Knowledge graph dynamic presentation method based on forgetting law

Abstract

This invention discloses a dynamic presentation method for knowledge graphs based on the forgetting curve. The knowledge graph uses knowledge points as nodes and relationships as edges. A memory strength model with memory stability is established for each node. The current memory retention of each node is calculated in real time according to the forgetting curve, and the visual presentation intensity of each node is adjusted accordingly to dynamically present the knowledge graph. When reviewing a node, the increase in its memory stability is diffused to adjacent nodes along the edges of the graph according to the relationship weight, achieving association reinforcement. Simultaneously, the method predicts when the memory retention of each node will fall below the review threshold, generates a review queue, and updates it periodically. This invention achieves dynamic presentation of the knowledge graph, collaborative consolidation of memory based on knowledge relationships, and scientific review scheduling, improving learning and review efficiency.

Description

Technical Field

[0001] This invention belongs to the field of knowledge visualization technology, specifically relating to a method and system for dynamically presenting knowledge graphs based on forgetting patterns. Background Technology

[0002] Knowledge graphs use nodes to represent knowledge points and edges to represent the relationships between them. They can intuitively present the structure and context of knowledge and are widely used in learning assistance, knowledge management, and other fields. In learning scenarios, learners' mastery of each knowledge point will decrease over time, which is consistent with the forgetting curve. To consolidate memory, it is necessary to review easily forgotten knowledge points at appropriate times.

[0003] Regarding the presentation and review of knowledge, existing technologies have the following main shortcomings. First, most existing knowledge graphs are static, with fixed visual representations of each node. This fails to reflect the dynamic changes in learners' mastery of each knowledge point, making it difficult for learners to intuitively determine which knowledge points are nearing forgetting and require priority review. Second, existing review reminder schemes often employ fixed-interval mechanical repetition, failing to consider the memory decay patterns and review history of each knowledge point. This leads to over-review of firmly grasped knowledge points while neglecting timely review of easily forgotten points, resulting in low review efficiency. Third, existing schemes often treat each knowledge point in isolation, failing to utilize the relationships between them and ignoring the stimulating effect of reviewing one knowledge point on the memory of related knowledge points. This fails to reflect the synergistic effect of knowledge associations in the knowledge graph on memory consolidation.

[0004] The common problems with the above solutions are that they fail to organically combine the forgetting curve of memory with the presentation of knowledge graphs and the scheduling of review. They are unable to dynamically present knowledge graphs based on the real-time changes in the retention of each knowledge point and to conduct scientific review scheduling accordingly. They also fail to use knowledge associations to achieve synergistic consolidation of memory.

[0005] To address the above problems, this invention proposes a dynamic presentation method for knowledge graphs based on forgetting patterns. Summary of the Invention

[0006] The purpose of this invention is to address the problems of static presentation, mechanical review intervals, and neglect of knowledge association and synergy in existing knowledge graph presentation and review schemes, and to provide a dynamic presentation method for knowledge graphs based on the forgetting curve.

[0007] To achieve the above-mentioned objectives, the specific technical solution adopted by this invention is as follows: A dynamic knowledge graph presentation method based on forgetting curves, wherein the knowledge graph is composed of knowledge points as nodes and the relationships between knowledge points as edges, with each node corresponding to a knowledge point to be memorized, and the method includes the following steps executed sequentially: Step S1: Establish a memory strength model for each node in the knowledge graph. The memory strength model includes memory stability, which represents the strength of memory retention for that node, and records the last review time for that node. When a user studies or reviews a node, the memory stability of that node is updated based on the time interval between the current review and the previous review, as well as the recall quality of the current review, so that the memory stability of that node increases after review.

[0008] Step S2: For each node, based on its memory stability and the time interval since the last review, calculate the memory retention of that node at the current moment according to the forgetting curve. The memory retention decreases as the time interval since the last review increases and the decay slows down as memory stability increases.

[0009] Step S3: Based on the current memory retention of each node, modulate the visual presentation intensity of each node in the knowledge graph so that the visual presentation intensity of the node changes continuously with the memory retention of that node, and dynamically present the knowledge graph on the display interface accordingly.

