A method, system, electronic device, and storage medium for generating test questions that combine multiple knowledge points.

Abstract

This invention discloses a method for generating cross-knowledge point combined exam questions, comprising the following steps: decomposing each knowledge point data into atomic knowledge points; mapping each atomic knowledge point to a leaf node of an Abstract Syntax Tree (AST); calculating the comprehensive correlation between two leaf nodes; constructing composite nodes; determining the linguistic expression order of the corresponding leaf nodes based on the association logic labels; generating initial exam question information based on the linguistic expression order, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type; performing semantic verification on the initial exam question information; and outputting the initial exam question information that passes the verification to obtain the target exam question information. This invention uses an abstract syntax tree as a knowledge modeling and combination carrier to achieve the organic coupling of multiple independent knowledge points, ultimately generating comprehensive exam questions that cover multi-dimensional skills and conform to actual job logic. This invention also discloses a system, electronic device, and storage medium for implementing the above method.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, and in particular to a method, system, electronic device, and storage medium for generating test questions that combine multiple knowledge points. Background Technology

[0002] With the deep integration of the digital economy and the real economy, skills-based service platforms, serving as crucial hubs connecting skills providers (such as skilled workers and artisans) and demanders (such as households, SMEs, and community organizations), are gradually becoming key vehicles for promoting flexible employment, facilitating skills monetization, and improving the quality of public services. These platforms typically set entry thresholds, requiring registrants to pass specific skills assessments to verify their professional capabilities and ensure the quality and reliability of service provision. However, the current traditional model for generating and applying skills assessment questions not only has significant limitations in the scientific rigor of the assessments themselves but also suffers from a disconnect from the actual needs of platform-based services. This limits the reference value of the assessment results and makes it difficult to effectively support the core logic that "passing the assessment equates to possessing platform service capabilities."

[0003] Traditional skills test questions often break down a complete skills system into unrelated, fragmented knowledge points or operational segments, failing to reflect a candidate's comprehensive understanding and application of the inherent connections, causal logic, and application scenarios within the skills system. Furthermore, the loose combination logic of traditional test questions makes it difficult to assess a candidate's comprehensive adaptability and integrated application abilities in complex situations. Although some traditional assessments have introduced so-called "comprehensive questions" to cover more skills, their design is not closely integrated with the logic of actual operations, making it difficult to comprehensively assess the diverse competencies required by candidates in real platform service environments and failing to meet the precise assessment needs of skills-based service platforms for the comprehensive service capabilities of their members. Summary of the Invention

[0004] To address the aforementioned problems in the existing technology, this invention provides a method, system, electronic device, and storage medium for generating cross-knowledge point combined test questions. The technical problem to be solved by this invention is achieved through the following technical solution:

[0005] The first aspect of this invention provides a method for generating test questions that combine multiple knowledge points, comprising the following steps:

[0006] Each knowledge point data is decomposed into atomic knowledge points; wherein, each atomic knowledge point includes a triplet attribute of knowledge point identifier, knowledge point type, and knowledge point skill weight; wherein, the knowledge point identifier is associated with the content of the knowledge point;

[0007] Each of the atomic knowledge points is mapped to a leaf node of the AST, resulting in a set of AST leaf nodes;

[0008] The comprehensive correlation between the two leaf nodes is determined based on the job scenario information of the two leaf nodes, the knowledge point type, and the knowledge point skill weight.

[0009] When the comprehensive correlation degree satisfies the correlation closed-loop condition, the corresponding leaf node is constructed as a composite node, and the correlation logical label and the composite weight of any two leaf nodes of the composite node are determined; wherein, the correlation logical label includes: causal relationship, sequential relationship, dependency relationship or parallel relationship;

[0010] The order of language representations for the corresponding leaf nodes is determined based on the associated logical labels.

[0011] Based on the order of the language expression, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type, generate initial test question information;

[0012] The initial test question information is semantically validated, and the validated initial test question information is output to obtain the target test question information.

[0013] In one embodiment of the present invention, determining the comprehensive correlation degree of the two leaf nodes based on the job scenario information of the two leaf nodes, the knowledge point type, and the knowledge point skill weight includes:

[0014] The skill scenario overlap is calculated based on the intersection and union of the task scenario sets of the two leaf nodes;

[0015] The type complementarity is determined based on the preset complementary weights of the knowledge point types of the two leaf nodes;

[0016] The weight synergy is calculated based on the difference in the knowledge point skill weights between the two leaf nodes;

[0017] The overall correlation degree of the two leaf nodes is obtained by weighting the skill scenario overlap, type complementarity, and weight synergy.

