Automatic problem-solving methods, devices, and electronic equipment for application test questions
By identifying and constructing condition nodes and question nodes for application test questions, a computational graph data structure is built, overcoming the limitations of photo-based question search technology and enabling automatic solutions and test paper generation in scenarios where there are no resources in the question bank.
Patent Information
- Application Number
- CN202210780252.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-07-04
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2042-07-04
AI Technical Summary
Existing photo-based question-solving technologies have high requirements for the capacity and maintenance of resource databases, and cannot automatically answer application questions not contained in the resource databases, thus having significant limitations.
By identifying condition nodes and question nodes in application test questions, entity attribute information and relationship information are extracted, and a computational graph data structure is constructed. Based on this, a solution path search is performed and solution steps are generated.
It enables automatic solutions and thought process guidance in scenarios where there are no questions in the question bank, reducing the need for manual solutions and generating test papers without original questions for ability testing.
Smart Images

Figure CN115168539B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent education technology, specifically to methods, devices, and electronic equipment for automatically solving test questions. Background Technology
[0002] Word problems are problems that use language or words to describe facts, reflect certain mathematical relationships (such as quantitative relationships, positional relationships, etc.), and solve for unknown quantities. Each word problem includes known conditions and the question to be asked.
[0003] In primary and secondary education, a large number of math and physics word problems exist to help students better grasp mathematical and physics knowledge. However, when students choose to answer these problems, those without sufficient knowledge cannot provide guidance. The emergence of image-based problem-solving technology has, to some extent, enabled the automatic solution of these word problems. This technology relies on a large database of problems, using OCR to obtain the text of the problem from a photograph, and then searching the database for a matching solution. If a match is found, the solution is returned. However, this technology cannot automatically solve word problems not found in the database. Therefore, image-based problem-solving technology places high demands on the capacity and maintenance of the database and still has significant limitations.
[0004] In view of this, the present invention aims to propose a solution for automatically answering application test questions autonomously, and hereby proposes this patent. Summary of the Invention
[0005] To address the aforementioned problems, this invention provides an automatic problem-solving method, apparatus, and electronic device for application test questions. Specifically, the following technical solution is adopted:
[0006] An automatic problem-solving method for application test questions includes:
[0007] For application test questions, the system performs question comprehension, identifies condition nodes and question nodes in the application test questions, and extracts entity attribute information, entity relationship information between condition nodes and question nodes, and between each question node;
[0008] Based on the extracted entity attribute information and entity relationship information, the solution path from the condition node to the question node for the application test question is determined;
[0009] The problem-solving steps for the application test questions are generated based on the described problem-solving path.
[0010] As an optional embodiment of the present invention, in the automatic problem-solving method for application test questions described in the present invention, the step of understanding the application test questions includes:
[0011] When it is determined that the application test question includes multiple question nodes, the application test question is split into question nodes corresponding to each question and the relationship information between each question node (order relationship, parallel relationship).
[0012] As an optional embodiment of the present invention, in the automatic problem-solving method for application test questions of the present invention, the step of extracting entity attribute information in the application test questions includes extracting entity attribute information from the condition nodes and question nodes in the application test questions respectively.
[0013] The entity attribute information in the condition node includes condition name, condition attribute category, and condition attribute value. The condition name includes the object name extracted from the condition node, the condition attribute category includes the category to which the object in the condition node belongs, and the condition attribute value is the attribute parameter and parameter value extracted from the condition node corresponding to each condition attribute category.
[0014] The entity attribute information of the problem node includes the problem name, the problem attribute category, and the problem attribute value. The problem name includes the name of the object extracted from the problem node, the problem attribute category includes the category to which the object extracted from the problem node belongs, and the problem attribute value is the attribute parameter to be determined and the parameter value of the object extracted from the problem node.
[0015] Optionally, the entity attribute information of the condition node is stored as {name: [], attribute category: [], attribute value: [{name:, value:, unit:}, {name:, value:, unit:}]};
[0016] Optionally, the entity attribute information of the problem node is stored as {name: [], attribute category: [], attribute value: [{name:, value: unknown, unit:}]}.
[0017] As an optional embodiment of the present invention, in the automatic problem-solving method for application test questions described in the present invention, the step of extracting entity relationship information from the application test questions includes:
[0018] Based on the information about the relationships between the problem nodes obtained from the understanding of the question, determine whether there are logical operation relationships between the problem nodes in the application test question;
[0019] If so, then extract the entity relationship information between each problem node based on the entity attribute information of each problem node.
[0020] As an optional embodiment of the present invention, in the automatic problem-solving method for application questions described in the present invention, the step of determining the problem-solving path from the condition node to the question node based on the extracted entity attribute information and entity relationship information includes:
[0021] By introducing basic knowledge points related to the entity attribute information and entity relationship information, a computational graph data structure for solving each problem node is constructed;
[0022] Searching using computational graph data structures reveals the solution path from condition nodes to problem nodes in application problems.