[0010] Step S4: When a user reviews a certain node, along the edge of that node in the knowledge graph, the improvement in the memory stability of that node by this review is proportionally diffused to the adjacent nodes according to the association weight of the edge, thereby strengthening the association of the memory stability of the adjacent nodes.

[0011] Step S5: Based on the memory strength model of each node, predict the time when the memory retention of each node will fall below the set review threshold, sort each node according to the predicted time to generate a review queue, and periodically return to step S2. Update the memory retention, visual presentation strength and review queue of each node in real time with the passage of time and user review behavior.

[0012] Furthermore, in step S2, the knowledge graph contains a total of The node, the Each node at the current moment Memory retention According to the forgetting curve, it is determined by the following formula: ;in, ; For the first The last review time for each node; For the first The memory stability of each node, with the unit consistent with time. The larger the value, the slower the memory retention rate decays over time; the higher the memory retention rate, the slower the memory retention rate decays over time. .

[0013] Furthermore, in step S1, the first The node at the th Memory stability after the second review From this node Memory stability after the second review The quality of recall during this review Update as follows: ;in, This is the stability gain coefficient, and its value range is... ; The range of values ​​for the recall quality of this review is: The higher the quality of recall, the greater the improvement in memory stability; The memory stability of each node during its initial learning is taken as the initial stability. .

[0014] Furthermore, in step S1, the recall quality of this review... Determined by user feedback on the node, or by the memory retention rate of the node before this review. estimate: In the formula For the first The node The next review time; when the user recalls the node correctly, a higher recall quality is used, and when the user recalls it incorrectly, a lower recall quality is used.

[0015] Furthermore, in step S3, the first The visual intensity of each node depends on the urgency of reviewing that node. Modulation is performed, and the visual presentation intensity includes the node diameter. Node opacity With node hue They are determined by the following formulas respectively: ; ; ;in, , These are the lower and upper limits of the node diameter, respectively; , These represent the lower and upper limits of node opacity, respectively. The hue baseline, measured in degrees, is used to display one hue for nodes with high memory retention and another hue for nodes with low memory retention. Nodes with higher review urgency have larger diameters and higher opacity, thus standing out more in the knowledge graph.

[0016] Furthermore, in step S4, the node the user is currently reviewing is recorded as the [number]. Each node, the amount of memory stability improvement caused by this review is [number]. ; Regarding the first Each node has an edge connecting it to any of its adjacent nodes. Its memory stability is reinforced by the following formula: ;in, To connect the first The node and the first The association weight of the edges of each node, with values ​​ranging from 1 to 2. The closer the association, the greater the association weight; The correlation diffusion coefficient has a range of values ​​of 100. It is used to control the strength of association reinforcement.

[0017] Furthermore, in step S4, the association reinforcement spreads hop-by-hop along the edges of the knowledge graph, targeting nodes further away from the current review node. for The improvement in memory stability gained at each hop node varies with the correlation diffusion coefficient over each hop. Attenuation, i.e. The diffusion enhancement amount of the jump node and Proportional; when the diffusion enhancement amount is less than the set diffusion cutoff amount It stopped spreading at that time.

[0018] Furthermore, in step S5, the first The memory retention rate of each node fell below the set review threshold. Predicted time Determined by the following formula: ;in, The review threshold has a range of values. ; All nodes will be set according to the predicted time. Nodes are sorted from smallest to largest to create a review queue, with nodes whose prediction time is earlier appearing at the top of the queue. A node's current memory retention is no greater than the review threshold. When that happens, place the node at the head of the review queue.

[0019] Furthermore, in step S5, the review priority of each node in the review queue is further calculated. ,in For the first The importance weight of each node is determined by the number of edges connecting that node in the knowledge graph; the more edges connected, the greater the importance weight. When multiple nodes have the same predicted review time, they are ranked according to review priority. Sort by size from largest to smallest; the update cycle of the periodic return step S2. satisfy Furthermore, when the current memory retention of a node falls below the review threshold, a review reminder will be pushed to the user.