[0018] In one embodiment of the present invention, the skill scene overlap The calculation formula is:

[0019]

[0020] in, Indicates the first leaf nodes A collection of work scenarios Indicates the first leaf nodes The corresponding set of work scenarios.

[0021] In one embodiment of the present invention, the formula for calculating the weighted synergy degree is:

[0022]

[0023] in, Indicates the first leaf nodes Knowledge point and skill weights Indicates the first leaf nodes The weight of knowledge points and skills.

[0024] In one embodiment of the present invention, the associated closed-loop condition is that the comprehensive correlation degree is greater than or equal to a preset correlation degree threshold;

[0025] The composite weight is the maximum value of the knowledge point skill weights of the leaf nodes in the composite node.

[0026] In one embodiment of the present invention, the semantic verification of the initial test question information includes:

[0027] Semantic fluency is determined based on the number of correct logical expressions in the initial test question information, the total number of logical expressions, and the scene matching degree.

[0028] Determine whether the semantic fluency is greater than or equal to the verification threshold;

[0029] If so, the verification passes.

[0030] In one embodiment of the present invention, the semantic fluency The calculation formula is:

[0031]

[0032] in, Indicates the number of correct logical statements. Indicates the total number of logical statements. Indicates the scene matching degree.

[0033] A second aspect of this invention provides a cross-knowledge point combined test question generation system, comprising:

[0034] The decomposition module is used to decompose each knowledge point data into atomic knowledge points; wherein, the atomic knowledge point includes a triplet attribute of knowledge point identifier, knowledge point type, and knowledge point skill weight; wherein, the knowledge point identifier is associated with the content of the knowledge point;

[0035] The mapping module is used to map each of the atomic knowledge points to a leaf node of the AST, thereby obtaining a set of AST leaf nodes;

[0036] The calculation module is used to determine the comprehensive correlation between the two leaf nodes based on the job scenario information of the two leaf nodes, the knowledge point type, and the knowledge point skill weight.

[0037] The construction module is used to construct the corresponding leaf node into a composite node when the comprehensive correlation degree meets the correlation closed loop condition, and to determine the correlation logical label and the composite weight of any two leaf nodes of the composite node; wherein, the correlation logical label includes: causal relationship, sequential relationship, dependency relationship or parallel relationship;

[0038] The determination module is used to determine the order of language representations of the corresponding leaf nodes based on the associated logical labels;

[0039] The generation module is used to generate initial test question information based on the order of the language expression, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type.

[0040] The verification module is used to perform semantic verification on the initial test question information. If the initial test question information passes the verification, the target test question information is output.

[0041] A third aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a method for generating cross-knowledge point combination test questions provided in the first aspect of the present invention.

[0042] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a method for generating cross-knowledge point combination test questions provided in the first aspect of the present invention.

[0043] The beneficial effects of this invention are:

[0044] This invention uses Abstract Syntax Trees (ASTs) as the carrier for knowledge modeling and combination. Through a complete process of "atomized decomposition of knowledge points → structured AST modeling → quantification of knowledge point correlations → cross-knowledge point AST fusion → semantic verification of exam questions → adaptive output," it achieves the organic coupling of multiple independent knowledge points, generating comprehensive questions that conform to the exam scenario (skill assessment with a mentor). It transforms scattered knowledge points into computable and combinable AST nodes, quantifies the correlation and rationality of knowledge point combinations through mathematical models, and solves the problems of "isolated knowledge points" and "loose combination logic" in traditional exam question generation. This ensures that comprehensive questions not only cover multi-dimensional skills but also conform to the logic of actual operations.

[0045] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0046] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0047] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0048] Figure 1 A flowchart illustrating a method for generating test questions that combine multiple knowledge points, provided in an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of a cross-knowledge point combined test question generation system provided in an embodiment of the present invention. Detailed Implementation

[0050] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0051] like Figure 1 As shown, the first aspect of this invention provides a method for generating test questions that combine multiple knowledge points, comprising the following steps:

[0052] Step 11: Decompose each knowledge point data into atomic knowledge points.

[0053] The atomic knowledge point includes a triplet attribute of knowledge point identifier, knowledge point type, and knowledge point skill weight; the knowledge point identifier is associated with the content of the knowledge point.

[0054] Step 12: Map each atomic knowledge point to a leaf node of the AST to obtain the set of AST leaf nodes.