[0023] As an optional embodiment of the present invention, in the automatic problem-solving method for application test questions described in the present invention, the step of introducing basic knowledge points and entity attribute information and entity relationship information to jointly construct a computational graph data structure includes:
[0024] The basic knowledge points that can be introduced are determined based on the entity attribute information and entity relationship information of the application test questions;
[0025] Based on the basic knowledge points, construct a computational graph data structure by taking the question attribute information of each condition node, the question attribute information of each problem node, and the question relationship information of each problem node;
[0026] The computational graph data structure includes condition node attribute information and edge attribute information. The node attribute information includes condition node attribute information, problem node attribute information, and attribute information of the predecessor nodes that need to be solved in the process of solving the problem node from the condition nodes. The edge attribute information includes all paths from the condition node attribute information to the problem node attribute information. As an optional embodiment of the present invention, in the automatic problem-solving method for application questions, the step of searching based on the computational graph data structure to find the solution path from the condition nodes to the node to be solved in the application question includes:
[0027] Based on a computational graph data structure, the loop continues until a problem node is successfully solved or the maximum number of loop iterations is reached;
[0028] If the problem node is solved successfully, the path of the self-looping process is formed into a node-path graph with the problem node as the starting point and the condition node as the ending point.
[0029] The shortest path is searched in the node-path graph as the solution path from the condition node to the unknown node in the application problem.
[0030] As an optional embodiment of the present invention, in the automatic problem-solving method for application test questions described in the present invention, during the shortest path search process, the relevant attributes of the current node are calculated in conjunction with the predecessor node, and it is determined whether the values are computable. If not, unit conversion is required, and then the relevant node attribute information in the computation graph data structure is updated.
[0031] As an optional embodiment of the present invention, in the automatic problem-solving method for application test questions described in the present invention, the step of generating application test question solutions based on the problem-solving path includes:
[0032] Based on the problem-solving approach and the corresponding unit conversions, the most basic calculation steps are formed;
[0033] Then, based on the calculation steps, the corresponding guidance steps are generated.
[0034] This invention also provides an automatic problem-solving device for application test questions, comprising:
[0035] The question comprehension module performs question comprehension on application test questions, extracting entity attribute information and entity relationship information from the application test questions;
[0036] The problem-solving reasoning module introduces basic knowledge points and entity attribute information and entity relationship information to jointly construct a computational graph data structure. Based on the computational graph data structure, a search is performed to find the solution path from the condition node to the unknown quantity node of the application problem. Based on the solution path, the solution steps of the application problem are generated.
[0037] As an optional embodiment of the present invention, an automatic problem-solving device for application test questions in this embodiment further includes:
[0038] The text preprocessing module preprocesses the acquired application test text into the input format of the question comprehension module;
[0039] The post-processing normalization module standardizes the format of the final calculation process.
[0040] The present invention also provides an electronic device, including a processor and a memory, the memory being used to store a computer-executable program, wherein when the computer program is executed by the processor, the processor executes the automatic problem-solving method for the application test questions.
[0041] The present invention also provides a computer-readable storage medium storing a computer-executable program, which, when executed, implements an automatic problem-solving method for the application test questions.
[0042] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0043] The automatic problem-solving method of this invention targets application problems including, but not limited to, mathematical application problems and physics application problems. Specifically, it addresses scenarios where there are no resources in the question bank for producing and grading application problems. It iteratively and automatically summarizes problem-solving methods and thought processes for application problems in teaching, and automatically provides solutions and thought process guidance for application problems for which there are no resources in the question bank. The automatic problem-solving method of this invention can solve the following problems:
[0044] 1) When there are no resources in the question bank during the photo search and grading process, the question will be automatically answered and the answer will be provided.
[0045] 2) Reduce costs and improve efficiency for manual production of our own question bank; some questions can be answered without human intervention.
[0046] In addition, the automatic problem-solving method of the present invention is designed for scenarios where there are no resources for generating test papers for application test questions. By changing the wording, entities, etc., it can generate questions on its own and generate some questions that are not found in the question bank for ability testing, etc. Attached image description:
[0047] Figure 1 A flowchart illustrating the automatic problem-solving method for test questions in this embodiment of the invention;
[0048] Figure 2 Example steps of graph traversal search algorithm in this invention Figure 1 ;
[0049] Figure 3 Example steps of graph traversal search algorithm in this invention Figure 2 ;
[0050] Figure 4 Example steps of graph traversal search algorithm in this invention Figure 3 ;
[0051] Figure 5 Example steps of graph traversal search algorithm in this invention Figure 4 ;
[0052] Figure 6 Example steps of graph traversal search algorithm in this invention Figure 5 ;
[0053] Figure 7 A schematic diagram of a graph traversal search algorithm based on a computational graph data structure according to an embodiment of the present invention;
[0054] Figure 8 Examples of applications of this invention include the automatic solution results of test questions;
[0055] Figure 9 A block diagram of an automatic problem-solving device for test questions according to an embodiment of the present invention. Detailed Implementation
[0056] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.