[0020] Based on the same inventive concept, this invention also provides a dynamic presentation system for knowledge graphs based on forgetting patterns. The system includes a knowledge graph storage module, a memory calculation module, an association reinforcement module, a dynamic presentation module, and a review scheduling module.

[0021] The knowledge graph storage module is used to store a knowledge graph with knowledge points as nodes and the relationships between knowledge points as edges, as well as a memory strength model for each node. The memory strength model includes the memory stability of each node and the last review time.

[0022] The memory calculation module is connected to the knowledge graph storage module and is used to calculate the current memory retention of each node and update the memory stability of the node being reviewed when the user reviews it.

[0023] The association reinforcement module is connected to the knowledge graph storage module and the memory computing module, and is used to reinforce the memory stability of adjacent nodes along the edges of the node when the user reviews a certain node.

[0024] The dynamic presentation module is connected to the memory calculation module and is used to adjust the node diameter, opacity and hue of each node according to the current memory retention of each node and the urgency of review, and dynamically present the knowledge graph on the display interface.

[0025] The review scheduling module is connected to the memory computing module and is used to predict when the memory retention of each node will fall below the review threshold, generate a review queue, and push review reminders to the user.

[0026] Compared with the prior art, the beneficial effects of this invention are: This invention establishes a memory strength model with memory stability for each node in a knowledge graph, calculates the memory retention of each node in real time according to the forgetting curve, and modulates the visual presentation intensity of each node accordingly, so that the knowledge graph is dynamically presented as the memory retention changes. At the same time, when reviewing a certain node, the associated nodes along the edge of the graph are reinforced, and a review queue is generated based on the predicted time when the memory retention of each node falls below the review threshold. This realizes the dynamic presentation of the knowledge graph, the synergistic consolidation of memory associations, and the scientific scheduling of review, thereby improving the efficiency of learning and review. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the composition of the knowledge graph dynamic presentation system based on the forgetting law of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention are described clearly and completely below. Obviously, the described embodiments are only some embodiments of this invention, not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0029] It should be noted that the method and system of this invention are applicable to learning and memory consolidation scenarios that organize knowledge points using knowledge graphs and require dynamic presentation and review scheduling based on the forgetting curve. They are particularly suitable for scenarios such as foreign language vocabulary memorization, subject knowledge point review, professional qualification exam preparation, and professional knowledge learning that requires a large amount of memorization, such as medicine and law.

[0030] Furthermore, this invention can also be applied to scenarios with similar knowledge association structures and memory consolidation needs, such as corporate training knowledge management and long-term maintenance of personal knowledge bases. For scenarios that do not involve long-term memory consolidation or where there is no significant correlation between knowledge points, the association reinforcement mechanism and review scheduling strategy can be simplified or adjusted according to actual needs.

[0031] Example 1 This embodiment provides a method for dynamically presenting knowledge graphs based on forgetting patterns. To clearly describe the method of this invention, a small knowledge graph from a foreign language vocabulary learning scenario is selected as the object. This knowledge graph contains... There are 1 node, each corresponding to one of the six vocabulary knowledge points, denoted as node. , , , , , .

[0032] Nodes are connected by semantic relationships, with edges representing the relationships between them. The weight of each edge is: Node With nodes between ,node With nodes between ,node With nodes between ,node With nodes between ,node With nodes between .

[0033] Model parameters are taken as stability gain coefficients Initial stability Correlation diffusion coefficient Review threshold The visual modulation parameters are taken as the lower limit of the node diameter. Upper limit Lower bound of node opacity Upper limit Hue reference value Update cycle Taking the start time of learning as the origin, the current time is the nth time since the start of learning. Heaven, that is .

[0034] Step S1: Establish a memory strength model for each node in the knowledge graph. The memory strength model includes memory stability, which represents the strength of memory retention for that node, and records the last review time for that node. When a user studies or reviews a node, the memory stability of that node is updated based on the time interval between the current review and the previous review, as well as the recall quality of the current review, so that the memory stability of that node increases after review.