[0055] Step 13: Determine the overall correlation between the two leaf nodes based on their job scenario information, knowledge point type, and knowledge point skill weight.

[0056] Step 14: When the overall correlation degree meets the correlation closed-loop condition, construct the corresponding leaf node into a composite node, and determine the correlation logical label and composite weight of any two leaf nodes of the composite node.

[0057] Among them, the logical association labels include: causal relationship, sequential relationship, dependency relationship or parallel relationship.

[0058] Step 15: Determine the order of language representations for the corresponding leaf nodes based on the associated logical labels.

[0059] Step 16: Generate initial test question information based on the order of language expression, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type.

[0060] Step 17: Perform semantic verification on the initial test question information, and output the initial test question information that passes the verification to obtain the target test question information.

[0061] In this embodiment, an Abstract Syntax Tree (AST) is used as the knowledge modeling and combination carrier. Through a complete process of "atomized decomposition of knowledge points → structured AST modeling → quantification of knowledge point correlations → cross-knowledge point AST fusion → semantic verification of exam questions → adaptive output," multiple independent knowledge points are organically coupled to generate comprehensive questions that conform to the exam scenario (skill assessment with a mentor). Scattered knowledge points are transformed into computable and combinable AST nodes. Mathematical models quantify the correlations and rationality of knowledge point combinations, solving the problems of "isolated knowledge points" and "loose combination logic" in traditional exam question generation. This ensures that comprehensive questions cover multi-dimensional skills and conform to actual operational logic.

[0062] Based on the first aspect of the present invention, the second aspect of the present invention provides a method for generating cross-knowledge point combined exam questions in further detail. The second aspect of the present invention provides a method for generating cross-knowledge point combined exam questions, including the following steps:

[0063] Step 21: Decompose each knowledge point data into atomic knowledge points.

[0064] The atomic knowledge point includes a triplet attribute of knowledge point identifier, knowledge point type, and knowledge point skill weight; the knowledge point identifier is associated with the content of the knowledge point.

[0065] In this step, each independent knowledge point is broken down into "atomic knowledge points" to ensure that they are indivisible and have a clear skill orientation (to meet the dual requirements of practical operation and theory for the master's assessment).

[0066] Atomic knowledge points: Let the set of atomic knowledge points be... Each atomic knowledge point Includes triplet properties:

[0067] This indicates a unique identifier for a knowledge point (e.g., "Refrigerant R22 characteristics" is K1, "Refueling gun calibration operation" is K2).

[0068] Indicates the type of knowledge point (an enumerated value: theoretical category) Practical Training / Calculation ,like ∈ , ∈ );

[0069] This indicates the weight of knowledge points and skills (with a preset value of 0-1, representing the importance of the knowledge point in the entry assessment, such as the knowledge points of core practical skills). =0.9, auxiliary theoretical knowledge points =0.3).

[0070] Specifically, all the individual knowledge points required for the mentor onboarding assessment are broken down into atomic units, with each atomic knowledge point... Assignment Triples .

[0071] Example:

[0072] Boiling point / freezing point characteristics of refrigerant R22 (theoretical)

[0073] Environmental characteristics of refrigerant R410A (theoretical)

[0074] Calculation of refrigerant charge based on air conditioner horsepower (calculation type)

[0075] Calibration of filling gun pressure (hands-on practice)

[0076] Leak detection procedures during the filling process (practical).

[0077] Step 22: Map each atomic knowledge point to a leaf node of the AST to obtain the set of AST leaf nodes.

[0078] In this step, each atomic knowledge point will be... The leaf nodes are mapped to the AST, and the attributes of the leaf nodes are... The triples correspond one-to-one, i.e., the leaf nodes of the AST. Output: Set of leaf nodes in the AST (Abstract Syntax Tree) The set of leaf nodes can also be represented as .

[0079] Step 23: Determine the overall correlation between the two leaf nodes based on their job scenario information, knowledge point type, and knowledge point skill weight.

[0080] In this step, the overall relevance is calculated through three dimensions: skill scenario overlap, type complementarity, and weight synergy. This ensures that the combination of leaf nodes not only conforms to the scenario logic but also takes into account the complementarity of theory, practice, and computation.

[0081] Specifically, step 23 includes steps 231-234:

[0082] Step 231: Calculate the skill scenario overlap based on the intersection and union of the task scenario sets of the two leaf nodes.