[0057] Therefore, the following detailed description of embodiments of the present invention is not intended to limit the scope of the claimed invention, but merely illustrates some embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.
[0058] It should be noted that, unless otherwise specified, the embodiments and features and technical solutions in the present invention can be combined with each other.
[0059] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0060] In the description of this invention, it should be noted that the terms "upper," "lower," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use, or the orientation or positional relationship commonly understood by those skilled in the art. These terms are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention. In addition, the terms "first," "second," etc., are only used to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0061] See Figure 1 As shown in this embodiment, an automatic problem-solving method for application test questions includes:
[0062] For application test questions, the system performs question comprehension, identifies condition nodes and question nodes in the application test questions, and extracts entity attribute information, entity relationship information between condition nodes and question nodes, and between each question node;
[0063] Based on the extracted entity attribute information and entity relationship information, the solution path from the condition node to the question node for the application test question is determined;
[0064] The problem-solving steps for the application test questions are generated based on the described problem-solving path.
[0065] The automatic problem-solving method in this embodiment targets application problems including, but not limited to, mathematical and physics application problems. This method addresses scenarios where there are no resources in the question bank for producing and grading application problems. It iteratively and automatically summarizes the problem-solving methods and thought processes for application problems in teaching, and automatically solves and provides thought process guidance for application problems for which there are no resources in the question bank. This automatic problem-solving method can solve the following problems:
[0066] 1) When there are no resources in the question bank during the photo search and grading process, the question will be automatically answered and the answer will be provided.
[0067] 2) Reduce costs and improve efficiency for manual production of our own question bank; some questions can be answered without human intervention.
[0068] In addition, the automatic problem-solving method in this embodiment is designed for scenarios where there are no resources for generating test papers for application questions. By changing the wording, entities, etc., it can generate questions on its own, such as generating questions that are not found in the question bank, for ability testing, etc.
[0069] As an optional implementation of this embodiment, one method for automatically solving application test questions includes understanding the test questions:
[0070] When it is determined that the application test question includes multiple question nodes, the application test question is split into question nodes corresponding to each question and the relationship information between each question node.
[0071] When an application problem contains multiple questions, it's crucial to determine if there are logical or parallel relationships between them. Questions with logical relationships are interconnected and must be solved sequentially, with the answer to the previous question serving as a condition for solving the next. Parallel questions are unrelated and can be solved individually using their respective condition nodes. This embodiment's problem understanding not only requires identifying the condition and question nodes in the application problem but also further uncovering the implicit conditions to achieve automatic solutions for the question nodes.
[0072] As an optional implementation of this embodiment, in an automatic problem-solving method for application questions in this embodiment, the step of extracting entity attribute information in the application questions includes extracting entity attribute information from condition nodes and question nodes in the application questions respectively.
[0073] The entity attribute information in the condition node includes condition name, condition attribute category, and condition attribute value. The condition name includes the object name extracted from the condition node, the condition attribute category includes the category to which the object in the condition node belongs, and the condition attribute value is the attribute parameter and parameter value corresponding to each condition attribute category extracted from the condition node.
[0074] The entity attribute information of the problem node includes the problem name, the problem attribute category, and the problem attribute value. The problem name includes the name of the object extracted from the problem node, the problem attribute category includes the category to which the object extracted from the problem node belongs, and the problem attribute value is the attribute parameter to be determined and the parameter value of the object extracted from the problem node.
[0075] This embodiment extracts information from condition nodes and problem nodes, lists the known objects and their attribute information, and lists the solution objects and their attribute information, and then combines it with existing basic knowledge to construct the problem-solving process and results of solving problem nodes from known nodes.
[0076] Optionally, the entity attribute information of the condition node is stored as {name: [], attribute category: [], attribute value: [{name:, value:, unit:}, {name:, value:, unit:}]};
[0077] Optionally, the entity attribute information of the problem node is stored as {name: [], attribute category: [], attribute value: [{name:, value: unknown, unit:}]}.
[0078] As an optional implementation of this embodiment, in an automatic problem-solving method for application test questions, the step of extracting entity relationship information from the application test questions includes:
[0079] Based on the information about the relationships between the problem nodes obtained from the understanding of the question, determine whether there are logical operation relationships between the problem nodes in the application test question;
[0080] If so, then extract the entity relationship information between each problem node based on the entity attribute information of each problem node.
[0081] When the application test contains multiple question nodes with sequential relationships, it is necessary to identify each question node, determine the entity relationship information between each question node, and automatically answer each question node in sequence based on the entity relationship information.
[0082] As an optional implementation of this embodiment, the method for determining the solution path from the condition node to the question node based on the extracted entity attribute information and entity relationship information includes:
[0083] By introducing basic knowledge points related to the entity attribute information and entity relationship information, a computational graph data structure for solving each problem node is constructed;
[0084] Searching using computational graph data structures reveals the solution path from condition nodes to problem nodes in application problems.
