Teaching association semantic matching method and system for road and bridge professional concepts

Abstract

The invention discloses a road and bridge major concept-oriented teaching association semantic matching method and system, and the method comprises the following steps: obtaining a road and bridge major concept set and a road and bridge major teaching content set, and carrying out the preprocessing of the road and bridge major teaching content set; constructing an engineering cognitive object representation for each road and bridge professional concept; constructing a concept dependence directed relation set, and generating a road and bridge professional concept dependence directed graph; constructing a teaching evolution constraint mechanism; performing teaching stage labeling on the teaching content standardized text unit set; determining a stage reachable concept candidate set according to a teaching evolution constraint mechanism, and constructing a matching feature vector; based on the matching feature vector, asymmetric teaching semantic matching calculation is executed, and a directional matching result set is generated. According to the method, asymmetric semantic matching of road and bridge professional concepts and teaching contents is realized through engineering cognitive modeling and teaching evolution constraint, and the method has the advantages of accurate matching and high cognitive consistency.

Description

Technical Field

[0001] This invention relates to the field of semantic matching, and in particular to a teaching-related semantic matching method and system for road and bridge engineering concepts. Background Technology

[0002] With the application of information technology in engineering education, existing teaching methods are gradually incorporating technologies such as text processing, keyword matching, and semantic analysis to calculate the relationship between teaching content and professional concepts. This is used to support the organization of teaching resources, the recommendation of teaching content, and the planning of learning paths. In these technologies, teaching content is usually processed in text form, while professional concepts are mostly matched in the form of terms or keywords. The matching process mainly relies on text similarity, word frequency statistics, or vector representation results.

[0003] However, existing technologies generally treat road and bridge engineering concepts as symbolic units at the linguistic level, failing to characterize the engineering cognitive structure of these concepts and the dependencies between them during the calculation process. Furthermore, the teaching stage often treats these concepts as external process information and does not participate in the semantic matching calculation itself. This leads to a tendency for the matching results between teaching content and engineering concepts to be distorted in terms of the engineering cognitive sequence. It also makes it difficult to distinguish between the explanatory relationship of teaching content to concepts and the supporting relationship of concepts to teaching content, thus affecting the accuracy and usability of the semantic matching results. Summary of the Invention

[0004] One objective of this invention is to propose a teaching-related semantic matching method and system for road and bridge professional concepts. This invention achieves asymmetric semantic matching between road and bridge professional concepts and teaching content through engineering cognitive modeling and teaching evolution constraints, and has the advantages of accurate matching and high cognitive consistency.

[0005] A teaching-related semantic matching method for road and bridge engineering concepts according to an embodiment of the present invention includes the following steps: Obtain the set of concepts and teaching content for the road and bridge engineering major; perform preprocessing on the teaching content set for the road and bridge engineering major to generate a set of standardized text units for the teaching content. Generate a unique concept identifier for each road and bridge professional concept in the road and bridge professional concept set, and construct an engineering cognitive object representation for each road and bridge professional concept to generate a set of road and bridge professional concept engineering cognitive objects; Based on the set of cognitive objects of road and bridge professional concept engineering, construct a set of directed relational concepts and generate a directed graph of road and bridge professional concept dependencies; A teaching evolution constraint mechanism is constructed based on the directed graph of road and bridge professional concepts and the set of road and bridge professional concept engineering cognitive objects. Perform teaching stage annotation on the standardized text unit set of teaching content, mapping each standardized text unit of teaching content to a teaching stage in the teaching stage set, and generating a set of teaching content stage annotations; For each standardized text unit of teaching content in the stage annotation set of teaching content, a candidate set of stage reachable concepts is determined based on the teaching evolution constraint mechanism, and a matching feature vector is constructed for each road and bridge professional concept engineering cognitive object in the stage reachable concept candidate set; Based on the matching feature vector, asymmetric instructional semantic matching calculation is performed on each standardized text unit of teaching content and the corresponding stage reachable concept candidate set to generate a directional matching result set.

[0006] Optionally, the generation of the standardized text unit set of teaching content specifically includes: Obtain the set of teaching content for road and bridge engineering, generate text carrier identifiers for the teaching content, and store and index the teaching content based on the text carrier identifiers; Perform character standardization on each piece of road and bridge engineering teaching content to generate character-standardized road and bridge engineering teaching content; Segmenting the standardized road and bridge engineering teaching content into sentences and segments to generate a set of teaching content fragments; For each teaching content segment in the set of teaching content segments, perform term boundary recognition and bind the boundary information of road and bridge professional terms to the corresponding teaching content segment to generate a set of term boundary labeled segments; Noise fragment removal is performed on each term boundary annotation fragment in the term boundary annotation fragment set to generate a valid fragment set; Based on the text carrier identifier, the set of valid fragments is aggregated, and each valid fragment in the set of valid fragments is identified as a standardized text unit of teaching content, thereby generating a set of standardized text units of teaching content.

[0007] Optionally, the generation of the road and bridge professional conceptual engineering cognitive object set specifically includes: Obtain the set of road and bridge professional concepts, and perform concept name normalization on each road and bridge professional concept in the set to generate a set of normalized concept names; Generate a unique concept identifier for each normalized concept name in the set of normalized concept names, and generate a set of concept identifiers; Engineering dependency identification is performed on the standardized concept name corresponding to the concept identifier to generate engineering dependency tags. The engineering dependencies include prerequisite dependencies, structural composition relationships, stress mechanism dependencies, construction technology dependencies, and specification parameter constraint relationships. Perform engineering cognition level determination on the standardized concept name corresponding to the concept identifier, and generate engineering abstract level label; Perform engineering attribute coverage statistics and dependency complexity statistics on the standardized concept names corresponding to the concept identifiers to generate cognitive load tags; The project dependency tags, project abstraction level tags, and cognitive load tags are combined to form a project cognitive object representation. All concept identifiers in the road and bridge professional concept set are processed to generate a road and bridge professional concept project cognitive object set.

[0008] Optionally, the generation of the directed graph that relies on the road and bridge professional concepts specifically includes: Using concept identifiers as node identifiers, a set of concept nodes is constructed, and the engineering cognitive object representations corresponding to the concept identifiers are normalized and bound to the corresponding concept nodes to generate a set of concept nodes. Based on the engineering dependency tags bound to each concept node in the concept node set, a directed association relationship is constructed between concept nodes. For concept identifier pairs with dependencies, a directed relationship edge is established between the corresponding concept nodes. Based on the dependency direction indicated in the project dependency label, determine the direction of the directed relationship edge and bind the project dependency type identifier to the corresponding directed relationship edge; A convergence process is performed on the set of concept nodes and the set of directed relation edges to generate a set of concept-dependent directed relations, and a road and bridge professional concept-dependent directed graph is constructed based on the set of concept nodes and the set of concept-dependent directed relations. For each directed relation edge in the directed graph of road and bridge professional concept dependency, a relation strength assessment is performed based on the engineering dependency relation type identifier bound to the directed relation edge, the degree of difference in the engineering abstraction level label of the corresponding concept node, and the degree of consistency of engineering attributes in the engineering cognitive object representation of the corresponding concept node. The relation strength assessment result is used as the relation strength weight of the directed relation edge. The set of concept-dependent directed relations is subjected to validity screening based on relation strength weight. When the relation strength weight corresponding to a directed relation edge is less than the preset relation strength threshold, the directed relation edge is removed from the set of concept-dependent directed relations, and the road and bridge professional concept-dependent directed graph is updated simultaneously.

