Knowledge graph-based native folk culture reconstruction method and system

Abstract

The invention relates to the technical field of knowledge maps, in particular to a local folk culture reconstruction method and system based on a knowledge map, and the method comprises the following steps: obtaining festival node time and structure information, collecting the festival node time and structure information into culture behaviors, extracting types and features, connecting the types and features into a relation chain, labeling field frequencies to form core nodes, and rearranging paragraphs to establish corresponding mapping in order. According to the method, structure grouping is achieved through joint comparison of time, regions and behavior sequences of festival behavior data, core nodes are identified by behavior fields according to frequency and semantics in sequence, structure mapping is completed by role matching and semantic combination of cultural paragraphs, dependence on an external labeling system is avoided, and the method is simple and convenient to operate. A complete process from behavior extraction to atlas updating is formed, so that folk information has clear structure expression ability, semantic organization precision is improved, and adaptability and expression integrity of the atlas under heterogeneous cultural contexts are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of knowledge graphs, and in particular to a method and system for reconstructing local folk culture based on a knowledge graph. Background Art

[0002] The technical field of knowledge graphs includes related technologies for identifying, organizing, and semantically modeling entities and their relationships in structured and unstructured data. The core of this technical field lies in constructing a graph structure with semantic representation capabilities through processes such as knowledge modeling, knowledge extraction, knowledge fusion, knowledge storage, and knowledge calculation, so as to achieve semantic organization and logical reasoning of massive data. This technical field is widely applied in directions such as search engines, intelligent recommendation, question-and-answer systems, intelligent decision-making, and semantic understanding. It mainly relies on means such as natural language processing, entity recognition, relationship extraction, knowledge representation, and reasoning to construct a semantic connection network between entity nodes and realize the structured expression and efficient query of information.

[0003] Among them, the method for reconstructing local folk culture based on a knowledge graph refers to an operation method that uses knowledge graph technology to perform semantic recognition, structured expression, and associative integration on the content of local folk culture. The technical matters targeted by this method cover identifying and annotating cultural entities and their attributes in the original folk text data, extracting and analyzing the event relationships in cultural behaviors and festival customs, and performing semantic association and graph modeling on different cultural regional characteristics and time evolution processes. Specifically, language text analysis technology is used to achieve word segmentation and annotation of cultural entities, semantic relationship recognition rules are used to summarize the logical relationships between entities, knowledge fusion means are used to unify the standards of multi-source cultural data, ontology construction methods are used to organize the category system and attribute hierarchy of cultural elements, and semantic query structures are used to visually call and navigate and manage cultural information.

[0004] Traditional knowledge graph technology focuses on the static extraction of entities and relationships and usually ignores the dynamics and complexity of cultural data. Especially when dealing with folk culture data with rich spatio-temporal backgrounds and deep cultural meanings, traditional methods often seem inadequate when connecting cultural entities and behavior chains and lack effective dynamic semantic modeling capabilities. This static and one-sided data processing method is difficult to cope with the diversity and variability of cultural behaviors in different regions and different time backgrounds, resulting in low efficiency of the generated knowledge graph in dealing with complex queries and diverse application scenarios, and the practicality and accuracy of information are difficult to meet high-standard requirements. In addition, the lack of an effective cultural semantic connection mechanism makes it easy to generate information islands during the integration and update process of cultural knowledge, affecting the systematicness and scalability of cultural knowledge. Summary of the Invention

[0005] The object of the present invention is to solve the deficiencies existing in the prior art, and a method and system for reconstructing local folk culture based on a knowledge graph are proposed, including the following steps: To achieve the above object, the present invention adopts the following technical solutions: A method for reconstructing local folk culture based on a knowledge graph, including the following steps: S1: Obtain the festival node information in the knowledge graph, perform time and structural sequence comparison on the cultural element types and folk ritual behavior fields corresponding to the nodes, and obtain the content of the festival cultural behavior set; S2: Based on the cultural behaviors in the content of the festival cultural behavior set, extract the folk ritual types and cultural association characteristics, and connect them in sequence according to the logical order and content fitting degree to obtain the folk culture relationship chain; S3: Based on the node sequence in the folk culture relationship chain, extract the ritual behavior fields and cultural descriptions in the nodes, and mark the repeated manifestation forms of the behavior fields in the sequence to obtain the folk core ritual nodes; S4: Based on the core nodes in the folk core ritual nodes, obtain the cultural paragraphs in the graph that match the ritual category and role characteristics, and reorder the role participation scope and behavior description sequence to obtain the ritual-driven folk culture paragraphs; S5: Based on the associated paragraphs of the ritual-driven folk culture paragraphs, extract the ritual content, local characteristics and role interaction relationships in the paragraphs, analyze the semantic connection methods between the paragraphs, and map the combined content into the graph for cultural ontology update to obtain the reconstructed ontology of the local folk culture knowledge graph.

