferred embodiment, in step S4:

[0031] The optimization of the adaptive learning path adopts a deep deterministic strategy gradient algorithm to dynamically adjust the path branches, and the path includes a knowledge forgetting curve compensation strategy.

[0032] The compensation strategy is as follows: when the user's memory strength for a specific knowledge node is lower than the threshold of 0.6, an interval repetition training task is automatically inserted.

[0033] In a preferred embodiment, in step S4:

[0034] M=λ1·R meta +λ2·R crit +λ3·R int

[0035] Among them, R meta R represents the improvement rate of metacognitive ability. crit R is the critical thinking index. int For knowledge integration efficiency, λ1, λ2, and λ3 are weighting coefficients and satisfy λ1+λ2+λ3=1.

[0036] The cognitive enhancement algorithm based on human-computer interaction provided by this invention has the following advantages:

[0037] This invention achieves a precise balance between cognitive challenge and individual adaptive learning by dynamically adjusting the fusion of cross-domain task complexity and neuroscience incentive mechanisms through a fuzzy logic controller. Compared to traditional mental training algorithms, this invention introduces semantic network analysis technology into the generation of cross-domain knowledge association tasks, and combines a three-dimensional cognitive graph constructed by a graph attention network with deep reinforcement learning optimization strategies to effectively solve the problems of static tasks, singular incentives, and linear paths in traditional algorithms. Attached Figure Description

[0038] Figure 1 This is a schematic diagram illustrating the steps of a human-computer interaction-based cognitive enhancement algorithm proposed in this invention. Detailed Implementation

[0039] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and marked in the accompanying drawings can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0040] This invention discloses a mindset enhancement algorithm based on human-computer interaction.

[0041] Reference Figure 1 As shown, a mindset enhancement algorithm based on human-computer interaction includes the following steps:

[0042] S1. Collect user behavior data, cognitive ability indicators and learning preferences through the human-computer interaction interface, and construct an initial cognitive feature vector including attention concentration and cognitive style classification.

[0043] User behavior data, including click frequency, task response time, and error rate, is collected through a human-computer interaction interface. This data is then combined with a cognitive ability test question bank (including Raven's Reasoning questions and spatial rotation tasks) to obtain user cognitive style classifications and construct an initial cognitive feature vector.

[0044] V = [0.82, 1, 0.45, 0.67] T

[0045] The dimensions represent attention concentration (0-1), cognitive style (0 = field dependent, 1 = field independent), working memory capacity (standardized value), and logical reasoning ability score, respectively.

[0046] S2. Based on the cognitive challenge module, multi-dimensional interactive tasks are dynamically generated according to the user's current cognitive level. The tasks include logical reasoning, abstract modeling, and cross-domain knowledge association tasks. The task difficulty coefficient is adjusted by a fuzzy logic controller to activate the user's higher-order thinking.

[0047] The cognitive challenge module is invoked, and the user's current cognitive level (score P) is assessed. c =0.73) and historical performance (P h =0.68), the task difficulty coefficient D = 0.62 is calculated according to the difficulty formula described in claim 2;

[0048] Generate L3 level cross-domain tasks, such as "derive the correlation between the Fibonacci sequence and the signal period in circuit design", forcibly associating knowledge points in the fields of mathematics and electrical engineering, with an association path length of 2 layers of nodes;

[0049] The task complexity is adjusted by a fuzzy logic controller, with the output logical reasoning steps being 5 and the abstract modeling requirement being intermediate.

[0050] S3. Through the feedback and incentive module, the user's task completion rate and cognitive growth trajectory are analyzed in real time. Combined with the neuroscience-driven dopamine reward simulation mechanism, personalized incentive signals and cognitive diagnostic reports including cognitive load threshold and knowledge transfer efficiency are generated.

[0051] When a user breaks through a cognitive bottleneck for the first time in a task (such as independently completing cross-domain derivation), the dopamine pulse excitation of claim 4 is triggered, with a pulse intensity of 0.8 (difficulty level L3).

[0052] Generate a cognitive diagnostic report, including 12 dimensions of indicators such as cross-domain knowledge transfer efficiency = 73% and metacognitive monitoring frequency = 2.1 times / minute.

[0053] S4. Based on the personalized learning module, cognitive graph modeling technology is used to analyze the characteristics of users' thinking patterns, and an adaptive learning path that matches the user's cognitive development stage is generated through knowledge transfer reinforcement strategies.

[0054] A three-dimensional cognitive graph is constructed based on a graph attention network. The node weights are calculated using the convolutional neural network of claim 6, and the output weight matrix is ​​W = [0.72, 0.65, 0.58].