[0085] Furthermore, in one embodiment of the automatic problem-solving method for application test questions, the step of introducing basic knowledge points and entity attribute information and entity relationship information to jointly construct a computational graph data structure includes:
[0086] The basic knowledge points that can be introduced are determined based on the entity attribute information and entity relationship information of the application test questions;
[0087] Based on the basic knowledge points, construct a computational graph data structure by taking the question attribute information of each condition node, the question attribute information of each problem node, and the question relationship information of each problem node;
[0088] The computational graph data structure includes condition node attribute information and edge attribute information. The node attribute information includes condition node attribute information, problem node attribute information, and the attribute information of the predecessor nodes that need to be solved in the middle of solving the problem node from the condition node. The edge attribute information includes all paths to solve the problem node attribute information from the condition node attribute information.
[0089] This embodiment of the automatic problem-solving method for application test questions designs a graph-based data structure. Prior knowledge is introduced by the teaching and research teacher and, together with entity attributes and entity relationships, constructs a computational graph. Here, prior knowledge refers to basic common sense knowledge, such as rabbits have 4 legs, chickens have 2 legs, the area of a rectangle equals length times width, and the volume of a cone equals base area times height divided by 3, etc.
[0090] This step mainly involves introducing knowledge about the conditions and problems involved in the previous steps, thereby expanding the solid attributes and relationships, and completing the construction of the graph (node attributes and edge attributes).
[0091] For ease of understanding, this embodiment uses a computational graph data structure to search for specific examples of the solution path from condition nodes to the node of the unknown quantity in the application problem, as follows:
[0092] Specific examples:
[0093] Problem: A batch of small stones, used for road construction, was transported to a certain section of road and piled into a cone-shaped pile. The circumference of the base was measured to be 18.84 meters, and the height was 2.8 meters. If this pile of stones were used to pave a 2-meter-wide road surface with a thickness of 2 centimeters, approximately how long would it cover?
[0094] Assuming the diagram structure is assembled, such as Figure 2 As shown, this includes prior knowledge such as formulas and common sense, relationships predicted by the model, and identifying which nodes are known conditions and which are problem nodes. In the initial state, nodes with known numerical attributes are as follows: Figure 2 As shown in the diagram of the circular nodes (condition nodes), the problem to be solved is... Figure 2 The pentagonal node (problem node) in the diagram.
[0095] First traversal Figure 3 As shown, it will determine whether a node can be solved based on each node and its relationship. In the first traversal, the "circumference of the circle" of the triangle node (predecessor node) is calculated.
[0096] The second traversal is as follows Figure 4As shown, the triangle node (predecessor node) from the previous traversal has now become a condition node (circle node: circumference of the circle). In the second traversal, the "diameter of the circle" of the triangle node (predecessor node) is calculated.
[0097] And so on, the third iteration yields the area of the circle, the fourth iteration yields the base area of the cone, and the fifth iteration yields the volume of the cone.
[0098] The volume of the cuboid is obtained on the sixth traversal, such as... Figure 5 .
[0099] The seventh iteration yields the final problem's solution: "how long can it be tiled?", which is the length of the cuboid. Figure 6 .
[0100] Furthermore, after constructing the computational graph data structure, in the automatic problem-solving method for application questions in this embodiment, the step of searching based on the computational graph data structure to find the solution path from the condition node to the question node of the application question includes:
[0101] Based on a computational graph data structure, the loop continues until a problem node is successfully solved or the maximum number of loop iterations is reached;
[0102] If the problem node is solved successfully, the path of the self-looping process is formed into a node-path graph with the problem node as the starting point and the condition node as the ending point.
[0103] The shortest path is searched in the node-path graph as the solution path from the condition node to the unknown node in the application problem.
[0104] For the computational graph data structures mentioned above, a special graph traversal search algorithm is designed, such as... Figure 7 As shown, the main steps include:
[0105] 1. Initialize the condition nodes and problem nodes;
[0106] 2. The loop continues until the problem node is successfully solved or STAT reaches n;
[0107] 3. If the problem node is solved successfully, the path of the above process is formed into a graph with Ques as the starting point and Conds as the ending point. Each point in the graph contains the point set state of the predecessor conditions.
[0108] 4. Search for the shortest path.
[0109] This step mainly involves searching within the computational graph data structure mentioned above to find the shortest path from the condition node to the node of the unknown quantity.
[0110] The specific process includes:
[0111] Initialize an empty graph structure G;
[0112] While {problem node solved or traversal depth reaches the number of nodes};
[0113] for { iterate through each node};
[0114] for {the set of predecessor nodes for each node};
[0115] Update node status;
[0116] Record the number of steps (depth) of the shortest path to the current node and the node ID of the previous step, and update the graph structure G;
[0117] Perform a breadth-first search on the graph structure G to find the shortest path from the problem node (pentagon node) to the initial condition node (circular node).