[0009] Optionally, the construction of the teaching evolution constraint mechanism specifically includes: Obtaining road and bridge engineering concepts relies on a directed graph and a set of engineering cognitive objects related to road and bridge engineering concepts; Construct a set of teaching stages, which includes the basic object stage, structural construction stage, mechanical performance stage, construction process stage, testing and evaluation stage, and specification constraint stage, and generate a unique stage identifier for each teaching stage in the set of teaching stages; Based on the engineering abstract level tags, a hierarchical mapping relationship is constructed, and the engineering abstract level tags are mapped to the teaching stage set to generate a set of engineering abstract level mapping relationships; For each engineering cognitive object representation in the set of conceptual engineering cognitive objects in road and bridge engineering, the earliest reachable teaching stage is determined based on the set of engineering abstraction level labels and engineering abstraction level mapping relationships, and then bound to the corresponding engineering cognitive object representation to generate the earliest reachable stage label set; Based on the cognitive load labels, cognitive load constraint adjustment processing is performed on the earliest reachable stage label set to generate the earliest reachable stage label set after cognitive load constraint adjustment. Based on the directed graph of road and bridge professional concept dependency, the earliest reachable stage annotation set after cognitive load constraint adjustment is processed by dependency constraint propagation, and the preceding concept node constraint convergence is performed on each concept node in the directed graph of road and bridge professional concept dependency. Based on the earliest reachable stage annotation set after dependency constraint propagation processing, construct a stage reachable concept set mapping relationship for each teaching stage in the teaching stage set; The mapping relationship between the set of teaching stages and the set of concepts reachable at each stage is used as a constraint mechanism for the evolution of teaching.

[0010] Optionally, the generation of the teaching content stage annotation set specifically includes: Obtain a standardized set of text units for teaching content and a set of teaching stages; For each teaching stage in the set of teaching stages, construct a set of stage keywords corresponding to that teaching stage; For each standardized text unit in the set of standardized text units for teaching content, count the number of matches between that standardized text unit and the set of stage keywords corresponding to each teaching stage, and generate stage match count values ​​for each teaching stage. Based on the stage matching count value, the teaching stage is determined for the standardized text unit of the teaching content, and the teaching stage with the largest stage matching count value is selected as the teaching stage corresponding to the standardized text unit of the teaching content. When multiple teaching stages have the same maximum stage matching count value, the teaching stage with the earlier stage order is selected from the multiple teaching stages according to the preset stage order in the teaching stage set as the teaching stage corresponding to the standardized text unit of the teaching content. Each standardized text unit of teaching content is bound to its corresponding teaching stage, generating a teaching content stage annotation record, which is then aggregated to form a teaching content stage annotation set.

[0011] Optionally, the construction of the matching feature vector specifically includes: For each standardized text unit of teaching content in the set of teaching content stage annotations, read its corresponding teaching stage, and read the set of stage reachable concepts corresponding to that teaching stage according to the teaching evolution constraint mechanism. The set of reachable concepts for a given stage is identified as the candidate set of reachable concepts for that stage, and then bound to the corresponding standardized text unit identifier of the teaching content to generate a record of the candidate set of reachable concepts for that stage. For each road and bridge professional concept engineering cognitive object in the candidate set of reachable concepts, its concept identifier and engineering cognitive object representation are read, and the corresponding standardized text units of teaching content are processed by segment extraction to generate a term fragment set and a context fragment set. Term consistency feature values ​​are generated based on the consistency relationship between the term fragment set and the standardized concept name corresponding to the concept identifier, and the term consistency feature values ​​are bound to the concept identifier. Contextual referential feature values ​​are generated based on the referential relationship between referential statements and concept identifiers in the context fragment set, and the contextual referential feature values ​​are bound to the concept identifiers; Based on the directed path relationship between concept identifiers and concept identifiers within the stage reachable concept set in the road and bridge professional concept dependency directed graph, concept dependency path feature values ​​are generated, and concept dependency path feature values ​​are bound to concept identifiers; Based on the consistency relationship between engineering attribute information in standardized text units of teaching content and engineering attribute labels in the representation of engineering cognitive objects, engineering attribute consistency feature values ​​are generated, and engineering attribute consistency feature values ​​are bound to concept identifiers. The terminology consistency feature value, context reference feature value, concept dependency path feature value, and engineering attribute consistency feature value corresponding to the same concept identifier are concatenated and normalized to form a matching feature vector.

[0012] Optionally, the generation of the directional matching result set specifically includes: Obtain the set of matching feature vectors and the candidate set of stage-reachable concepts; For each matching feature vector, perform asymmetric instructional semantic matching calculation, which includes support reachability matching calculation and interpretation referential matching calculation; Perform support reachability matching calculation, taking concept identifiers as the calculation object and concept dependency path feature values, engineering attribute consistency feature values, and terminology consistency feature values ​​as input features. Weight the input features to obtain the support reachability matching score, and bind the support reachability matching score to the corresponding concept identifier and the corresponding standardized text unit identifier of the teaching content. For the same matching feature vector, perform interpretation reference matching calculation. Take the standardized text unit identifier of teaching content as the calculation object, and take the terminology consistency feature value and context reference feature value as input features. The weighted result is used to obtain the interpretation reference matching score, and the interpretation reference matching score is bound to the corresponding concept identifier and the corresponding standardized text unit identifier of teaching content. For standardized text unit identifiers of the same concept and the same teaching content, the combination of support can achieve matching scores and explanatory reference matching scores to generate directional matching results; The matching score for support reachability is compared with the matching score for interpretation reference to generate a matching direction identifier. Specifically, when the matching score for support reachability is greater than the matching score for interpretation reference, the matching direction identifier is determined as the support reachability matching direction; when the matching score for interpretation reference is greater than the matching score for support reachability, the matching direction identifier is determined as the interpretation reference matching direction; when the matching score for support reachability is equal to the matching score for interpretation reference, the matching direction identifier is determined as the bidirectional matching direction, and the matching direction identifier is bound to the directional matching result record. Perform aggregation processing on all directional matching results to generate a set of directional matching results.

[0013] Optionally, the directional matching results include standardized text unit identifiers of teaching content, concept identifiers, support reachability matching scores, and explanatory referential matching scores.

[0014] A teaching-related semantic matching system for road and bridge engineering concepts, according to an embodiment of the present invention, includes: The teaching content preprocessing module is used to obtain the set of teaching content for road and bridge engineering and perform preprocessing. The Engineering Cognitive Object Construction Module is used to construct representations of engineering cognitive objects and generate a set of engineering cognitive objects for road and bridge engineering concepts. The concept-dependent directed graph construction module is used to construct a set of concept-dependent directed relations based on the set of cognitive objects of road and bridge professional concept engineering, and generate a concept-dependent directed graph of road and bridge professional. The teaching evolution constraint construction module is used to construct a teaching evolution constraint mechanism based on the directed graph of road and bridge professional concept dependency and the set of road and bridge professional concept engineering cognitive objects; The teaching stage annotation module is used to perform teaching stage annotation on a standardized set of text units of teaching content, and generate a set of teaching content stage annotations. The stage-reachable concept filtering module is used to determine the candidate set of stage-reachable concepts for each standardized text unit of teaching content in the stage-annotated set of teaching content, based on the teaching evolution constraint mechanism, and to construct a matching feature vector. The asymmetric instructional semantic matching module is used to perform asymmetric instructional semantic matching calculations on standardized text units of teaching content and corresponding stage reachable concept candidate sets based on matching feature vectors, generating a set of directional matching results.

[0015] The beneficial effects of this invention are: This invention introduces an engineering cognitive modeling and teaching evolution constraint mechanism for road and bridge engineering into a computer system. This enables refined, structured, and directional calculation of the relationship between teaching content and professional concepts. Compared with existing processing methods that rely solely on text similarity or keyword matching, this invention improves the rationality and consistency of teaching semantic matching results at the engineering cognitive level. By modeling road and bridge engineering concepts as engineering cognitive objects containing engineering dependencies, engineering abstraction levels, and cognitive load information, professional concepts no longer participate in matching calculations merely as linguistic symbols, but as computational objects with clear engineering semantic structures throughout the entire process. This avoids the mismatch problem between teaching content and concepts in the engineering logical order.