[0035] No. The node at the th Memory stability after the second review From this node Memory stability after the second review The quality of recall during this review Update as follows: ;in, This is the stability gain coefficient, and its value range is... ; The range of values ​​for the recall quality of this review is: The higher the quality of recall, the greater the improvement in memory stability; The memory stability of each node during its initial learning is taken as the initial stability. .

[0036] The memory stability of each node during its initial learning is taken as the initial stability. Each time you review, update the formula based on memory stability. Update memory stability.

[0037] The quality of recall during this review Determined by user feedback on the node, or by the memory retention rate of the node before this review. estimate: In the formula For the first The node The timing of the next review session; higher recall quality is used when the user recalls the node correctly, and lower recall quality is used when the user recalls it incorrectly. The review history and stability evolution of each node are as follows.

[0038] node At the Review every day (recall quality) Stability is determined by Updated to ; at the Review every day ( Stability updated to Last review time .

[0039] node After the first , , After three high-quality reviews over three days, the stability was updated to [value missing]. , .

[0040] node Only in the first Review once a day ( Stability , .

[0041] node After the first , After reviewing twice a day , .

[0042] node At the Review once a day ( ), , .node At the Review once a day ( ), , .

[0043] Step S2: For each node, based on its memory stability and the time interval since the last review, calculate the memory retention of that node at the current moment according to the forgetting curve. The memory retention decreases as the time interval since the last review increases and the decay slows down as memory stability increases.

[0044] The knowledge graph contains a total of The node, the Each node at the current moment Memory retention According to the forgetting curve, it is determined by the following formula: ;in, ; For the first The last review time for each node; For the first The memory stability of each node, with the unit consistent with time. The larger the value, the slower the memory retention rate decays over time; the higher the memory retention rate, the slower the memory retention rate decays over time. .

[0045] Substitute the current time into the memory retention formula. Calculate the current memory retention of each node. : ;node : ;node : ;node : ;node : ;node : .

[0046] It is evident that nodes with a high number of reviews and high stability The nodes with the highest memory retention, but with less review and lower stability. , , Memory retention has significantly decreased, and the individual is on the verge of being forgotten.

[0047] Step S3: Based on the current memory retention of each node, modulate the visual presentation intensity of each node in the knowledge graph so that the visual presentation intensity of the node changes continuously with the memory retention of that node, and dynamically present the knowledge graph on the display interface accordingly.

[0048] No. The visual intensity of each node depends on the urgency of reviewing that node. Modulation is performed, and the visual presentation intensity includes the node diameter. Node opacity With node hue They are determined by the following formulas respectively: ; ; ;in, , These are the lower and upper limits of the node diameter, respectively; , These represent the lower and upper limits of node opacity, respectively. The hue baseline, measured in degrees, is used to display one hue for nodes with high memory retention and another hue for nodes with low memory retention. Nodes with higher review urgency have larger diameters and higher opacity, thus standing out more in the knowledge graph.

[0049] Based on the urgency of review and visual modulation formula , , Modulate the visual intensity of each node. Focus on nodes with strong memory. With the node on the verge of being forgotten For example: node The urgency of review Node diameter Opacity hue (Greenish hue indicates mastery); Node The urgency of review Node diameter Opacity hue (Reddish hue indicates near oblivion).

[0050] Based on this, a knowledge graph is dynamically displayed on the interface, showing nodes on the verge of being forgotten. Highlight nodes with larger diameters, higher opacity, and a reddish hue to ensure they are easily remembered. It is then displayed with a smaller diameter, lower opacity, and a greenish tint to make it easier for learners to visually identify the knowledge points that need to be reviewed first.

[0051] Step S4: When a user reviews a certain node, along the edge of that node in the knowledge graph, the improvement in the memory stability of that node by this review is proportionally diffused to the adjacent nodes according to the association weight of the edge, thereby strengthening the association of the memory stability of the adjacent nodes.

[0052] The user's current review node is recorded as the [number]. Each node, the amount of memory stability improvement caused by this review is [number]. ; Regarding the first Each node has an edge connecting it to any of its adjacent nodes. Its memory stability is reinforced by the following formula: ;in, To connect the first The node and the first The association weight of the edges of each node, with values ​​ranging from 1 to 2. The closer the association, the greater the association weight; The correlation diffusion coefficient has a range of values ​​of 100. It is used to control the strength of association reinforcement.