[0006] As a further solution of the present invention, the content of the festival cultural behavior set includes the node time sequence number, regional identification field, and behavior process segment annotation. The folk culture relationship chain includes the ritual type classification label, cultural association attribute group, and paragraph connection structure table. The folk core ritual nodes include the high-frequency behavior field list, node position information group, and ritual category attribution identifier. The ritual-driven folk culture paragraphs include the role matching paragraph set, behavior field mapping table, and paragraph sequence rearrangement table. The reconstructed ontology of the local folk culture knowledge graph includes semantic structure fragments, cultural role interaction nodes, and mapping link items in the knowledge graph.

[0007] As a further solution of the present invention, the specific steps of S1 are: S101: Obtain the festival node information in the knowledge graph, collect the cultural element types, ritual behavior fields and time annotations of the nodes, combine and analyze the time field and regional attributes, analyze the sequential position of the nodes in the festival process, and generate a festival node sequence annotation table; S102: Based on the festival node sequence annotation table, extract the content of the ritual behavior fields of the nodes, identify the process structure among the behavior fields, analyze the associated connection status of the action sequences and behavior types between adjacent nodes, and obtain the festival node process connection group; S103: Invoke the node combinations in the festival node process connection group with continuous actions and similar cultural elements, and merge the combination content according to process continuity to obtain the content of the festival cultural behavior set.

[0008] As a further solution of the present invention, the specific steps of S2 are: S201: Based on the same cultural behavior in the content of the festival cultural behavior set, extract the corresponding folk ritual types, classify the ritual type names, and group the cultural behaviors of the same category into the same group to generate a ritual type merge set; S202: Invoke the data of the same group in the ritual type merge set, extract the cultural association fields in each group, compare the positions of the time series and event sequence in the fields, calculate the sequence offset amount between the behaviors, and filter the behavior combinations with continuous sequence and similar structure to obtain the sequence structure fitting group; S203: Based on the behavior combinations in the sequence structure fitting group, extract the starting nodes and connection nodes in the combination, analyze the connection relationship of the nodes in the process sequence, and construct the combinations with sequential connection and semantic connection into a linear structure to obtain the folk culture relationship chain.

[0009] As a further solution of the present invention, the specific calculation formula of the sequence offset amount between the behaviors is: ; Wherein, represents the sequence offset amount between the behaviors, represents the time sequence position value of the th item in the behavior sequence , represents the time sequence position value of the th item in the behavior sequence , represents the cultural field sequence structure score value of the th item in the behavior sequence , represents the cultural field sequence structure score value of the th item in the behavior sequence , represents the total number of comparable behavior items.

[0010] As a further solution of the present invention, the specific steps of S3 are: S301: Based on the completed node sequence in the folk culture relationship chain, extract the ritual behavior fields and cultural description content of the nodes, classify and label the ritual fields, and construct a field index table according to the node arrangement order to obtain a ritual field index structure; S302: Invoke the field sequence in the ritual field index structure, record the cumulative occurrence times of each type of field in the node sequence, and combine the distribution ratio in the total number of nodes to calculate the repeated density value of the field distribution in the chain to obtain the ritual field repeated density measure; S303: Based on the node positions corresponding to the high-density fields in the ritual field repeated density measure, extract the ritual categories and the process numbers where the nodes are located, and set the nodes that meet the repeated density threshold and have a unified category attribution as core nodes to obtain folk core ritual nodes.