[0016] Meanwhile, this invention introduces the teaching stage as an inherent constraint condition for semantic computing. By constructing a teaching evolution constraint mechanism, it limits the reachability of concepts in different teaching stages, ensuring that teaching content can only participate in matching calculations with concepts that are reasonably reachable in terms of engineering cognition and teaching progress. This prevents higher-order concepts from being incorrectly associated in the initial teaching stage, fundamentally improving the consistency between the teaching semantic matching results and the actual teaching evolution process. Furthermore, by performing constraint propagation processing on engineering dependencies, the pre-dependencies between concepts are globally consistent in the teaching stage determination, ensuring that the order of concept introduction conforms to the internal logic of the engineering knowledge system.

[0017] Furthermore, this invention introduces an asymmetric instructional semantic matching computation mechanism to clearly distinguish between the explanatory referential relationship of instructional content to concepts and the supportive reach relationship of concepts to instructional content. This allows the matching results to not only reflect whether the two are related but also to characterize the direction of semantic action. Consequently, it provides more discriminative structured results for instructional resource organization, instructional path analysis, and instructional content optimization. By generating a set of directional matching results, the instructional system can accurately identify which concepts are the engineering foundation upon which the instructional content is based and which concepts are the objects being explained in the instructional content, thereby improving the usability and practical application value of instructional semantic analysis. Attached Figure Description

[0018] 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: Figure 1 The flowchart shows a teaching-related semantic matching method for road and bridge professional concepts proposed in this invention. Figure 2 This is a schematic diagram illustrating the construction of the teaching evolution constraint mechanism for a teaching-related semantic matching method for road and bridge professional concepts proposed in this invention. Figure 3 This is a schematic diagram of asymmetric teaching semantic matching, which is a teaching-related semantic matching method for road and bridge professional concepts proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3 A teaching-related semantic matching method for road and bridge engineering concepts includes the following steps: Obtain the set of concepts and teaching content for the road and bridge engineering major; perform preprocessing on the teaching content set for the road and bridge engineering major to generate a set of standardized text units for the teaching content. Generate a unique concept identifier for each road and bridge professional concept in the road and bridge professional concept set, and construct an engineering cognitive object representation for each road and bridge professional concept to generate a set of road and bridge professional concept engineering cognitive objects; Based on the set of cognitive objects of road and bridge professional concept engineering, construct a set of directed relational concepts and generate a directed graph of road and bridge professional concept dependencies; A teaching evolution constraint mechanism is constructed based on the directed graph of road and bridge professional concepts and the set of road and bridge professional concept engineering cognitive objects. Perform teaching stage annotation on the standardized text unit set of teaching content, mapping each standardized text unit of teaching content to a teaching stage in the teaching stage set, and generating a set of teaching content stage annotations; For each standardized text unit of teaching content in the stage annotation set of teaching content, a candidate set of stage reachable concepts is determined based on the teaching evolution constraint mechanism, and a matching feature vector is constructed for each road and bridge professional concept engineering cognitive object in the stage reachable concept candidate set; Based on the matching feature vector, asymmetric instructional semantic matching calculation is performed on each standardized text unit of teaching content and the corresponding stage reachable concept candidate set to generate a directional matching result set.

[0021] In this embodiment, the generation of the standardized text unit set of teaching content specifically includes: The system acquires a set of teaching content for road and bridge engineering, generates text carrier identifiers for the teaching content, and stores and indexes the teaching content based on the text carrier identifiers. The set of teaching content for road and bridge engineering includes textbook chapter texts, courseware texts, lecture notes texts, and exercise solution texts. For each piece of road and bridge professional teaching content, character standardization is performed. The character standardization includes converting full-width characters to half-width characters, converting traditional Chinese characters to simplified Chinese characters, unifying English letters to lowercase characters, merging consecutive whitespace characters into a single whitespace character, removing illegal characters, and removing control characters, thereby generating road and bridge professional teaching content with standardized characters. Segmenting the standardized road and bridge engineering teaching content into sentences and segments to generate a set of teaching content fragments; For each teaching content segment in the set of teaching content segments, terminology boundary recognition is performed. The terminology boundary recognition process is used to identify the start and end positions of road and bridge professional terms in the teaching content segments, and bind the boundary information of road and bridge professional terms to the corresponding teaching content segments to generate a set of terminology boundary-annotated segments. Noise fragment removal is performed on each term boundary annotation fragment in the term boundary annotation fragment set to generate a valid fragment set; Based on the text carrier identifier, the set of valid fragments is aggregated, and each valid fragment in the set of valid fragments is identified as a standardized text unit of teaching content, thereby generating a set of standardized text units of teaching content.

[0022] In this embodiment, the generation of the road and bridge professional conceptual engineering cognitive object set specifically includes: Obtain a set of road and bridge professional concepts, and perform concept name standardization on each road and bridge professional concept in the set. The concept name standardization process includes merging synonymous concept names, expanding abbreviation forms, unifying symbolic expressions, and resolving naming conflicts, thereby generating a set of standardized concept names. The aforementioned road and bridge professional concepts are standardized professional knowledge used to characterize engineering objects, structural composition, stress mechanism, construction process, testing and evaluation, and regulatory constraints. Generate a unique concept identifier for each normalized concept name in the set of normalized concept names, and generate a set of concept identifiers, wherein each concept identifier in the set of concept identifiers corresponds one-to-one with a normalized concept name in the set of normalized concept names; Engineering dependency identification is performed on the standardized concept name corresponding to the concept identifier to generate engineering dependency tags. The engineering dependencies include prerequisite dependencies, structural composition relationships, stress mechanism dependencies, construction technology dependencies, and specification parameter constraint relationships. The generation of the engineering dependency tags specifically includes: obtaining the standardized concept name corresponding to the concept identifier, and constructing a candidate concept pair set based on other standardized concept names in the road and bridge professional concept set; for each candidate concept pair, performing engineering relationship determination based on the co-occurrence position relationship, syntactic dependency relationship, and explicit relationship indicator of the standardized concept name in the engineering textbook catalog structure, teaching syllabus description text, and professional terminology definition text; when a standardized concept name is explicitly described in the definition text as a prerequisite for another standardized concept name, the corresponding relationship is determined as a pre-dependency relationship; when a standardized concept name is described in the structural description text as a component unit or constituent element of another standardized concept name, the corresponding relationship is determined as a structural composition relationship; when a standardized concept name... When a standardized concept name is used in a mechanical analysis description text to explain the stress behavior or performance changes of another standardized concept name, the correspondence is determined to be a stress mechanism dependency; when a standardized concept name is used in a construction process description text to limit or trigger the construction steps corresponding to another standardized concept name, the correspondence is determined to be a construction technology dependency; when a standardized concept name is used in a specification clause or technical parameter description text to constrain the value range, applicable conditions, or design requirements of another standardized concept name, the correspondence is determined to be a specification parameter constraint; the preceding dependencies, structural composition relationships, stress mechanism dependencies, construction technology dependencies, and specification parameter constraint relationships identified for all candidate concept pairs will be summarized to form corresponding engineering dependency labels; Perform an engineering cognitive level determination on the standardized concept name corresponding to the concept identifier, and generate an engineering abstract level label. The engineering cognitive level includes object level, structure level, mechanism level, process level, and constraint level. The engineering abstract level label is used to identify the engineering cognitive level to which the concept identifier belongs. Perform engineering attribute coverage statistics and dependency complexity statistics on the standardized concept names corresponding to the concept identifiers to generate cognitive load tags; The generation of the cognitive load label specifically includes: for each concept identifier in the concept identifier set, obtaining a standardized concept name that corresponds one-to-one with the concept identifier; based on the definition text and associated description content of the standardized concept name in the road and bridge engineering knowledge system, determining whether the standardized concept name involves structural attributes, mechanical attributes, construction attributes, and standard attributes, and generating corresponding attribute coverage mark values ​​for structural attributes, mechanical attributes, construction attributes, and standard attributes respectively, and summarizing the attribute coverage mark values ​​to form engineering attribute coverage statistical results; based on the association records of the concept identifier in the concept dependency directed relation set, counting the number of dependencies corresponding to the concept identifier, generating a dependency quantity value, and counting the number of dependency relation types involved in the dependency, generating a dependency relation type quantity value, the dependency quantity value and the dependency relation type quantity value together constitute the dependency relation complexity statistical results; determining the engineering attribute coverage level value according to the engineering attribute coverage statistical results, and determining the dependency complexity value according to the dependency relation complexity statistical results; combining the engineering attribute coverage level value and the dependency complexity value to generate a cognitive load label that uniquely corresponds to the concept identifier, and binding the cognitive load label to the concept identifier; For each conceptual identifier, the corresponding engineering attribute coverage statistics are read. When the engineering attribute coverage statistics show that the conceptual identifier only involves structural attributes, the engineering attribute coverage level is determined to be the first level. When the engineering attribute coverage statistics show that the conceptual identifier involves both structural and mechanical attributes, the engineering attribute coverage level is determined to be the second level. When the engineering attribute coverage statistics show that the conceptual identifier involves structural, mechanical, and construction attributes, the engineering attribute coverage level is determined to be the third level. When the engineering attribute coverage statistics show that the conceptual identifier involves structural, mechanical, construction, and specification attributes, the engineering attribute coverage level is determined to be the fourth level. For each concept identifier, the dependency complexity statistics are retrieved. If the dependency complexity statistics show that the number of dependencies for the concept identifier is zero, the dependency complexity level is determined to be Level 1. If the dependency complexity statistics show that the number of dependencies for the concept identifier is greater than zero and the number of dependency types is one, the dependency complexity level is determined to be Level 2. If the dependency complexity statistics show that the number of dependency types for the concept identifier is two or three, the dependency complexity level is determined to be Level 3. If the dependency complexity statistics show that the number of dependency types for the concept identifier is no less than four, the dependency complexity level is determined to be Level 4. The project dependency tags, project abstraction level tags, and cognitive load tags are combined to form a project cognitive object representation. All concept identifiers in the road and bridge professional concept set are processed to generate a road and bridge professional concept project cognitive object set.