[0055] When the system detects that the user's memory strength for the "circuit design" knowledge node has dropped to 0.55 (below the threshold of 0.6), it automatically inserts an interval repetition training task (the compensation strategy as described in claim 7).

[0056] S5. Quantify the improvement effect of user mindset through multi-dimensional evaluation indicators, including metacognitive ability, critical thinking index and knowledge integration efficiency, and iteratively optimize algorithm parameters.

[0057] Calculate the comprehensive index of mental enhancement according to the evaluation formula in claim 8:

[0058] M = 0.82 (λ1 = 0.4, λ2 = 0.3, λ3 = 0.3). Adjust the algorithm parameters according to the value of M, for example, increase the adjustment factor k of the fuzzy logic controller from 0.5 to 0.7.

[0059] Reference Figure 1 As shown, in a preferred embodiment, in step S2:

[0060] The logical complexity of the multi-dimensional interactive task is divided into L1 to L5 levels, where L1 is a single-step reasoning task and L5 is a cross-domain knowledge association task, and the association threshold is ≥0.7.

[0061] The adjustment rule for the task difficulty coefficient is as follows:

[0062]

[0063] Where D is the difficulty level, P h To score performance on historical tasks, P c The current cognitive level is rated, and k is a moderating factor;

[0064] In step S2:

[0065] The cross-domain knowledge association task is achieved through semantic network analysis technology, which forces the association of knowledge points from at least two different knowledge domains, and the association path length does not exceed 3 layers of nodes.

[0066] The abstract modeling task includes at least one of mathematical formula derivation and physical principle visualization modeling.

[0067] Using semantic network analysis techniques, we extract related nodes (e.g., "gradient descent" and "cognitive bias") from the knowledge base to construct association paths:

[0068] Gradient descent → Optimize objective function → Confirmation bias in cognitive bias (path length = 3 layers); the correlation coefficient is calculated to be 0.75 (≥0.7 threshold), which meets the requirements of L5 level task.

[0069] When the user's historical performance rating P h =0.75, current cognitive level P c =0.68, when the adjustment factor k=0.6, according to the formula:

[0070]

[0071] Generate medium-difficulty (Level 3) tasks, which include 4 steps of logical reasoning and 2 abstract modeling sub-tasks.

[0072] Reference Figure 1 As shown, in a preferred embodiment, in step S4:

[0073] The cognitive graph modeling technology constructs a three-dimensional graph through graph attention networks. The three-dimensional dimensions include knowledge domains, cognitive ability levels, and thinking pattern characteristics.

[0074] The knowledge transfer enhancement strategy is as follows: when a user completes a task in a specific domain, a cross-domain training task with a relevance of ≥0.7 is forcibly inserted;

[0075] The node weights of the 3D graph are calculated using a convolutional neural network, and the node weight update formula is as follows:

[0076]

[0077] Where σ is the activation function, α ij h represents the node association attention coefficient. j Represent latent vectors for historical tasks;

[0078] In step S4:

[0079] The optimization of the adaptive learning path adopts a deep deterministic strategy gradient algorithm to dynamically adjust the path branches, and the path includes a knowledge forgetting curve compensation strategy.

[0080] The compensation strategy is as follows: when the user's memory strength for a specific knowledge node is lower than the threshold of 0.6, an interval repetition training task is automatically inserted.

[0081] Node weights are calculated using a convolutional neural network, with the latent vector h. j Based on user's historical task data, the system detected that the user's memory strength for a certain node was 0.58 (threshold 0.6), triggering an interval repetition strategy.

[0082] Day 1: Complete the basic probability questions;

[0083] Day 3: Insert questions relating Bayes' theorem to psychological statistics;

[0084] Day 7: Memory strength recovered to 0.68 after intensive training.

[0085] Reference Figure 1 As shown, in a preferred embodiment, in step S3:

[0086] The dopamine reward simulation mechanism generates phased incentive pulses based on the time point when the user breaks through the cognitive bottleneck. The time point is predicted by a hidden Markov model with a prediction error of ≤5%.

[0087] The cognitive diagnostic report includes 12 dimensions of indicators, among which field-independent / dependent cognitive style classification and cross-domain knowledge transfer efficiency are mandatory indicators.

[0088] In step S4:

[0089] M=λ1·R meta +λ2·R crit +λ3·R int

[0090] Among them, R meta R represents the improvement rate of metacognitive ability. crit R is the critical thinking index. int For knowledge integration efficiency, λ1, λ2, and λ3 are weighting coefficients that satisfy λ1 + λ2 + λ3 = 1.