[0083] Skill scene overlap This characterizes whether two knowledge points belong to the same work scenario (e.g., "refrigerant filling calculation" and "refrigerant gun calibration" both belong to the "refrigerant filling scenario" and have a high degree of overlap). The skill scenario overlap is determined by the number of any two leaf nodes in the leaf node set. The calculation formula is:

[0084]

[0085] in, Indicates the first leaf nodes A collection of work scenarios Indicates the first leaf nodes The corresponding set of work scenarios.

[0086] A database of relationships between leaf nodes and job scenarios is pre-established, and the job scenarios of the corresponding nodes are extracted from this database.

[0087] Example, leaf node of leaf node of , , ,but .

[0088] Step 232: Determine the type complementarity based on the preset complementary weights of the knowledge point types of the two leaf nodes.

[0089] In this step, type complementarity This characterizes whether the types (theory / practice / calculation) of two knowledge points are complementary (entry assessment requires a combination of theory + practice + calculation; higher complementarity means greater combined value). The complementarity of the types of any two leaf nodes. The calculation formula is:

[0090]

[0091] in, This represents the preset complementary weights, where the preset complementary weight matrix is ​​shown in Table 1:

[0092] Table 1

[0093]

[0094] Example, and belong combination, , and belong combination,

[0095] Step 233: Calculate the weight synergy based on the difference between the knowledge point skill weights of the two leaf nodes.

[0096] Weight synergy characterizes whether the skill weights of two knowledge points are synergistic (the synergy is higher when core skills are combined with core skills). The formula for calculating the weight synergy of any two leaf nodes is:

[0097]

[0098] The smaller the weight difference, the higher the degree of cooperation; when the difference is at most 1, the degree of cooperation is 0.5 (to avoid complete repulsion). For example, , ,but .

[0099] Step 234: Perform a weighted calculation on the skill scenario overlap, type complementarity, and weight synergy to obtain the comprehensive correlation between the two leaf nodes.

[0100] In this step, the three sub-dimensions calculated in steps 231-233 are weighted and summed. The weight coefficients are based on the preset assessment focus (practical operation + calculation priority, therefore...). =0.3, =0.4, =0.3, which satisfies The formula for calculating the overall correlation degree is:

[0101]

[0102] Example 1: Calculation (Refill volume calculation + refill gun calibration)

[0103] ;

[0104] Example 2: Calculation (R22 Features + Filling Gun Calibration)

[0105] → 1 / 4 = 0.25

[0106] (that is) ),

[0107] ;

[0108] Example 3: Calculation (R22 features + leak detection)

[0109] .

[0110] Step 24: When the overall correlation degree meets the correlation closed-loop condition, construct the corresponding leaf node into a composite node, and determine the correlation logical label and composite weight of any two leaf nodes of the composite node.

[0111] Among them, the logical association labels include: causal relationship, sequential relationship, dependency relationship or parallel relationship.

[0112] In this step, knowledge point pairs that meet the criteria are selected based on relevance, and composite nodes are generated through AST node fusion, clarifying the logical association labels between nodes. .

[0113] First, based on the preset correlation threshold Candidate nodes are selected based on their overall relevance being greater than or equal to a preset relevance threshold. The two leaf nodes form a set of candidate nodes. In this embodiment, .

[0114]

[0115] in, Indicates satisfaction The set of candidate nodes. For example, .

[0116] Then, composite nodes are constructed from the candidate node set; in this embodiment, they are used... Each composite node represents a set of composite nodes. .

[0117] Specifically, the leaf nodes within a composite node must satisfy the condition that any two leaf nodes... The closed-loop condition is met, and each composite node contains at least 2 leaf nodes (atomic knowledge points).

[0118] Finally, assign association logical labels to any two leaf nodes in the composite node that meet the conditions. This means assigning association logic labels to the edges within a composite node and determining the composite weight of the composite node. The composite weight is the maximum value of the knowledge point skill weights of the leaf nodes in the composite node.

[0119] here, Based on the knowledge point type and scenario logic, These are enumerated values, including: causal relationships. Sequence Relationship Dependency and parallel relationship For example, "Refrigerant Characteristics → Charge Calculation" is... "Tool Calibration → Filling Operation" is .

[0120] Example: Building composite nodes (fusion) )

[0121] Verify closed loop:

[0122] ≥0.6,

[0123] ≥0.6, ≥0.6 satisfies closed-loop condition;

[0124] Assign edge labels: (Refrigerant characteristics → charge amount calculation, cause and effect relationship) (Refill volume calculation → refill gun calibration, dependency relationship) (Refrigerant characteristics and filling gun calibration are listed in parallel).