[0053] The reinforcement of associations spreads hop-by-hop along the edges of the knowledge graph, affecting nodes further from the current review node. for The improvement in memory stability gained at each hop node varies with the correlation diffusion coefficient over each hop. Attenuation, i.e. The diffusion enhancement amount of the jump node and Proportional; when the diffusion enhancement amount is less than the set diffusion cutoff amount It stopped spreading at that time.

[0054] Set the user's current review node (memory quality) Its memory stability is from before review Updated to The increase in memory stability caused by this review For nodes If there are adjacent nodes connected by an edge, then according to the association strengthening formula... Strengthen associations: neighboring nodes ( stability increment Stability from Updated to Adjacent nodes ( Stability increment Stability from Updated to Adjacent nodes ( Stability increment Stability from Updated to Therefore, review nodes It not only strengthens one's own memory, but also synergistically consolidates the memory of adjacent knowledge points along knowledge connections.

[0055] Step S5: Based on the memory strength model of each node, predict the time when the memory retention of each node will fall below the set review threshold, sort each node according to the predicted time to generate a review queue, and periodically return to step S2. Update the memory retention, visual presentation strength and review queue of each node in real time with the passage of time and user review behavior.

[0056] No. The memory retention rate of each node fell below the set review threshold. Predicted time Determined by the following formula: ;in, The review threshold has a range of values. ; All nodes will be set according to the predicted time. Nodes are sorted from smallest to largest to create a review queue, with nodes whose prediction time is earlier appearing at the top of the queue. A node's current memory retention is no greater than the review threshold. When that happens, place the node at the head of the review queue.

[0057] The review priority is further calculated for each node in the review queue. ,in For the first The importance weight of each node is determined by the number of edges connecting that node in the knowledge graph; the more edges connected, the greater the importance weight. When multiple nodes have the same predicted review time, they are ranked according to review priority. Sort by size from largest to smallest.

[0058] The update cycle of the periodic return step S2 satisfy Furthermore, when the current memory retention of a node falls below the review threshold, a review reminder will be pushed to the user.

[0059] According to the formula for predicting time Take the review threshold ( ), calculate the predicted time when the memory retention of each node falls below the review threshold: , , , , , .

[0060] Since the current time is the [number]th ,node , , , , The current memory retention rate is no greater than the review threshold. Place it at the head of the review queue and sort it by review priority. Ranking, with importance weights Determined by the number of edges connected to each node (node) connect Strip, ;node connect Strip, ;node , , Each connection Strip, ).

[0061] Calculate the review priority for each node that has been breached: , , , , ; Nodes are sorted by priority from largest to smallest. , , , , .node Current memory retention The node that has not yet fallen below the threshold is placed after the node that has fallen below it, according to its predicted time. Therefore, the final review queue is: node , , , , , .

[0062] The system is updated on a regular schedule. The system periodically returns to step S2, updating the memory retention, visual presentation intensity, and review queue of each node in real time based on the user's review behavior over time. When the memory retention of a node falls below the review threshold, a review reminder is pushed to the user.

[0063] The above results show that the method in this embodiment calculates memory retention in real time and dynamically presents the knowledge graph based on the forgetting patterns of each node. Knowledge points that are about to be forgotten are highlighted on the graph. The review queue prioritizes important knowledge points that are about to be forgotten. When reviewing a node, the memory of adjacent knowledge points is consolidated along the knowledge association. This verifies the effectiveness of the method of the present invention in terms of dynamic presentation, scientific review scheduling and memory consolidation.

[0064] Example 2 This embodiment provides a dynamic knowledge graph presentation system based on forgetting patterns, such as... Figure 1 As shown, the system includes a knowledge graph storage module, a memory computing module, an association reinforcement module, a dynamic presentation module, and a review scheduling module.

[0065] The knowledge graph storage module is used to store a knowledge graph with knowledge points as nodes and the relationships between knowledge points as edges, as well as a memory strength model for each node; the memory strength model includes the memory stability of each node. Compared to the last review time The edge stores the association weights connecting the corresponding two nodes. In this embodiment, the knowledge graph storage module stores six nodes and the semantic association edges and association weights between them.