[0011] As a further solution of the present invention, the calculation formula for the repeated density value of the field distribution in the chain is specifically: ; Wherein, represents the repeated density value of the field distribution in the chain, represents the total number of nodes in the node sequence, represents the cumulative occurrence times of the f-th type of field in the e-th node, represents the extreme difference of the cumulative occurrence times of the f-th type of field in all nodes, represents the proportion of the occurrence times of the f-th type of field in the e-th node in the total number of fields of this node, represents the average value of the occurrence proportions of the f-th type of field in all nodes, represents the average value of the cumulative occurrence times of the f-th type of field in all nodes, represents the variance of the occurrence times of the fields in the e-th node.

[0012] As a further solution of the present invention, the specific steps of S4 are: S401: Based on the folk core ritual nodes, extract the corresponding ritual categories and cultural role information of the nodes, screen the paragraph content in the knowledge graph with the same ritual category and including the corresponding role tags, and generate a paragraph set corresponding to the nodes; S402: Invoke the paragraph text in the paragraph set corresponding to the nodes, extract the cultural role names that appear in the paragraphs, and perform corresponding matching with the role tags in the nodes, and mark the paragraph text whose role names are all included in the node role tags to obtain a list of paragraphs with complete role matching; S403: Based on the paragraphs marked in the role coverage marking result, extract the original position order in the graph structure, compare the node positions with the paragraph sequence numbers, and perform structural position rearrangement to align the paragraph structure with the node sequence, so as to obtain the ritual-driven folk culture paragraphs.

[0013] As a further solution of the present invention, the specific steps of S5 are as follows: S501: Based on the paragraphs associated with the ritual-driven folk culture paragraphs, extract the folk ritual content, local cultural characteristics, and cultural role fields in the paragraphs, classify and group the fields, and construct an element set corresponding to the paragraphs to generate a paragraph attribute set; S502: Invoke the ritual and role fields in the paragraph attribute set, analyze the consistency of the content order and role combination between paragraphs, identify paragraph groups with continuous structure and semantic connection characteristics, and obtain the paragraph logical sequence structure; S503: Based on the paragraph groups in the paragraph logical sequence structure, extract the original sequence numbers and logical positions of the paragraphs, adjust the sequence and combine and relocate their structural relationships in the graph to obtain the reconstructed ontology of the local folk culture knowledge graph.

[0014] The local folk culture reconstruction system based on the knowledge graph includes: The node collection module extracts the time field, region field, and ritual behavior field based on the festival event nodes in the knowledge graph, numbers the time in sequence, classifies and codes the regions, analyzes the node set with consistent time continuity and region attribution, combines the process stage content of the behavior field, identifies the action connection relationship between adjacent nodes, and filters the connectable node groups to obtain the content of the festival culture behavior set; The type recognition module extracts the ritual keywords and scene element phrases based on the cultural behaviors in the content of the festival culture behavior set, classifies and labels the keywords, performs semantic fitting analysis on the places, used items, and action methods, combines the behaviors with consistent attribution into a type sequence, and identifies the paragraphs with associated levels to obtain the folk culture relationship chain; The behavior extraction module extracts the behavior keywords of the nodes based on the node sequence in the folk culture relationship chain, counts the occurrence frequency in the paragraphs, records the keywords that are continuously stable and positionally overlapping, extracts the node positions and the upper and lower text fields, sorts them according to the frequency and coverage range, and marks the high-density aggregation areas to obtain the folk core ritual nodes; The paragraph correspondence module extracts the paragraphs in the graph that include the same behavior field and similar role content based on the high-frequency behavior fields and numbers in the folk core ritual nodes, identifies the intersection of the role and action fields, filters the paragraph content with high coincidence, and rearranges the paragraph list according to the action sequence number to obtain the ritual-driven folk culture paragraphs; The structure reorganization module extracts the action connection fields, scene names, and role phrases within the ritual-driven folk culture paragraphs, analyzes the stage continuation relationships of the action fields, establishes connection chains for paragraph groups with connection characteristics, splices and maps them into the knowledge graph in sequence, records the semantic paths between paragraphs and role phrases, and obtains the reconstructed ontology of the local folk culture knowledge graph.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, the festival behavior data is grouped by structure through the combined comparison of time, region, and action sequence. The action fields identify the core nodes according to frequency and semantic order. The cultural paragraphs complete the structure mapping through role matching and semantic combination, avoiding dependence on external annotation systems, forming a complete process from behavior extraction to knowledge graph update, enabling the folk information to have a clear structural expression ability, improving the semantic organization accuracy, and enhancing the adaptability and expression integrity of the knowledge graph in heterogeneous cultural contexts. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following-described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 is a schematic diagram of the step flow of the present invention; Figure 2 is a system module diagram of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The following will describe the technical solutions in the present invention with reference to the drawings.