[0118] The graph structure G in the process is Figure 7 In the diagram to the right of the blue arrow, the pentagonal nodes are the nodes that are ultimately calculated (corresponding to...). Figure 2 The pentagonal nodes), and the circular nodes are the initial known condition nodes (corresponding to...). Figure 2 (circular nodes).
[0119] Furthermore, in one embodiment of the automatic problem-solving method for application questions, during the shortest path search process, the relevant attributes of the current node are calculated in conjunction with the predecessor node. At the same time, it is determined whether the values are computable. If not, unit conversion is required, and then the relevant node attribute information in the computation graph data structure is updated.
[0120] In this embodiment, the automatic problem-solving method, when solving the shortest path from the condition node to the node to be solved, excludes operations such as entity attribute updates. The overall computational complexity is MAX O(n*n*Small_C), specifically n*n*Small_C+n*Small_C+Small_C*Small_C, where Small_C is the average number of precondition sets for the node (between 1 and 5).
[0121] In an automatic problem-solving method for application test questions according to this embodiment, the step of generating application test question solutions based on the problem-solving path includes:
[0122] Based on the problem-solving approach and the corresponding unit conversions, the most basic calculation steps are formed;
[0123] Then, based on the calculation steps, the corresponding guidance steps are generated.
[0124] For example:
[0125] example:
[0126] A batch of small stones, specifically for road construction, was transported to a certain section of road and piled into a cone-shaped pile. The circumference of the base was measured to be 18.84 meters, and the height was 2.8 meters. What is the volume of this pile in cubic meters? If this pile of stones were used to pave a 2-meter-wide road surface with a thickness of 2 centimeters, approximately how long would the road surface be?
[0127] Step 1: Text Preprocessing
[0128] Process the text into the model's input format.
[0129] Step 2: Problem Breakdown
[0130] The problem that ultimately needs to be solved is broken down into smaller parts.
[0131] For example, this problem can be broken down into two questions: (1) What is the volume of this pile of stones in cubic meters? (2) If this pile of stones is laid on a 2-meter-wide road with a thickness of 2 centimeters, approximately how long can it be laid?
[0132] Step 3: Entity Attribute Extraction
[0133] Extract the known conditions and the quantities to be determined from the problem.
[0134] The given conditions are:
[0135] {Name: Pile of Stones, Attribute Category: Geometry - Cone, Attribute Value: [{Name: Circumference of Cone Base, Value: 18.84, Unit: meters}, {Name: Height of Cone, Value: 2.8, Unit: meters}]}
[0136] {Name: Road Surface, Attribute Category: Geometry - Cuboid, Attribute Value: [{Name: Width of the Cuboid, Value: 2, Unit: meters}, {Name: Height of the Cuboid, Value: 2, Unit: centimeters}]}
[0137] The quantity to be determined is:
[0138] {Name: Pile of Stones, Attribute Category: Geometry - Cone, Attribute Value: [{Name: Volume of Cone, Value: Unknown, Unit: Cubic Meters}]}
[0139] {Name: Road Surface, Attribute Category: Geometry - Cuboid, Attribute Value: [{Name: Length of the Cuboid, Value: Unknown, Unit: No Requirement}]}
[0140] Step 4: Entity Relationship Extraction
[0141] Extract the relationships between entities in the problem.
[0142] In the second part of this problem, "the pile of stones is spread 2 centimeters thick on a 2-meter wide road surface," which means that the volume of the stone pile cone should be equal to the volume of the road surface paved with a cuboid. Therefore, we get an implicit condition: "the volume of the stone pile is equal to the volume of the paved road surface."
[0143] Step 5: Construction of the computational graph data structure
[0144] A computational graph data structure was designed, in which prior knowledge was introduced by the teaching and research teacher, and together with entity attributes and entity relationships, a computational graph was constructed. The prior knowledge here refers to some basic common sense knowledge, such as rabbits have 4 legs, chickens have 2 legs, the area of a rectangle is equal to length times width, the volume of a cone is equal to base area times height divided by 3, etc.
[0145] This step mainly involves introducing knowledge about the conditions and problems involved in the previous steps, thereby expanding the solid attributes and relationships, and completing the construction of the graph (node attributes and edge attributes).
[0146] Step 6: Reasoning and Calculation
[0147] For the computational graph data structure mentioned in step 5, a special graph traversal search algorithm was designed (see diagram). Figure 7 (As shown).
[0148] This step mainly involves searching within the computation graph data structure constructed in step 5 to find the shortest path from the condition node to the node of the unknown quantity.
[0149] During the search process, the relevant attributes of the current node are calculated by combining the predecessor nodes. At the same time, it is determined whether the values are compatible. If not, unit conversion is required (i.e., the unit conversion module in the graph), and then the relevant node attributes in the graph are updated. (For example, the speed value "5 m / s" cannot be directly multiplied with the time value "10 minutes". The time needs to be converted to "600 seconds" or the speed needs to be converted to "m / min" for the two values to be compatible.)