[0023] In this embodiment, the generation of the directed graph that relies on road and bridge engineering concepts specifically includes: Using concept identifiers as node identifiers, a set of concept nodes is constructed, and the engineering cognitive object representations corresponding to the concept identifiers are normalized and bound to the corresponding concept nodes to generate a set of concept nodes. Based on the engineering dependency tags bound to each concept node in the concept node set, a directed association relationship is constructed between concept nodes. For concept identifier pairs with dependencies, a directed relationship edge is established between the corresponding concept nodes. Based on the dependency direction indicated in the project dependency label, determine the direction of the directed relationship edge and bind the project dependency type identifier to the corresponding directed relationship edge; A convergence process is performed on the set of concept nodes and the set of directed relation edges to generate a set of concept-dependent directed relations, and a road and bridge professional concept-dependent directed graph is constructed based on the set of concept nodes and the set of concept-dependent directed relations. For each directed relation edge in the directed graph of road and bridge professional concept dependency, a relation strength assessment is performed based on the engineering dependency relation type identifier bound to the directed relation edge, the degree of difference in the engineering abstraction level label of the corresponding concept node, and the degree of consistency of engineering attributes in the engineering cognitive object representation of the corresponding concept node. The relation strength assessment result is used as the relation strength weight of the directed relation edge. The generation of the relationship strength weight specifically includes: taking the engineering dependency relationship type identifier bound to the directed relationship edge; searching for the corresponding relationship type weight value in a preset relationship type weight table based on the engineering dependency relationship type identifier; the relationship type weight table assigns uniquely determined relationship type weight values ​​for each of the following: pre-dependent relationships, structural composition relationships, stress mechanism dependencies, construction technology dependencies, and specification parameter constraint relationships; obtaining the relationship type weight value of the directed relationship edge; acquiring the starting concept node and target concept node corresponding to the directed relationship edge; reading the engineering abstract level label from the engineering cognitive object representation bound to the two concept nodes; calculating the level difference between the two engineering abstract level labels; and mapping the level difference. The hierarchy difference weight value is used as follows: Engineering attribute sets are read from the engineering cognitive object representations bound to the starting concept node and the target concept node corresponding to the directed relation edge. Attribute consistency calculation is performed on the two engineering attribute sets to count the number of identical attribute items in the two sets. Based on the ratio of the number of identical attribute items to the total number of attribute items, an engineering attribute consistency weight value is generated. The relation type weight value, hierarchy difference weight value, and engineering attribute consistency weight value are used as three independent relation strength components. The three relation strength components are weighted to generate the relation strength evaluation result corresponding to the directed relation edge. The relation strength evaluation result is bound to the corresponding directed relation edge as the relation strength weight of that directed relation edge. The engineering attributes include structural attributes, mechanical attributes, construction attributes, and specification attributes. Structural attributes are used to characterize the composition and connection relationship of the engineering object corresponding to the road and bridge professional concept. Mechanical attributes are used to characterize the stress characteristics and mechanical behavior of the engineering object corresponding to the road and bridge professional concept under load. Construction attributes are used to characterize the construction methods, construction sequence, and technological constraints of the engineering object corresponding to the road and bridge professional concept. Specification attributes are used to characterize the parameter constraints of the design specifications, construction specifications, and acceptance specifications that the engineering object corresponding to the road and bridge professional concept is subject to. The set of concept-dependent directed relations is subjected to validity screening based on relation strength weight. When the relation strength weight corresponding to a directed relation edge is less than the preset relation strength threshold, the directed relation edge is removed from the set of concept-dependent directed relations, and the road and bridge professional concept-dependent directed graph is updated simultaneously.