[0066] The memory computing module is connected to the knowledge graph storage module and is used to calculate the memory retention formula. Calculate the current memory retention of each node; and update the formula based on memory stability when a user studies or reviews a particular node. Update the memory stability of the reviewed node, and write the updated memory stability and the last review time back to the knowledge graph storage module.

[0067] The association reinforcement module is connected to the knowledge graph storage module and the memory calculation module, and is used to, when a user reviews a certain node, follow the edge of that node in the knowledge graph according to the association reinforcement formula. The memory stability of adjacent nodes is reinforced by association, and the reinforced memory stability is written back to the knowledge graph storage module; the association reinforcement spreads hop by hop along the edge, and stops spreading when the amount of reinforcement is less than the diffusion cutoff amount.

[0068] The dynamic presentation module is connected to the memory calculation module and is used to adjust the review urgency based on the current memory retention of each node. For the node diameter of each node Opacity With hue The knowledge graph is modulated and dynamically presented on the display interface, highlighting nodes that are on the verge of being forgotten.

[0069] The review scheduling module is connected to the memory calculation module and is used to calculate the prediction time formula. Predict the time when memory retention at each node falls below the review threshold, sort the nodes by predicted time in ascending order, and then prioritize the nodes whose current memory retention has fallen below the review threshold. Place it at the head of the queue to generate a review queue, and push a review reminder to the user when the memory retention of a node falls below the review threshold.

[0070] In this embodiment, during system operation, the knowledge graph storage module provides the knowledge graph and memory strength models for each node; the memory calculation module periodically calculates the current memory retention of each node and updates the memory stability of the reviewed node during user review; the association reinforcement module strengthens the association between adjacent nodes during review; the dynamic presentation module modulates the visual presentation intensity of each node according to memory retention and dynamically presents the knowledge graph; and the review scheduling module predicts review times, generates review queues, and pushes review reminders. These modules work collaboratively to achieve dynamic presentation of the knowledge graph based on forgetting patterns and scientific review scheduling.

[0071] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic presentation method for knowledge graphs based on forgetting patterns, characterized in that, The knowledge graph uses knowledge points as nodes and the relationships between knowledge points as edges. Each node corresponds to a knowledge point to be memorized. The method includes the following steps executed sequentially: Step S1: Establish a memory strength model for each node in the knowledge graph. The memory strength model includes memory stability, which represents the strength of memory retention for that node, and records the last review time for that node. When a user studies or reviews a node, the memory stability of that node is updated based on the time interval between the current review and the previous review, as well as the recall quality of the current review, so that the memory stability of that node increases after review. Step S2: For each node, based on its memory stability and the time interval since the last review, calculate the memory retention of that node at the current moment according to the forgetting curve. The memory retention decreases as the time interval since the last review increases and the decay slows down as memory stability increases. Step S3: Based on the current memory retention of each node, modulate the visual presentation intensity of each node in the knowledge graph so that the visual presentation intensity of the node changes continuously with the memory retention of that node, and dynamically present the knowledge graph on the display interface accordingly. Step S4: When a user reviews a certain node, along the edge of that node in the knowledge graph, the improvement in the memory stability of that node by this review is proportionally diffused to the adjacent nodes according to the association weight of the edge, thereby strengthening the association of the memory stability of the adjacent nodes. Step S5: Based on the memory strength model of each node, predict the time when the memory retention of each node will fall below the set review threshold, sort each node according to the predicted time to generate a review queue, and periodically return to step S2. Update the memory retention, visual presentation strength and review queue of each node in real time with the passage of time and user review behavior.

2. The method according to claim 1, characterized in that, In step S2, the knowledge graph contains a total of The node, the Each node at the current moment Memory retention According to the forgetting curve, it is determined by the following formula: ;in, ; For the first The last review time for each node; For the first The memory stability of each node, with the unit consistent with time. The larger the value, the slower the memory retention rate decays over time; the higher the memory retention rate, the slower the memory retention rate decays over time. .