[0150] Step 7: Step Generation
[0151] For step 6, the shortest path solution and corresponding unit conversions are outlined, forming the most basic calculation steps. Then, based on these calculation steps, corresponding guidance steps will be generated. See [link / reference]. Figure 8 As shown.
[0152] Step 8: Post-normalization processing
[0153] The format of the final calculation process is standardized.
[0154] See Figure 9As shown, this embodiment also provides an automatic problem-solving device for application test questions, including:
[0155] The question comprehension module performs question comprehension on application test questions, extracting entity attribute information and entity relationship information from the application test questions;
[0156] The problem-solving reasoning module introduces basic knowledge points and entity attribute information and entity relationship information to jointly construct a computational graph data structure. Based on the computational graph data structure, a search is performed to find the solution path from the condition node to the unknown quantity node of the application problem. Based on the solution path, the solution steps of the application problem are generated.
[0157] The question comprehension module in this embodiment includes:
[0158] The problem decomposition module breaks down the problem that ultimately needs to be solved into smaller parts.
[0159] The entity attribute extraction module extracts the known conditions and unknowns from the problem.
[0160] Identify the condition nodes and question nodes in the application test questions; extract entity attribute information from the application test questions, including extracting entity attribute information from the condition nodes and question nodes in the application test questions respectively.
[0161] The entity attribute information in the condition node includes condition name, condition attribute category, and condition attribute value. The condition name includes the object name extracted from the condition node, the condition attribute category includes the category to which the object in the condition node belongs, and the condition attribute value is the attribute parameter and parameter value extracted from the condition node corresponding to each condition attribute category.
[0162] The entity attribute information of the problem node includes the problem name, the problem attribute category, and the problem attribute value. The problem name includes the name of the object extracted from the problem node, the problem attribute category includes the category to which the object extracted from the problem node belongs, and the problem attribute value is the attribute parameter to be determined and the parameter value of the object extracted from the problem node.
[0163] Optionally, the entity attribute information of the condition node is stored as {name: [], attribute category: [], attribute value: [{name:, value:, unit:}, {name:, value:, unit:}]};
[0164] Optionally, the entity attribute information of the problem node is stored as {name: [], attribute category: [], attribute value: [{name:, value: unknown, unit:}]}.
[0165] The entity relationship extraction module, when it is determined that the application test question includes multiple question nodes, performs question splitting on the application test question, and extracts the question nodes corresponding to each question and the relationship information between each question node.
[0166] The entity relationship extraction module extracts entity relationship information from the application test questions, including:
[0167] The entity relationship extraction module determines whether the relationship between the problem nodes in the application test question is a sequential relationship based on the relationship information between the problem nodes obtained from the understanding of the question; if so, it extracts the entity relationship information between the problem nodes based on the entity attribute information of each problem node.
[0168] The problem-solving reasoning module in this embodiment includes:
[0169] The graph construction module introduces basic knowledge points and entity attribute information and entity relationship information to jointly construct a computational graph data structure.
[0170] The graph construction module determines the basic knowledge points that can be introduced based on the entity attribute information and entity relationship information of the application test questions. Based on the basic knowledge points, the graph construction module constructs the node attributes and edge attributes of the computational graph data structure from the test question attribute information of each condition node, the test question attribute information of each question node, and the test question relationship information of each question node.
[0171] The reasoning and calculation module uses a computation graph data structure to iterate until the problem node is solved successfully or the number of iterations reaches the upper limit. If the solution is successful, the path of the iterative process is used to form a node-path graph with the problem node as the starting point and the condition node as the ending point. The shortest path is searched in the node-path graph as the solution path for the application problem from the condition node to the node to be solved.
[0172] The step generation module generates the most basic calculation steps based on the problem-solving path and the corresponding unit conversions; then, based on the calculation steps, it generates corresponding guidance steps.
[0173] The unit conversion module determines whether the values are compatible. If not, it performs unit conversion and then updates the relevant node attributes in the computation graph data structure.
[0174] Furthermore, the automatic problem-solving device for application test questions described in this embodiment also includes:
[0175] The text preprocessing module preprocesses the acquired application test text into the input format of the question comprehension module;
[0176] The post-processing normalization module standardizes the format of the final calculation process.
[0177] The automatic problem-solving device in this embodiment targets application problems including, but not limited to, mathematical application problems and physics application problems. The automatic problem-solving method in this embodiment addresses scenarios where there are no resources in the question bank for producing and grading application problems. It iteratively and automatically summarizes the problem-solving methods and thought processes for application problems in teaching, and automatically solves and provides thought process guidance for application problems for which there are no resources in the question bank. The automatic problem-solving method in this embodiment can solve the following problems:
[0178] 1) When there are no resources in the question bank during the photo search and grading process, the question will be automatically answered and the answer will be provided.