[0024] In this embodiment, the construction of the teaching evolution constraint mechanism specifically includes: Obtaining road and bridge engineering concepts relies on a directed graph and a set of engineering cognitive objects related to road and bridge engineering concepts; Construct a set of teaching stages, which includes the basic object stage, structural construction stage, mechanical performance stage, construction process stage, testing and evaluation stage, and specification constraint stage, and generate a unique stage identifier for each teaching stage in the set of teaching stages; Based on the engineering abstract level tags, a hierarchical mapping relationship is constructed, and the engineering abstract level tags are mapped to the teaching stage set to generate a set of engineering abstract level mapping relationships; For each engineering cognitive object representation in the set of conceptual engineering cognitive objects in road and bridge engineering, the earliest reachable teaching stage is determined based on the set of engineering abstraction level labels and engineering abstraction level mapping relationships, and then bound to the corresponding engineering cognitive object representation to generate the earliest reachable stage label set; The generation of the earliest reachable stage annotation set specifically includes: reading the engineering abstract level label bound to the engineering cognitive object representation, wherein the engineering abstract level label is used to uniquely indicate the engineering cognitive level to which the engineering cognitive object representation belongs; retrieving the mapping record matching the engineering abstract level label in the engineering abstract level mapping relationship set, wherein the mapping record explicitly gives the teaching stage set corresponding to the engineering abstract level label; selecting the teaching stage with the smallest sequence number from the teaching stage set according to the preset order of the teaching stage set, and determining it as the earliest reachable teaching stage of the engineering cognitive object representation; binding the determined earliest reachable teaching stage with the corresponding concept identifier to form the earliest reachable stage annotation; performing the above processing sequentially on all engineering cognitive object representations in the road and bridge professional concept engineering cognitive object set, and aggregating all the generated earliest reachable stage annotations to obtain the earliest reachable stage annotation set; Based on the cognitive load labels, cognitive load constraint adjustment processing is performed on the earliest reachable stage label set to generate the earliest reachable stage label set after cognitive load constraint adjustment. The specific steps of the cognitive load constraint adjustment process include: reading the cognitive load threshold corresponding to each teaching stage in the teaching stage set, where the cognitive load threshold represents the maximum cognitive load level allowed to be introduced in that teaching stage; for each engineering cognitive object representation in the earliest reachable stage label set, obtaining its bound earliest reachable teaching stage and corresponding cognitive load label; comparing the cognitive load label with the cognitive load threshold corresponding to the earliest reachable teaching stage to obtain a comparison result; when the comparison result indicates that the value of the cognitive load label is greater than the corresponding cognitive load threshold, adjusting the earliest reachable teaching stage of the engineering cognitive object representation to the next teaching stage in the teaching stage set, and updating the earliest reachable stage label; obtaining the corresponding cognitive load threshold again for the adjusted earliest reachable teaching stage and repeating the comparison process until the comparison result indicates that the value of the cognitive load label is not greater than the corresponding cognitive load threshold; rebinding the adjusted earliest reachable teaching stage with the corresponding concept identifier, and aggregating to generate the earliest reachable stage label set after cognitive load constraint adjustment; Based on the directed graph of road and bridge professional concept dependency, the earliest reachable stage annotation set after cognitive load constraint adjustment is processed by dependency constraint propagation, and the preceding concept node constraint convergence is performed on each concept node in the directed graph of road and bridge professional concept dependency. When performing dependency constraint propagation processing on the earliest reachable stage annotation set after cognitive load constraint adjustment, each concept node is processed sequentially according to the topological order of the concept nodes in the road and bridge professional concept dependency directed graph. For the current concept node, the earliest reachable teaching stage already bound in the engineering cognitive object representation corresponding to the concept node is read as the initial stage value of the current concept node. All directed relation edges pointing to the concept node in the road and bridge professional concept dependency directed graph are retrieved to obtain the corresponding set of predecessor concept nodes. For each predecessor concept node in the set of predecessor concept nodes, the earliest reachable teaching stage already determined in its engineering cognitive object representation is read to form a set of predecessor stage values. The set of predecessor stage values ​​is processed to perform stage maximum value aggregation processing to obtain the predecessor stage constraint value. The predecessor stage constraint value is compared with the initial stage value of the current concept node, and the teaching stage that is not earlier than the previous one is taken as the updated earliest reachable teaching stage. The updated earliest reachable teaching stage is then rebound to the engineering cognitive object representation corresponding to the current concept node to complete the dependency constraint propagation processing of the current concept node. This process continues until all concept nodes in the road and bridge professional concept dependency directed graph have been processed. Based on the earliest reachable stage annotation set after dependency constraint propagation processing, a stage reachable concept set mapping relationship is constructed for each teaching stage in the teaching stage set. Specifically, the teaching stage set is obtained, and each teaching stage is selected as the current teaching stage according to the preset order of the teaching stage set. For the current teaching stage, the earliest reachable stage annotation set after dependency constraint propagation processing is traversed, and the earliest reachable teaching stage bound to each engineering cognitive object representation is read one by one. The current teaching stage is compared with the read earliest reachable teaching stage. When the current teaching stage is not earlier than the earliest reachable teaching stage in the teaching stage set, the concept identifier of the corresponding engineering cognitive object representation is added to the stage reachable concept set of the current teaching stage, and a mapping record containing the current teaching stage identifier and the concept identifier is generated. When the current teaching stage is earlier than the earliest reachable teaching stage in the teaching stage set, the corresponding concept identifier is not added to the stage reachable concept set of the current teaching stage. The above processing is performed for each teaching stage in the teaching stage set in sequence, and the stage reachable concept set mapping relationship between the teaching stage set and each stage reachable concept set is generated. The mapping relationship between the set of teaching stages and the set of concepts reachable at each stage is used as a constraint mechanism for the evolution of teaching.

[0025] In this embodiment, the generation of the teaching content stage annotation set specifically includes: Obtain a standardized set of text units for teaching content and a set of teaching stages; For each teaching stage in the set of teaching stages, a set of stage keywords corresponding to that teaching stage is constructed. The set of stage keywords consists of road and bridge professional terms and commonly used expressions that characterize the teaching content of that teaching stage. For each standardized text unit in the set of standardized text units for teaching content, count the number of matches between that standardized text unit and the set of stage keywords corresponding to each teaching stage, and generate stage match count values ​​for each teaching stage. Based on the stage matching count value, the teaching stage is determined for the standardized text unit of the teaching content, and the teaching stage with the largest stage matching count value is selected as the teaching stage corresponding to the standardized text unit of the teaching content. When multiple teaching stages have the same maximum stage matching count value, the teaching stage with the earlier stage order is selected from the multiple teaching stages according to the preset stage order in the teaching stage set as the teaching stage corresponding to the standardized text unit of the teaching content. Each standardized text unit of teaching content is bound to its corresponding teaching stage, generating a teaching content stage annotation record, which is then aggregated to form a teaching content stage annotation set.

[0026] In this embodiment, the construction of the matching feature vector specifically includes: For each standardized text unit of teaching content in the set of teaching content stage annotations, read its corresponding teaching stage, and read the set of stage reachable concepts corresponding to that teaching stage according to the teaching evolution constraint mechanism. The set of reachable concepts for a given stage is identified as the candidate set of reachable concepts for that stage, and then bound to the corresponding standardized text unit identifier of the teaching content to generate a record of the candidate set of reachable concepts for that stage. For each road and bridge professional concept engineering cognitive object in the candidate set of reachable concepts, its concept identifier and engineering cognitive object representation are read, and the corresponding standardized text unit of teaching content is processed by segment extraction to generate a term fragment set and a context fragment set. The term fragment set is the set of road and bridge professional term fragments that are hit in the standardized text unit of teaching content, and the context fragment set is the set of adjacent sentence fragments surrounding the term fragment. Term consistency feature values ​​are generated based on the consistency relationship between the term fragment set and the standardized concept name corresponding to the concept identifier, and the term consistency feature values ​​are bound to the concept identifier. Extract a set of term fragments from the standardized text units of the teaching content; read the standardized concept name corresponding to the current concept identifier; compare each term fragment in the term fragment set with the standardized concept name one by one, count the number of term fragments that are completely consistent, and obtain the concept hit value; count the total number of term fragments in the term fragment set and obtain the total term hit value; calculate the ratio between the concept hit value and the total term hit value to generate a term consistency feature value; Contextual referential feature values ​​are generated based on the referential relationship between referential statements and concept identifiers in the context fragment set, and the contextual referential feature values ​​are bound to the concept identifiers; Extract a set of contextual fragments adjacent to terminology fragments from standardized text units of teaching content; retrieve referential statements from the contextual fragment set, the referential statements including fixed expressions used to explain, define, refer to, or interpret concepts; count the number of referential statements in the contextual fragment set that form a referential relationship with the current concept identifier to obtain a referential hit count value; count the total number of statement fragments in the contextual fragment set to obtain a total contextual fragment count value; calculate the ratio between the referential hit count value and the total contextual fragment count value to generate a contextual referential feature value; The referential statement refers to a statement in the teaching content that explicitly points the current text content to a specific road and bridge professional concept through the expression of explanation, definition, reference or interpretation. The characteristic of the statement is that it contains fixed expressions for explaining the meaning, composition, attribution or use of the concept, and can directly correspond the semantics of the statement to the road and bridge professional concept. Based on the directed path relationship between concept identifiers and concept identifiers within the stage reachable concept set in the road and bridge professional concept dependency directed graph, concept dependency path feature values ​​are generated, and concept dependency path feature values ​​are bound to concept identifiers; Read the set of reachable concepts corresponding to the current teaching stage; in the directed graph of concept dependency in the road and bridge engineering major, use the concept identifiers in the set of reachable concepts as the starting node set and the current concept identifier as the target node, and perform a directed path search; if there is at least one directed path from the starting node to the target node, count the number of dependency levels contained in the shortest directed path and generate a path length value; if there is no directed path, record the path length value as unreachable; generate a concept dependency path feature value based on the path length value, wherein the path length is inversely proportional to the concept dependency path feature value. When there is no directed path from the set of reachable concepts to the current concept identifier in the directed graph of concept dependency in the road and bridge engineering major, the path length value is set to infinity, and the concept dependency path feature value is set to 0. Based on the consistency relationship between engineering attribute information in standardized text units of teaching content and engineering attribute labels in the representation of engineering cognitive objects, engineering attribute consistency feature values ​​are generated, and engineering attribute consistency feature values ​​are bound to concept identifiers. The process involves: retrieving engineering attribute tags corresponding to the current concept identifier from the engineering cognitive object representation; extracting a set of engineering attribute information fragments from standardized text units of teaching content; comparing the set of engineering attribute information fragments with the set of engineering attribute tags one by one, counting the number of consistent engineering attribute types, and obtaining the attribute hit type value; counting the total number of types of engineering attribute tags in the set of engineering attribute tags, and obtaining the total attribute type value; and calculating the ratio between the attribute hit type value and the total attribute type value to generate an engineering attribute consistency feature value. The terminology consistency feature value, context reference feature value, concept dependency path feature value, and engineering attribute consistency feature value corresponding to the same concept identifier are concatenated and normalized to form a matching feature vector.