[0125] The resulting composite node attributes are: ,in, (The skill weight of a composite node is the maximum value of the leaf node).

[0126] In this embodiment, AST is used to structure atomic knowledge points, so that the "attributes, logical relationships, and application scenarios" of knowledge points can be parsed and combined by machines.

[0127] AST structure: Knowledge points for each atom In an AST (Abstract Syntax Tree), a "leaf node" corresponds to a node in the AST. Multiple leaf nodes are aggregated into a "composite node" (corresponding to the knowledge point combination unit in a comprehensive exam question) through "association logical nodes". The formal representation of the AST is as follows:

[0128]

[0129] Root : Root node (unique, representing the "set of knowledge points for comprehensive exam questions");

[0130] : Node set (Leaf nodes = atomic knowledge points) (Composite node = cross-knowledge point combination unit); Edge set, Logical labels for relationships between nodes (enumeration values: causal relationships) / Order Relationship / Dependency Parallel relationship For example, "Refrigerant Characteristics → Charge Calculation" is... "Tool Calibration → Filling Operation" is ).

[0131] Step 25: Determine the order of language representations of the corresponding leaf nodes based on the association logic labels of the composite nodes.

[0132] In this step, the order of language expression for knowledge points is based on the logical association labels within composite nodes: causal relationships ( According to "cause → effect", dependency relationship ( ) By “Prerequisite → Operation”, parallel relationships ( ) sorted in descending order by skill weight.

[0133] Step 26: Generate initial test question information based on the order of language expression, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type.

[0134] In this step, the content of the leaf node, which is the content of the knowledge point, can be obtained based on the identifier of the leaf node. The basic information corresponding to the knowledge point type includes formulas for calculation-type knowledge points (such as refrigerant charge calculation), steps for practical knowledge points (such as tool calibration), and characteristics for theoretical knowledge points (such as refrigerant parameters). The initial test question information in natural language is generated by combining the order of language expression with the content of the knowledge point and the basic information.

[0135] Preferably, a large language model is used to generate initial test question information in natural language.

[0136] Step 27: Perform semantic verification on the initial test question information, and output the initial test question information that passes the verification to obtain the target test question information.

[0137] Specifically, semantic fluency is determined based on the number of correct logical expressions in the initial test question information, the total number of logical expressions, and the scene matching degree. It is then determined whether the semantic fluency is greater than or equal to the verification threshold. If so, the verification passes; otherwise, adjustment suggestions are output (such as adding related logical words or adjusting the order of expressions).

[0138] semantic fluency The calculation formula is:

[0139]

[0140] in, Indicates the number of correct logical statements. Indicates the total number of logical statements. This indicates the scenario matching degree. The scenario matching degree is the degree of fit between the exam questions and the actual homework scenarios. It is labeled by industry experts and has a value of 0-1.

[0141] For example, if the semantic fluency of the initial test question is ≥0.8, the validation passes, and the validated initial test question information is directly output as the target test question information.

[0142] Preferably, a large language model is used for semantic verification.

[0143] Example, based on Generate exam questions:

[0144] Order of presentation: (R22 feature) → (Calculation of injection volume) → (Filling gun calibration);

[0145] The exam question reads: "Given that the boiling point of refrigerant R22 is -40.8℃ and the freezing point is -160℃, the formula for calculating the refrigerant charge amount for a 1.5 horsepower air conditioner is..." m =0.12× P +0.3 (where P is the air conditioner's horsepower): (1) Calculate the amount of R22 to be added to the air conditioner; (2) Briefly describe the pressure calibration steps for the filling gun before adding; (3) Explain the influence of the characteristics of R22 on the filling operation.

[0146] Validation: Number of correct logical statements = 3, Total number of logical statements = 3, Scenario matching degree = 0.95

[0147] The verification passed.

[0148] In this embodiment, for different technical service content, the skills service and life service platform generates corresponding test questions according to the above steps. When technical service providers join the skills service and life service platform, they are assessed through these test questions to see if they meet the joining standards.

[0149] In one feasible implementation, steps A1-A3 are performed after step 24 and before step 25 above:

[0150] Step A1: Obtain the pretest scores for each knowledge point type of the registered personnel.

[0151] In this step, new members can take a pre-test (on the new member platform or other platforms). The pre-test questions are divided into three categories: theoretical, practical, and calculation. After the pre-test is completed, the score rate for each type of question is recorded.