3. The method according to claim 2, characterized in that, In step S1, the first The node at the th Memory stability after the second review From this node Memory stability after the second review The quality of recall during this review Update as follows: ;in, This is the stability gain coefficient, and its value range is... ; The range of values ​​for the recall quality of this review is: The higher the quality of recall, the greater the improvement in memory stability; The memory stability of each node during its initial learning is taken as the initial stability. .

4. The method according to claim 3, characterized in that, In step S1, the recall quality of this review... Determined by user feedback on the node, or by the memory retention rate of the node before this review. estimate: In the formula For the first The node The next time to review.

5. The method according to claim 4, characterized in that, In step S3, the first The visual intensity of each node depends on the urgency of reviewing that node. Modulation is performed, and the visual presentation intensity includes the node diameter. Node opacity With node hue They are determined by the following formulas respectively: ; ; ;in, , These are the lower and upper limits of the node diameter, respectively; , These represent the lower and upper limits of node opacity, respectively. The hue baseline, measured in degrees, is used to display one hue for nodes with high memory retention and another hue for nodes with low memory retention. Nodes with higher review urgency have larger diameters and higher opacity, thus standing out more in the knowledge graph.

6. The method according to claim 5, characterized in that, In step S4, the node the user is currently reviewing is recorded as the node number. Each node, the amount of memory stability improvement caused by this review is [number]. ; Regarding the first Each node has an edge connecting it to any of its adjacent nodes. Its memory stability is reinforced by the following formula: ;in, To connect the first The node and the first The association weight of the edges of each node, with values ​​ranging from 1 to 2. The closer the association, the greater the association weight; The correlation diffusion coefficient has a range of values ​​of 100. It is used to control the strength of association reinforcement.

7. The method according to claim 6, characterized in that, In step S4, the association reinforcement spreads hop-by-hop along the edges of the knowledge graph, affecting nodes further from the current review node. for The improvement in memory stability gained at each hop node varies with the correlation diffusion coefficient over each hop. Attenuation, i.e. The diffusion enhancement amount of the jump node and Proportional; when the diffusion enhancement amount is less than the set diffusion cutoff amount It stopped spreading at that time.

8. The method according to claim 7, characterized in that, In step S5, the first The memory retention rate of each node fell below the set review threshold. Predicted time Determined by the following formula: ;in, The review threshold has a range of values. ; All nodes will be set according to the predicted time. Nodes are sorted from smallest to largest to create a review queue, with nodes whose prediction time is earlier appearing at the top of the queue. A node's current memory retention is no greater than the review threshold. When that happens, place the node at the head of the review queue.

9. The method according to claim 8, characterized in that, In step S5, the review priority of each node in the review queue is further calculated. ,in For the first The importance weight of each node is determined by the number of edges connecting that node in the knowledge graph; the more edges connected, the greater the importance weight. When multiple nodes have the same predicted review time, they are ranked according to review priority. Sort by size from largest to smallest.

10. A dynamic presentation system for knowledge graphs based on forgetting patterns, used to perform the method described in any one of claims 1-9, the system comprising a knowledge graph storage module, a memory computation module, an association reinforcement module, a dynamic presentation module, and a review scheduling module; The knowledge graph storage module is used to store a knowledge graph with knowledge points as nodes and the relationships between knowledge points as edges, as well as the memory strength model of each node. The memory strength model includes the memory stability of each node and the last review time. The memory calculation module is connected to the knowledge graph storage module and is used to calculate the current memory retention of each node and update the memory stability of the node being reviewed when the user reviews it. The association reinforcement module is connected to the knowledge graph storage module and the memory computing module, and is used to reinforce the memory stability of adjacent nodes along the edge of the node when the user reviews a certain node. The dynamic presentation module is connected to the memory calculation module and is used to adjust the node diameter, opacity and hue of each node according to the current memory retention of each node and the urgency of review, and dynamically present the knowledge graph on the display interface. The review scheduling module is connected to the memory computing module and is used to predict when the memory retention of each node will fall below the review threshold, generate a review queue, and push review reminders to the user.

Images