[0027] In this embodiment, the generation of the directional matching result set specifically includes: Obtain the set of matching feature vectors and the candidate set of stage-reachable concepts; For each matching feature vector, perform asymmetric instructional semantic matching calculation, which includes support reachability matching calculation and interpretation referential matching calculation; Perform support reachability matching calculation, taking concept identifiers as the calculation object and concept dependency path feature values, engineering attribute consistency feature values, and terminology consistency feature values ​​as input features. Weight the input features to obtain the support reachability matching score, and bind the support reachability matching score to the corresponding concept identifier and the corresponding standardized text unit identifier of the teaching content. For the same matching feature vector, perform interpretation reference matching calculation. Take the standardized text unit identifier of teaching content as the calculation object, and take the terminology consistency feature value and context reference feature value as input features. The weighted result is used to obtain the interpretation reference matching score, and the interpretation reference matching score is bound to the corresponding concept identifier and the corresponding standardized text unit identifier of teaching content. For standardized text unit identifiers of the same concept and the same teaching content, the combination of support can achieve matching scores and explanatory reference matching scores to generate directional matching results; The matching score for support reachability is compared with the matching score for interpretation reference to generate a matching direction identifier. Specifically, when the matching score for support reachability is greater than the matching score for interpretation reference, the matching direction identifier is determined as the support reachability matching direction; when the matching score for interpretation reference is greater than the matching score for support reachability, the matching direction identifier is determined as the interpretation reference matching direction; when the matching score for support reachability is equal to the matching score for interpretation reference, the matching direction identifier is determined as the bidirectional matching direction, and the matching direction identifier is bound to the directional matching result record. The matching direction identifiers are used to clarify the semantic direction of interaction between standardized text units of teaching content and road and bridge professional concepts. The support-reachable matching direction indicates that, under the constraints of engineering cognitive structure and dependency relationships, road and bridge professional concepts can serve as prerequisite or basic cognitive units, supporting and carrying the engineering knowledge involved in the standardized text units of teaching content. In other words, the engineering cognitive object of this concept satisfies the conditions for being invoked and relied upon by the teaching content at the teaching stage and engineering dependency level. The explanation-referential matching direction indicates that the standardized text units of teaching content directly explain, define, analyze, or exemplify road and bridge professional concepts at the semantic expression level. That is, the standardized text units themselves undertake the function of explaining and referring to this concept. The bidirectional matching direction indicates that road and bridge professional concepts not only support the teaching content at the engineering cognitive structure level, but also that the standardized text units of teaching content explain and refer to this concept at the semantic expression level. The two form a mutually corresponding and mutually reinforcing relationship at both the engineering cognitive and linguistic semantic levels. Perform aggregation processing on all directional matching results to generate a set of directional matching results.

[0028] In this embodiment, the directional matching result includes standardized text unit identifiers of teaching content, concept identifiers, support reachability matching scores, and explanatory referential matching scores.

[0029] A teaching-related semantic matching system for road and bridge engineering concepts includes: The teaching content preprocessing module is used to obtain the set of teaching content for road and bridge engineering and perform preprocessing. The Engineering Cognitive Object Construction Module is used to construct representations of engineering cognitive objects and generate a set of engineering cognitive objects for road and bridge engineering concepts. The concept-dependent directed graph construction module is used to construct a set of concept-dependent directed relations based on the set of cognitive objects of road and bridge professional concept engineering, and generate a concept-dependent directed graph of road and bridge professional. The teaching evolution constraint construction module is used to construct a teaching evolution constraint mechanism based on the directed graph of road and bridge professional concept dependency and the set of road and bridge professional concept engineering cognitive objects; The teaching stage annotation module is used to perform teaching stage annotation on a standardized set of text units of teaching content, and generate a set of teaching content stage annotations. The stage-reachable concept filtering module is used to determine the candidate set of stage-reachable concepts for each standardized text unit of teaching content in the stage-annotated set of teaching content, based on the teaching evolution constraint mechanism, and to construct a matching feature vector. The asymmetric instructional semantic matching module is used to perform asymmetric instructional semantic matching calculations on standardized text units of teaching content and corresponding stage reachable concept candidate sets based on matching feature vectors, generating a set of directional matching results.

[0030] Example 1: To verify the feasibility and effectiveness of the present invention in practical applications, it was applied to the teaching content management and teaching support analysis scenario of a road and bridge engineering major in a higher education institution. In this scenario, the teaching content comes from a wide range of sources, including digitized texts of printed textbooks, as well as courseware texts, lecture notes, and accompanying exercise solutions created by teachers. Over a long period of teaching, the teaching content has accumulated, and there are problems such as overlapping concepts, mixed engineering levels, and inconsistent teaching order among the content. This makes it difficult to accurately describe the relationship between the teaching content and road and bridge engineering concepts. Teachers mainly rely on manual experience when organizing teaching resources and adjusting teaching paths, which is inefficient and inconsistent.

[0031] In this scenario, existing road and bridge engineering teaching content is imported into the computer system. Character processing, text segmentation, and terminology recognition are performed on the teaching content to form a set of standardized text units with a clear structure. Subsequently, commonly used professional concepts in the field of road and bridge engineering are imported into the system. The concept names are standardized, and an engineering cognitive object representation containing engineering dependencies, engineering abstraction levels, and cognitive load information is constructed for each concept. Through analysis of the textbook's table of contents, teaching syllabus text, and professional terminology definition text, the system automatically identifies the relationships between concepts and constructs a directed graph of road and bridge engineering concept dependencies accordingly, so that the engineering logical relationships between professional concepts are uniformly expressed in the system.

[0032] In the teaching application process, the system sets up a teaching evolution constraint mechanism based on the engineering abstraction level and teaching stage, which clearly limits the scope of concepts that can be introduced in different teaching stages. When the teaching content is mapped to the corresponding teaching stage, the system automatically selects the set of concepts that are accessible in terms of engineering cognition within that stage, avoiding the participation of high-order concepts that do not yet have an engineering foundation in semantic matching calculation. On this basis, the system combines the usage of terms in the teaching content, contextual referential relationships, concept dependency paths, and engineering attribute consistency information to construct a matching feature vector and perform asymmetric teaching semantic matching calculation to evaluate the explanatory referential relationship of the teaching content to the concept and the supportive accessibility relationship of the concept to the teaching content.

[0033] By comparing the correlation of teaching content before and after the application of this invention, it can be found that after the introduction of this invention, the teaching correlation semantic results generated by the system are more in line with the internal logic of the road and bridge engineering knowledge system in terms of the engineering cognitive sequence. The teaching content is no longer associated with concepts that do not match the engineering stage. At the same time, the system can clearly distinguish whether the teaching content is explaining a certain professional concept or using a certain professional concept to expand the engineering description, thereby providing a more reliable basis for the organization of teaching content, the recommendation of teaching resources, and the analysis of teaching paths.