[0179] 2) Reduce costs and improve efficiency for manual production of our own question bank; some questions can be answered without human intervention.
[0180] In addition, the automatic problem-solving device in this embodiment is designed for scenarios where there are no resources for generating test papers using application questions. By changing the wording, entities, etc., it can generate questions on its own, such as questions that are not found in the question bank, for ability testing, etc.
[0181] This embodiment also provides a computer-readable storage medium storing a computer-executable program, which, when executed, implements the aforementioned method for automatically solving application test questions.
[0182] The computer-readable storage medium described in this embodiment may include data signals propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable medium may send, propagate, or transmit programs for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.
[0183] This embodiment also provides an electronic device, including a processor and a memory, wherein the memory is used to store a computer-executable program, and when the computer program is executed by the processor, the processor executes the automatic problem-solving method for an application test question.
[0184] The electronic device is manifested in the form of a general-purpose computing device. It may contain one or more processors that work collaboratively. This invention also does not preclude distributed processing, meaning that processors may be distributed across different physical devices. The electronic device of this invention is not limited to a single entity, but may also be the sum of multiple physical devices.
[0185] The memory stores a computer-executable program, typically machine-readable code. The computer-readable program can be executed by the processor to enable the electronic device to perform the method of the present invention, or at least some steps of the method.
[0186] The memory includes volatile memory, such as random access memory (RAM) and / or cache memory, and may also be non-volatile memory, such as read-only memory (ROM).
[0187] It should be understood that the electronic device of the present invention may also include elements or components not shown in the examples above. For example, some electronic devices also include display units such as a display screen, and some electronic devices also include human-computer interaction elements such as buttons and keyboards. Any electronic device capable of executing a computer-readable program in its memory to implement the method of the present invention or at least some steps of the method can be considered as an electronic device covered by the present invention.
[0188] From the above description of the embodiments, those skilled in the art will readily understand that the present invention can be implemented by hardware capable of executing specific computer programs, such as the system of the present invention, and the electronic processing unit, server, client, mobile phone, control unit, processor, etc. included in the system. The present invention can also be implemented by computer software that executes the methods of the present invention, for example, by control software executed by a microprocessor, electronic control unit, client, server, etc. However, it should be noted that the computer software executing the methods of the present invention is not limited to execution in one or a specific set of hardware entities; it can also be implemented in a distributed manner by unspecified hardware. For computer software, the software product can be stored in a computer-readable storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or distributed across a network, as long as it enables electronic devices to execute the methods according to the present invention.
[0189] The above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described herein. Although the present invention has been described in detail with reference to the above embodiments, the present invention is not limited to the specific embodiments described above. Therefore, any modifications or equivalent substitutions to the present invention, as well as all technical solutions and improvements that do not depart from the spirit and scope of the invention, are covered within the scope of the claims of the present invention.
Claims
1. An automatic problem-solving method for application test questions, characterized in that, include: For application test questions, the process involves preprocessing the acquired application test question text into an input format for question comprehension, identifying the condition nodes and question nodes in the application test questions, and extracting entity attribute information, entity relationship information between condition nodes and question nodes, and information between question nodes. Based on the extracted entity attribute information and entity relationship information, the solution path from the condition node to the question node for the application test question is determined; Generate application test question solution steps based on the described problem-solving path; Extracting entity attribute information from application questions includes extracting entity attribute information from condition nodes and question nodes in application questions, respectively. The entity attribute information in the condition node includes condition name, condition attribute category, and condition attribute value. The condition name includes the object name extracted from the condition node, the condition attribute category includes the category to which the object in the condition node belongs, and the condition attribute value is the attribute parameter and parameter value extracted from the condition node corresponding to each condition attribute category. The entity attribute information of the problem node includes the problem name, the problem attribute category, and the problem attribute value. The problem name includes the name of the object extracted from the problem node, the problem attribute category includes the category to which the object extracted from the problem node belongs, and the problem attribute value is the attribute parameter to be determined and the parameter value of the object extracted from the problem node. Extracting entity relationship information from application test questions includes: Based on the information about the relationships between the problem nodes obtained from the understanding of the question, determine whether there are logical operation relationships between the problem nodes in the application test question; If so, extract the entity relationship information between each problem node based on the entity attribute information of each problem node; The process of determining the problem-solving path from the condition node to the question node based on the extracted entity attribute information and entity relationship information includes: By introducing basic knowledge points related to the entity attribute information and entity relationship information, a computational graph data structure for solving each problem node is constructed; Searching using computational graph data structures reveals the solution path from condition nodes to question nodes in application problems, including: Based on a computational graph data structure, the loop continues until a problem node is successfully solved or the maximum number of loop iterations is reached; If the problem node is solved successfully, the path of the self-looping process is formed into a node-path graph with the problem node as the starting point and the condition node as the ending point. The shortest path is searched in the node-path graph to serve as the solution path for the application problem from the condition node to the unknown node.