[0152] Step A2: Calculate the percentage of each type of pre-test score in the re-examination based on the pre-test scores.

[0153] In this step, the percentage of marks is calculated based on the score rate for each type of question. Specifically, the formula for calculating the percentage of marks for each type of question in the re-examination is as follows:

[0154]

[0155]

[0156]

[0157] in, Indicates the theoretical score rate. This indicates the score rate for practical skills. This indicates the calculated class score rate. This indicates the percentage of candidates taking the re-examination for theoretical subjects. This indicates the percentage of practical skills-based interviews. This indicates the percentage of students taking the retest in computational subjects.

[0158] Here, the lower the score rate, the higher the weight of the retest, which can increase the weight in subsequent tests and exams, thus achieving adaptive assessment to identify and fill gaps in knowledge.

[0159] Step A3: Determine the target composite node in the composite node set based on the maximum percentage of type composite nodes.

[0160] In this step, the composite nodes with a high proportion of knowledge point types corresponding to the maximum percentage of the retest are selected as target nodes.

[0161] Example, theoretical score rate =0.6, practical operation score rate =0.7, calculate the score rate =0.5, then the proportion of the re-examination is 0.5. =(1-0.6) / (0.4+0.3+0.5)=0.33, the proportion =0.3 / 1.2=0.25, percentage =0.5 / 1.2=0.42, which shows that the score rate of calculation-related questions is relatively low. Therefore, calculation-related questions account for the highest proportion of the retest. In the future, we need to select the composite leaf nodes with a high proportion of calculation-related knowledge points to generate the final test questions in order to achieve adaptive assessment that fills in the gaps in knowledge.

[0162] Specifically, in a composite node with multiple leaf nodes, the ratio of the knowledge point skill weight of the leaf node to the sum of the knowledge point skill weights of all leaf nodes is calculated. t c Example, composite node middle, , , Calculate the leaf nodes of the class The knowledge point skill weight is 0.8. Dividing 0.8 by (0.7 + 0.8 + 0.9) yields the weight percentage of the leaf node in the composite node calculation class. t c1 Here, if there are multiple leaf nodes of the computational type in the composite node, the knowledge point skill weights are added together and then compared with the sum of the knowledge point skill weights of all leaf nodes.

[0163] After all composite nodes have been calculated, the leaf nodes of the composite node calculation class are sorted by weight percentage from highest to lowest. Then, the difference value is calculated for each pair of nodes. The difference value is calculated as the larger of the two weight percentages. t cx With smaller value t cn The difference between the smaller values t cn In comparison, the difference value is obtained. For example, the formula for calculating the difference value is: ( t cx - t cn ) / t cn Then, the difference value and the difference threshold. If the difference is greater than 100%, then the comparison is made. If the composite node has the larger weight ratio of the leaf node among the two composite nodes, then the composite node with the larger weight ratio of the leaf node is the target node, and then the next set of difference values ​​is calculated.

[0164] If the difference value is less than or equal to Then compare the composite node skill weights of the corresponding two composite leaf nodes. ,Pick Larger composite nodes are chosen as target nodes because their skill weights essentially extract the importance of the "most core skills" covered by the node, directly reflecting the extent to which the corresponding exam questions can verify the core competencies of the applicants. When the weight proportions of leaf nodes in the calculation category are not significantly different, composite nodes with larger skill weights are prioritized. Here, if... If they are equal, the composite node with the larger weight of the leaf node in the composite node calculation class is taken as the target node, and then the next set of difference values ​​is calculated.

[0165] After the calculation in step A3 is completed, all target nodes are obtained. Then, steps 25-27 are executed for the target nodes.

[0166] For example, since calculation-related knowledge points account for the highest proportion in the retest, and there are a large number of calculation-related knowledge points in the target leaf nodes, the final generated target test questions will have a higher proportion of calculation-related knowledge points, so as to achieve adaptive assessment for identifying and filling gaps in knowledge.

[0167] like Figure 2 As shown, a third aspect of the present invention provides a cross-knowledge point combined test question generation system, comprising:

[0168] The decomposition module 31 is used to decompose each knowledge point data into atomic knowledge points; wherein, the atomic knowledge point includes a triplet attribute of knowledge point identifier, knowledge point type and knowledge point skill weight; wherein, the knowledge point identifier is associated with the content of the knowledge point;

[0169] Mapping module 32 is used to map each atomic knowledge point to a leaf node of the AST, thereby obtaining a set of AST leaf nodes;

[0170] The calculation module 33 is used to determine the comprehensive correlation between two leaf nodes based on the job scenario information, knowledge point type, and knowledge point skill weight of the two leaf nodes.