[0034] To verify the performance of this invention, it was compared with traditional text similarity matching methods. The comparison results are shown in Table 1.

[0035] Table 1. Performance Comparison of the Invention and Traditional Text Similarity Matching Methods

[0036] As can be seen from Table 1, traditional text similarity-based matching methods have significant limitations in the overall processing of the association between teaching content and road and bridge engineering concepts. The accuracy of concept matching is mainly limited by keyword co-occurrence and vocabulary similarity. When multiple engineering concepts or implicit engineering logic appear in the teaching content, misjudgments are likely to occur, resulting in a low accuracy rate. At the same time, since this method does not consider the order of engineering cognition and concept dependencies, the consistency rate of engineering cognition order is low, and the number of conflicts in the concept preconditions is high, reflecting that there are many unreasonable associations in the teaching content at the level of engineering logic.

[0037] In comparison, the method of this invention shows improvement in several key indicators. The improved concept matching accuracy indicates that by introducing engineering cognitive object representation and concept dependency directed graph, the system can comprehensively consider the engineering dependency structure and teaching stage constraints in the matching calculation, thereby effectively avoiding erroneous associations based solely on language similarity. In terms of engineering cognitive order consistency rate, the method of this invention has a significant advantage, indicating that the association results between teaching content and professional concepts are more in line with the inherent order of the road and bridge engineering knowledge system.

[0038] Regarding the mismatch rate in the teaching stage, traditional methods, because they do not take the teaching stage as a computational constraint, are prone to introducing higher-order concepts into lower-level teaching content, resulting in a higher mismatch rate. However, this invention, through a teaching evolution constraint mechanism, strictly limits the reach of concepts, thereby reducing mismatches. This improvement is directly reflected in the decrease in the number of conflicts in the concept's pre-relationship and the proportion of redundant associations in the teaching content, indicating that the system output is more focused and logically consistent.

[0039] Of particular note is the comparison of the accuracy of directional semantic discrimination. Traditional methods assume that semantic matching is symmetrical, which cannot distinguish whether the teaching content is explaining a concept or expanding on a concept, resulting in poor performance of this indicator. This invention uses asymmetric teaching semantic matching calculation to evaluate support reachability relations and explanatory referential relations separately, enabling the system to clearly identify the direction of semantic action, thereby improving the directional discrimination ability. This improvement directly reduces the proportion of manually corrected matching results and increases the stability score of teaching content association results, indicating that the system has higher usability and reliability in actual teaching assistance applications.

[0040] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A teaching-related semantic matching method for road and bridge engineering concepts, characterized in that, Includes the following steps: Obtain the set of concepts and teaching content for the road and bridge engineering major; perform preprocessing on the teaching content set for the road and bridge engineering major to generate a set of standardized text units for the teaching content. Generate a unique concept identifier for each road and bridge professional concept in the road and bridge professional concept set, and construct an engineering cognitive object representation for each road and bridge professional concept to generate a set of road and bridge professional concept engineering cognitive objects; Based on the set of cognitive objects of road and bridge professional concept engineering, construct a set of directed relational concepts and generate a directed graph of road and bridge professional concept dependencies; A teaching evolution constraint mechanism is constructed based on the directed graph of road and bridge professional concepts and the set of road and bridge professional concept engineering cognitive objects. Perform teaching stage annotation on the standardized text unit set of teaching content, mapping each standardized text unit of teaching content to a teaching stage in the teaching stage set, and generating a set of teaching content stage annotations; For each standardized text unit of teaching content in the stage annotation set of teaching content, a candidate set of stage reachable concepts is determined based on the teaching evolution constraint mechanism, and a matching feature vector is constructed for each road and bridge professional concept engineering cognitive object in the stage reachable concept candidate set; Based on the matching feature vector, asymmetric instructional semantic matching calculation is performed on each standardized text unit of teaching content and the corresponding stage reachable concept candidate set to generate a directional matching result set.

2. The teaching-related semantic matching method for road and bridge professional concepts according to claim 1, characterized in that, The generation of the standardized text unit set of the teaching content specifically includes: Obtain the set of teaching content for road and bridge engineering, generate text carrier identifiers for the teaching content, and store and index the teaching content based on the text carrier identifiers; Perform character standardization on each piece of road and bridge engineering teaching content to generate character-standardized road and bridge engineering teaching content; Segmenting the standardized road and bridge engineering teaching content into sentences and segments to generate a set of teaching content fragments; For each teaching content segment in the set of teaching content segments, perform term boundary recognition and bind the boundary information of road and bridge professional terms to the corresponding teaching content segment to generate a set of term boundary labeled segments; Noise fragment removal is performed on each term boundary annotation fragment in the term boundary annotation fragment set to generate a valid fragment set; Based on the text carrier identifier, the set of valid fragments is aggregated, and each valid fragment in the set of valid fragments is identified as a standardized text unit of teaching content, thereby generating a set of standardized text units of teaching content.

3. The teaching-related semantic matching method for road and bridge engineering concepts according to claim 1, characterized in that, The generation of the aforementioned set of conceptual engineering cognitive objects for road and bridge engineering specifically includes: Obtain the set of road and bridge professional concepts, and perform concept name normalization on each road and bridge professional concept in the set to generate a set of normalized concept names; Generate a unique concept identifier for each normalized concept name in the set of normalized concept names, and generate a set of concept identifiers; Engineering dependency identification is performed on the standardized concept name corresponding to the concept identifier to generate engineering dependency tags. The engineering dependencies include prerequisite dependencies, structural composition relationships, stress mechanism dependencies, construction technology dependencies, and specification parameter constraint relationships. Perform engineering cognition level determination on the standardized concept name corresponding to the concept identifier, and generate engineering abstract level label; Perform engineering attribute coverage statistics and dependency complexity statistics on the standardized concept names corresponding to the concept identifiers to generate cognitive load tags; The project dependency tags, project abstraction level tags, and cognitive load tags are combined to form a project cognitive object representation. All concept identifiers in the road and bridge professional concept set are processed to generate a road and bridge professional concept project cognitive object set.

4. The teaching-related semantic matching method for road and bridge professional concepts according to claim 1, characterized in that, The aforementioned road and bridge engineering concepts rely on the generation of directed graphs, specifically including: Using concept identifiers as node identifiers, a set of concept nodes is constructed, and the engineering cognitive object representations corresponding to the concept identifiers are normalized and bound to the corresponding concept nodes to generate a set of concept nodes. Based on the engineering dependency tags bound to each concept node in the concept node set, a directed association relationship is constructed between concept nodes. For concept identifier pairs with dependencies, a directed relationship edge is established between the corresponding concept nodes. Based on the dependency direction indicated in the project dependency label, determine the direction of the directed relationship edge and bind the project dependency type identifier to the corresponding directed relationship edge; A convergence process is performed on the set of concept nodes and the set of directed relation edges to generate a set of concept-dependent directed relations, and a road and bridge professional concept-dependent directed graph is constructed based on the set of concept nodes and the set of concept-dependent directed relations. For each directed relation edge in the directed graph of road and bridge professional concept dependency, a relation strength assessment is performed based on the engineering dependency relation type identifier bound to the directed relation edge, the degree of difference in the engineering abstraction level label of the corresponding concept node, and the degree of consistency of engineering attributes in the engineering cognitive object representation of the corresponding concept node. The relation strength assessment result is used as the relation strength weight of the directed relation edge. The set of concept-dependent directed relations is subjected to validity screening based on relation strength weight. When the relation strength weight corresponding to a directed relation edge is less than the preset relation strength threshold, the directed relation edge is removed from the set of concept-dependent directed relations, and the road and bridge professional concept-dependent directed graph is updated simultaneously.