2. The automatic problem-solving method for application test questions according to claim 1, characterized in that, The process of understanding application-based test questions includes: When it is determined that the application test question includes multiple question nodes, the application test question is split into question nodes corresponding to each question and the relationship information between each question node.
3. The automatic problem-solving method for application test questions according to claim 2, characterized in that, The entity attribute information of the condition node is stored as {name: [ ], attribute category: [ ], attribute value: [{name:, value:, unit:}, {name:, value:, unit:}]}; The entity attribute information of the problem node is stored as {name: [ ], attribute category: [ ], attribute value: [{name:, value: unknown, unit:}]}.
4. The automatic problem-solving method for application test questions according to claim 1, characterized in that, The process of introducing fundamental knowledge points related to entity attribute information and entity relationship information to construct a computational graph data structure for solving each problem node includes: The basic knowledge points that can be introduced are determined based on the entity attribute information and entity relationship information of the application test questions; Based on the basic knowledge points, construct a computational graph data structure by taking the question attribute information of each condition node, the question attribute information of each problem node, and the question relationship information of each problem node; The computational graph data structure includes condition node attribute information and edge attribute information. The node attribute information includes condition node attribute information, problem node attribute information, and the attribute information of the predecessor nodes that need to be solved in the middle of the problem node from the condition node. The edge attribute information includes all paths from the condition node attribute information to the problem node attribute information.
5. The automatic problem-solving method for application test questions according to claim 1, characterized in that, During the shortest path search process, the relevant attributes of the current node are calculated in conjunction with the predecessor node. At the same time, it is determined whether the values are computable. If not, unit conversion is required, and then the relevant node attribute information in the computation graph data structure is updated.
6. The automatic problem-solving method for application test questions according to claim 1, characterized in that, The steps for generating application test questions based on the problem-solving path include: Based on the problem-solving approach and the corresponding unit conversions, the most basic calculation steps are formed; Then, based on the calculation steps, the corresponding guidance steps are generated.
7. An automatic problem-solving device for application test questions, characterized in that, include: The question comprehension module is designed to understand application test questions. It preprocesses the acquired application test question text into the input format of the question comprehension module and extracts entity attribute information and entity relationship information from the application test questions. The problem-solving reasoning module introduces basic knowledge points and entity attribute information and entity relationship information to jointly construct a computational graph data structure. Based on the computational graph data structure, it performs a search to find the solution path of the application problem from the condition node to the node of the unknown quantity. Based on the solution path, it generates the solution steps of the application problem. Extracting entity attribute information from application questions includes extracting entity attribute information from condition nodes and question nodes in application questions, respectively. The entity attribute information in the condition node includes condition name, condition attribute category, and condition attribute value. The condition name includes the object name extracted from the condition node, the condition attribute category includes the category to which the object in the condition node belongs, and the condition attribute value is the attribute parameter and parameter value extracted from the condition node corresponding to each condition attribute category. The entity attribute information of the problem node includes the problem name, the problem attribute category, and the problem attribute value. The problem name includes the name of the object extracted from the problem node, the problem attribute category includes the category to which the object extracted from the problem node belongs, and the problem attribute value is the attribute parameter to be determined and the parameter value of the object extracted from the problem node. Extracting entity relationship information from application test questions includes: Based on the information about the relationships between the problem nodes obtained from the understanding of the question, determine whether there are logical operation relationships between the problem nodes in the application test question; If so, extract the entity relationship information between each problem node based on the entity attribute information of each problem node; The introduced basic knowledge points, along with the entity attribute information and entity relationship information, jointly construct a computational graph data structure. Based on this computational graph data structure, a search is performed to find the solution path from the condition nodes to the unsolved quantity nodes in the application problem, including: By introducing basic knowledge points related to the entity attribute information and entity relationship information, a computational graph data structure for solving each problem node is constructed; Searching using computational graph data structures reveals the solution path from condition nodes to question nodes in application problems, including: Based on a computational graph data structure, the loop continues until a problem node is successfully solved or the maximum number of loop iterations is reached; If the problem node is solved successfully, the path of the self-looping process is formed into a node-path graph with the problem node as the starting point and the condition node as the ending point. The shortest path is searched in the node-path graph to serve as the solution path for the application problem from the condition node to the unknown node.
8. An electronic device, comprising a processor and a memory, the memory being used to store a computer-executable program, characterized in that: When the computer executable program is executed by the processor, the processor executes the automatic problem-solving method for the application test questions as described in any one of claims 1-6.
9. A computer-readable storage medium storing a computer-executable program, characterized in that, When the computer executable program is executed, it implements the automatic problem-solving method for the application test questions as described in any one of claims 1-6.
Citation Information
Patent Citations
Automatic problem solving method and device, computer equipment and storage medium
CN112183048A
Multi-information-source intelligent education question and answer method based on multi-relational graph convolutional neural network
CN114168704A