[0171] The construction module 34 is used to construct the corresponding leaf node into a composite node when the comprehensive correlation degree meets the correlation closed loop condition, and to determine the correlation logical label and composite weight of any two leaf nodes of the composite node; wherein, the correlation logical label includes: causal relationship, sequential relationship, dependency relationship or parallel relationship;

[0172] Module 35 is used to determine the order of language representations of the corresponding leaf nodes based on the associated logical labels.

[0173] The generation module 36 is used to generate initial test question information based on the order of language expression, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type.

[0174] The verification module 37 is used to perform semantic verification on the initial test question information. The initial test question information that passes the verification is output to obtain the target test question information.

[0175] In one embodiment of the present invention, the comprehensive correlation between two leaf nodes is determined based on the job scenario information, knowledge point type, and knowledge point skill weight of the two leaf nodes, including:

[0176] The skill scenario overlap is calculated based on the intersection and union of the task scenario sets of the two leaf nodes;

[0177] The type complementarity is determined based on the preset complementary weights of the knowledge point types of the two leaf nodes;

[0178] The weight synergy is calculated based on the difference in the knowledge point skill weights between the two leaf nodes;

[0179] We calculate the comprehensive correlation between the two leaf nodes by weighting the overlap of skill scenarios, the complementarity of types, and the synergy of weights.

[0180] In one embodiment of the present invention, skill scenario overlap The calculation formula is:

[0181]

[0182] in, Indicates the first leaf nodes A collection of work scenarios Indicates the first leaf nodes The corresponding set of work scenarios.

[0183] In one embodiment of the present invention, the formula for calculating the weight synergy degree is:

[0184]

[0185] in, Indicates the first leaf nodes Knowledge point and skill weights Indicates the first leaf nodes The weight of knowledge points and skills.

[0186] In one embodiment of the present invention, the associated closed-loop condition is that the comprehensive correlation degree is greater than or equal to a preset correlation degree threshold.

[0187] The composite weight is the maximum value of the knowledge point skill weights of the leaf nodes in the composite node.

[0188] In one embodiment of the present invention, semantic verification of the initial test question information includes:

[0189] Semantic fluency is determined based on the number of correct logical expressions in the initial test question information, the total number of logical expressions, and the scene matching degree.

[0190] Determine whether the semantic fluency is greater than or equal to the verification threshold;

[0191] If so, the verification passes.

[0192] In one embodiment of the present invention, semantic fluency The calculation formula is:

[0193]

[0194] in, Indicates the number of correct logical statements. Indicates the total number of logical statements. Indicates the scene matching degree.

[0195] A fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the cross-knowledge point combination test question generation method provided by the present invention described above.

[0196] A fifth aspect of the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the cross-knowledge point combined test question generation method provided in the above-described embodiments of the present invention.

[0197] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage system located remotely from the aforementioned processor.

[0198] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware systems.

[0199] The method provided in this invention can be applied to electronic devices. Specifically, the electronic device can be a desktop computer, a portable computer, a smart mobile terminal, a server, etc. No limitation is made herein; any electronic device that can implement this invention falls within the protection scope of this invention.

[0200] For system / electronic device embodiments, since they are basically similar to method embodiments, the description is relatively simple, and relevant parts can be found in the description of the method embodiments.