5. The teaching-related semantic matching method for road and bridge professional concepts according to claim 1, characterized in that, The construction of the teaching evolution constraint mechanism specifically includes: Obtaining road and bridge engineering concepts relies on a directed graph and a set of engineering cognitive objects related to road and bridge engineering concepts; Construct a set of teaching stages, which includes the basic object stage, structural construction stage, mechanical performance stage, construction process stage, testing and evaluation stage, and specification constraint stage, and generate a unique stage identifier for each teaching stage in the set of teaching stages; Based on the engineering abstract level tags, a hierarchical mapping relationship is constructed, and the engineering abstract level tags are mapped to the teaching stage set to generate a set of engineering abstract level mapping relationships; For each engineering cognitive object representation in the set of conceptual engineering cognitive objects in road and bridge engineering, the earliest reachable teaching stage is determined based on the set of engineering abstraction level labels and engineering abstraction level mapping relationships, and then bound to the corresponding engineering cognitive object representation to generate the earliest reachable stage label set; Based on the cognitive load labels, cognitive load constraint adjustment processing is performed on the earliest reachable stage label set to generate the earliest reachable stage label set after cognitive load constraint adjustment. Based on the directed graph of road and bridge professional concept dependency, the earliest reachable stage annotation set after cognitive load constraint adjustment is processed by dependency constraint propagation, and the preceding concept node constraint convergence is performed on each concept node in the directed graph of road and bridge professional concept dependency. Based on the earliest reachable stage annotation set after dependency constraint propagation processing, construct a stage reachable concept set mapping relationship for each teaching stage in the teaching stage set; The mapping relationship between the set of teaching stages and the set of concepts reachable at each stage is used as a constraint mechanism for the evolution of teaching.

6. The teaching-related semantic matching method for road and bridge professional concepts according to claim 1, characterized in that, The generation of the teaching content stage annotation set specifically includes: Obtain a standardized set of text units for teaching content and a set of teaching stages; For each teaching stage in the set of teaching stages, construct a set of stage keywords corresponding to that teaching stage; For each standardized text unit in the set of standardized text units for teaching content, count the number of matches between that standardized text unit and the set of stage keywords corresponding to each teaching stage, and generate stage match count values ​​for each teaching stage. Based on the stage matching count value, the teaching stage is determined for the standardized text unit of the teaching content, and the teaching stage with the largest stage matching count value is selected as the teaching stage corresponding to the standardized text unit of the teaching content. When multiple teaching stages have the same maximum stage matching count value, the teaching stage with the earlier stage order is selected from the multiple teaching stages according to the preset stage order in the teaching stage set as the teaching stage corresponding to the standardized text unit of the teaching content. Each standardized text unit of teaching content is bound to its corresponding teaching stage, generating a teaching content stage annotation record, which is then aggregated to form a teaching content stage annotation set.

7. The teaching-related semantic matching method for road and bridge professional concepts according to claim 1, characterized in that, The construction of the matching feature vector specifically includes: For each standardized text unit of teaching content in the set of teaching content stage annotations, read its corresponding teaching stage, and read the set of stage reachable concepts corresponding to that teaching stage according to the teaching evolution constraint mechanism. The set of reachable concepts for a given stage is identified as the candidate set of reachable concepts for that stage, and then bound to the corresponding standardized text unit identifier of the teaching content to generate a record of the candidate set of reachable concepts for that stage. For each road and bridge professional concept engineering cognitive object in the candidate set of reachable concepts, its concept identifier and engineering cognitive object representation are read, and the corresponding standardized text units of teaching content are processed by segment extraction to generate a term fragment set and a context fragment set. Term consistency feature values ​​are generated based on the consistency relationship between the term fragment set and the standardized concept name corresponding to the concept identifier, and the term consistency feature values ​​are bound to the concept identifier. Contextual referential feature values ​​are generated based on the referential relationship between referential statements and concept identifiers in the context fragment set, and the contextual referential feature values ​​are bound to the concept identifiers; Based on the directed path relationship between concept identifiers and concept identifiers within the stage reachable concept set in the road and bridge professional concept dependency directed graph, concept dependency path feature values ​​are generated, and concept dependency path feature values ​​are bound to concept identifiers; Based on the consistency relationship between engineering attribute information in standardized text units of teaching content and engineering attribute labels in the representation of engineering cognitive objects, engineering attribute consistency feature values ​​are generated, and engineering attribute consistency feature values ​​are bound to concept identifiers. The terminology consistency feature value, context reference feature value, concept dependency path feature value, and engineering attribute consistency feature value corresponding to the same concept identifier are concatenated and normalized to form a matching feature vector.

8. The teaching-related semantic matching method for road and bridge professional concepts according to claim 1, characterized in that, The generation of the directional matching result set specifically includes: Obtain the set of matching feature vectors and the candidate set of stage-reachable concepts; For each matching feature vector, perform asymmetric instructional semantic matching calculation, which includes support reachability matching calculation and interpretation referential matching calculation; Perform support reachability matching calculation, taking concept identifiers as the calculation object and concept dependency path feature values, engineering attribute consistency feature values, and terminology consistency feature values ​​as input features. Weight the input features to obtain the support reachability matching score, and bind the support reachability matching score to the corresponding concept identifier and the corresponding standardized text unit identifier of the teaching content. For the same matching feature vector, perform interpretation reference matching calculation. Take the standardized text unit identifier of teaching content as the calculation object, and take the terminology consistency feature value and context reference feature value as input features. The weighted result is used to obtain the interpretation reference matching score, and the interpretation reference matching score is bound to the corresponding concept identifier and the corresponding standardized text unit identifier of teaching content. For standardized text unit identifiers of the same concept and the same teaching content, the combination of support can achieve matching scores and explanatory reference matching scores to generate directional matching results; The matching score for support reachability is compared with the matching score for interpretation reference to generate a matching direction identifier. Specifically, when the matching score for support reachability is greater than the matching score for interpretation reference, the matching direction identifier is determined as the support reachability matching direction; when the matching score for interpretation reference is greater than the matching score for support reachability, the matching direction identifier is determined as the interpretation reference matching direction; when the matching score for support reachability is equal to the matching score for interpretation reference, the matching direction identifier is determined as the bidirectional matching direction, and the matching direction identifier is bound to the directional matching result record. Perform aggregation processing on all directional matching results to generate a set of directional matching results.

9. A teaching-related semantic matching method for road and bridge engineering concepts according to claim 8, characterized in that, The directional matching results include standardized text unit identifiers for teaching content, concept identifiers, support reachability matching scores, and explanatory referential matching scores.

10. A teaching-related semantic matching system for road and bridge professional concepts, comprising executing the teaching-related semantic matching method for road and bridge professional concepts as described in any one of claims 1 to 9, characterized in that, include: The teaching content preprocessing module is used to obtain the set of teaching content for road and bridge engineering and perform preprocessing. The Engineering Cognitive Object Construction Module is used to construct representations of engineering cognitive objects and generate a set of engineering cognitive objects for road and bridge engineering concepts. The concept-dependent directed graph construction module is used to construct a set of concept-dependent directed relations based on the set of cognitive objects of road and bridge professional concept engineering, and generate a concept-dependent directed graph of road and bridge professional. The teaching evolution constraint construction module is used to construct a teaching evolution constraint mechanism based on the directed graph of road and bridge professional concept dependency and the set of road and bridge professional concept engineering cognitive objects; The teaching stage annotation module is used to perform teaching stage annotation on a standardized set of text units of teaching content, and generate a set of teaching content stage annotations. The stage-reachable concept filtering module is used to determine the candidate set of stage-reachable concepts for each standardized text unit of teaching content in the stage-annotated set of teaching content, based on the teaching evolution constraint mechanism, and to construct a matching feature vector. The asymmetric instructional semantic matching module is used to perform asymmetric instructional semantic matching calculations on standardized text units of teaching content and corresponding stage reachable concept candidate sets based on matching feature vectors, generating a set of directional matching results.

Images