[0091] User B broke through the cognitive bottleneck in the L4 level task, generating impulse intensity = 1.0. The experimental group's sustained task participation rate was 37% higher than that of the control group. Assuming the user's metacognitive ability improvement rate R... meta =22%, Critical Thinking Index R crit =0.75, knowledge integration efficiency R int =81%, weighted calculation:

[0092] M=0.4×22+0.3×0.75+0.3×81=42.25

[0093] Then optimize the algorithm based on the M value.

[0094] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. The substitutions may be replacements of some structures, devices, or method steps, or they may be complete technical solutions. Equivalent substitutions or modifications made to the technical solutions and inventive concepts of the present invention should all be covered within the scope of protection of the present invention.

Claims

1. A mindset enhancement algorithm based on human-computer interaction, characterized in that, The following steps are involved: S1. Collect user behavior data, cognitive ability indicators and learning preferences through the human-computer interaction interface, and construct an initial cognitive feature vector including attention concentration and cognitive style classification. S2. Based on the cognitive challenge module, multi-dimensional interactive tasks are dynamically generated according to the user's current cognitive level. The tasks include logical reasoning, abstract modeling, and cross-domain knowledge association tasks. The task difficulty coefficient is adjusted through a fuzzy logic controller to activate the user's higher-order thinking. S3. Through the feedback and incentive module, the user's task completion rate and cognitive growth trajectory are analyzed in real time. Combined with the neuroscience-driven dopamine reward simulation mechanism, personalized incentive signals and cognitive diagnostic reports including cognitive load threshold and knowledge transfer efficiency are generated. S4. Based on the personalized learning module, cognitive graph modeling technology is used to analyze the characteristics of users' thinking patterns, and an adaptive learning path that matches the user's cognitive development stage is generated through knowledge transfer reinforcement strategies. S5. Quantify the improvement effect of user mindset through multi-dimensional evaluation indicators, including metacognitive ability, critical thinking index and knowledge integration efficiency, and iteratively optimize algorithm parameters.

2. The cognitive enhancement algorithm based on human-computer interaction according to claim 1, characterized in that, In step S2: The logical complexity of the multi-dimensional interactive task is divided into L1 to L5 levels, where L1 is a single-step reasoning task and L5 is a cross-domain knowledge association task, and the association threshold is ≥0.

7. The adjustment rule for the task difficulty coefficient is as follows: Where D is the difficulty level, P h To score performance on historical tasks, P c The current cognitive level is rated, and k is a moderating factor.

3. The cognitive enhancement algorithm based on human-computer interaction according to claim 1, characterized in that, In step S2: The cross-domain knowledge association task is achieved through semantic network analysis technology, which forces the association of knowledge points from at least two different knowledge domains, and the association path length does not exceed 3 layers of nodes. The abstract modeling task includes at least one of mathematical formula derivation and physical principle visualization modeling.

4. The cognitive enhancement algorithm based on human-computer interaction according to claim 1, characterized in that, In step S3: The dopamine reward simulation mechanism generates phased incentive pulses based on the time point when the user breaks through the cognitive bottleneck. The time point is predicted by a hidden Markov model with a prediction error of ≤5%. The cognitive diagnostic report includes 12 dimensions of indicators, among which field-independent / dependent cognitive style classification and cross-domain knowledge transfer efficiency are mandatory indicators.

5. The cognitive enhancement algorithm based on human-computer interaction according to claim 1, characterized in that, In step S4: The cognitive graph modeling technology constructs a three-dimensional graph through graph attention networks. The three-dimensional dimensions include knowledge domains, cognitive ability levels, and thinking pattern characteristics. The knowledge transfer enhancement strategy is as follows: when a user completes a task in a specific domain, a cross-domain training task with a correlation degree ≥ 0.7 is forcibly inserted.

6. The cognitive enhancement algorithm based on human-computer interaction according to claim 1, characterized in that, The node weights of the 3D graph are calculated using a convolutional neural network, and the node weight update formula is as follows: Where σ is the activation function, α ij h represents the node association attention coefficient. j Represent the hidden vectors for historical tasks.

7. The cognitive enhancement algorithm based on human-computer interaction according to claim 1, characterized in that, In step S4: The optimization of the adaptive learning path adopts a deep deterministic strategy gradient algorithm to dynamically adjust the path branches, and the path includes a knowledge forgetting curve compensation strategy. The compensation strategy is as follows: when the user's memory strength for a specific knowledge node is lower than the threshold of 0.6, an interval repetition training task is automatically inserted.

8. The cognitive enhancement algorithm based on human-computer interaction according to claim 1, characterized in that, In step S4: M=λ1·R meta +λ2·R crit +λ3·R int Among them, R meta R represents the improvement rate of metacognitive ability. crit R is the critical thinking index. int For knowledge integration efficiency, λ1, λ2, and λ3 are weighting coefficients and satisfy λ1+λ2+λ3=1.

Images