[0201] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0202] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0203] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0204] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for generating test questions that combine multiple knowledge points, characterized in that, Includes the following steps: Each knowledge point data is decomposed into atomic knowledge points; wherein, each atomic knowledge point includes a triplet attribute of knowledge point identifier, knowledge point type, and knowledge point skill weight; wherein, the knowledge point identifier is associated with the content of the knowledge point; Each of the atomic knowledge points is mapped to a leaf node of the AST, resulting in a set of AST leaf nodes; The comprehensive correlation between the two leaf nodes is determined based on the job scenario information of the two leaf nodes, the knowledge point type, and the knowledge point skill weight. When the comprehensive correlation degree satisfies the correlation closed-loop condition, the corresponding leaf node is constructed as a composite node, and the correlation logical label and composite weight of any two leaf nodes of the composite node are determined; wherein, the correlation logical label includes: causal relationship, sequential relationship, dependency relationship or parallel relationship; The order of language representations for the corresponding leaf nodes is determined based on the associated logical labels. Based on the order of the language expression, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type, generate initial test question information; The initial test question information is semantically validated, and the validated initial test question information is output to obtain the target test question information; The step of determining the comprehensive correlation between the two leaf nodes based on the job scenario information of the two leaf nodes, the knowledge point type, and the knowledge point skill weight includes: The skill scenario overlap is calculated based on the intersection and union of the task scenario sets of the two leaf nodes; The type complementarity is determined based on the preset complementary weights of the knowledge point types of the two leaf nodes; The weight synergy is calculated based on the difference in the knowledge point skill weights between the two leaf nodes; The overall correlation degree of the two leaf nodes is obtained by weighting the skill scenario overlap, type complementarity, and weight synergy. The overlap of skill scenarios The calculation formula is: in, Indicates the first leaf nodes A collection of work scenarios Indicates the first leaf nodes The corresponding set of work scenarios; The weighted synergy The calculation formula is: in, Indicates the first leaf nodes Knowledge point and skill weights Indicates the first leaf nodes The weighting of knowledge points and skills; The type complementarity The calculation formula is: in, This represents the preset complementary weight of any combination of two knowledge point types. , Indicates the type of theory. Indicates the type of practical activity. Indicates the calculation type.

2. The method as described in claim 1, characterized in that, The associated closed-loop condition is that the comprehensive correlation degree is greater than or equal to the preset correlation degree threshold. The composite weight is the maximum value of the knowledge point skill weights of the leaf nodes in the composite node.

3. The method as described in claim 1, characterized in that, The semantic verification of the initial test question information includes: Semantic fluency is determined based on the number of correct logical expressions in the initial test question information, the total number of logical expressions, and the scene matching degree. Determine whether the semantic fluency is greater than or equal to the verification threshold; If so, the verification passes.

4. The method as described in claim 3, characterized in that, The semantic fluency The calculation formula is: in, Indicates the number of correct logical statements. Indicates the total number of logical statements. Indicates the scene matching degree.

5. A cross-knowledge point combined test question generation system, characterized in that, include: The decomposition module is used to decompose each knowledge point data into atomic knowledge points; wherein, the atomic knowledge point includes a triplet attribute of knowledge point identifier, knowledge point type, and knowledge point skill weight; wherein, the knowledge point identifier is associated with the content of the knowledge point; The mapping module is used to map each of the atomic knowledge points to a leaf node of the AST, thereby obtaining a set of AST leaf nodes; The calculation module is used to determine the comprehensive correlation between the two leaf nodes based on the job scenario information of the two leaf nodes, the knowledge point type, and the knowledge point skill weight. The construction module is used to construct the corresponding leaf node into a composite node when the comprehensive correlation degree meets the correlation closed loop condition, and to determine the correlation logical label and composite weight of any two leaf nodes of the composite node; wherein, the correlation logical label includes: causal relationship, sequential relationship, dependency relationship or parallel relationship; The determination module is used to determine the order of language representations of the corresponding leaf nodes based on the associated logical labels; The generation module is used to generate initial test question information based on the order of the language expression, the content of the corresponding leaf nodes, and the basic information corresponding to the knowledge point type. The verification module is used to perform semantic verification on the initial test question information. The initial test question information that passes the verification is output to obtain the target test question information. The step of determining the comprehensive correlation between the two leaf nodes based on the job scenario information of the two leaf nodes, the knowledge point type, and the knowledge point skill weight includes: The skill scenario overlap is calculated based on the intersection and union of the task scenario sets of the two leaf nodes; The type complementarity is determined based on the preset complementary weights of the knowledge point types of the two leaf nodes; The weight synergy is calculated based on the difference in the knowledge point skill weights between the two leaf nodes; The overall correlation degree of the two leaf nodes is obtained by weighting the skill scenario overlap, type complementarity, and weight synergy. The overlap of skill scenarios The calculation formula is: in, Indicates the first leaf nodes A collection of work scenarios Indicates the first leaf nodes The corresponding set of work scenarios; The weighted synergy The calculation formula is: in, Indicates the first leaf nodes Knowledge point and skill weights Indicates the first leaf nodes The weighting of knowledge points and skills; The type complementarity The calculation formula is: in, This represents the preset complementary weight of any combination of two knowledge point types. , Indicates the type of theory. Indicates the type of practical activity. Indicates the calculation type.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for generating cross-knowledge point combination test questions as described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for generating cross-knowledge point combination test questions as described in any one of claims 1 to 4.

Images