[0019] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to represent examples, illustrations, or explanations. Any embodiment or design solution described as "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two can be selected.

[0020] In the embodiments of the present invention, the terms "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, their intended meanings are the same. The terms "of", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference between them is not emphasized, their intended meanings are the same.

[0021] In the embodiments of the present invention, sometimes a subscript such as W 1 may be written in a non-subscript form such as W1. When the difference between them is not emphasized, their intended meanings are the same.

[0022] To make the technical problems, technical solutions, and advantages to be solved by the present invention clearer, the following will be described in detail with reference to the accompanying drawings and specific embodiments.

[0023] The embodiments of the present invention provide a method for reconstructing rural folk culture based on a knowledge graph, including the following steps: S1: Obtain the festival node information in the knowledge graph, check each item of the cultural element type and folk ritual behavior field associated with the node one by one, and conduct a continuous comparison based on the time and regional attributes in the cultural element type and the process structure described in the ritual behavior field. Converge the nodes that are sequential in time order and have associated descriptions of cultural elements into the same cultural behavior to obtain the content of the festival cultural behavior set; S2: Based on the same cultural behavior in the content of the festival cultural behavior set, extract the folk ritual type and cultural association characteristics, compare them according to the folk ritual type, and identify the paragraphs with associated levels according to the logical order and fitting degree of the ritual occurrence in the cultural association characteristics, and conduct structural connection to obtain the folk culture relationship chain; S3: Based on the node sequence completed in the folk culture relationship chain, extract the ritual behavior field and cultural description in the node, record the occurrence frequency according to the repeated occurrence of the ritual behavior field in the node, and use the ritual node with a continuous and stable occurrence frequency of the behavior field as the core node, and mark the ritual category and position of the node to obtain the folk core ritual node; S4: Based on the folk core ritual node, obtain the paragraph content in the knowledge graph corresponding to the node ritual category and associated cultural role, compare the role overlap degree between the node and the paragraph, and adjust the position order of the paragraph in the knowledge graph according to the ritual behavior content participated by the role, so that a corresponding mapping is formed between the node and the paragraph to obtain the ritual-driven folk culture paragraph; S5: For the paragraphs related to folk culture paragraphs driven by rituals, extract the folk ritual content, local cultural characteristics, and the interaction relationships between cultural roles in the paragraphs. Conduct logical verification between paragraphs based on the interaction relationships between ritual content and cultural roles to form a coherent semantic structure combination of local folk culture, and incorporate it into the knowledge graph to form an updated cultural ontology, obtaining the reconstructed ontology of the local folk culture knowledge graph.

[0024] The content of the festival culture behavior set includes the node time sequence number, regional identification field, and behavior process segment annotation. The folk culture relationship chain includes the ritual type classification label, cultural association attribute group, and paragraph connection structure table. The folk core ritual node includes the high-frequency behavior field list, node location information group, and ritual category attribution identifier. The ritual-driven folk culture paragraph includes the role matching paragraph set, behavior field mapping table, and paragraph order rearrangement list. The reconstructed ontology of the local folk culture knowledge graph includes semantic structure fragments, cultural role interaction nodes, and mapping link items in the knowledge graph.

[0025] The specific steps of S1 are as follows: S101: Obtain the festival node information in the knowledge graph, collect the cultural element types, ritual behavior fields, and time annotations of the nodes, combine and analyze the time fields and regional attributes, analyze the sequential positions of the nodes in the festival process, and generate a festival node sequence annotation table; First, extract all node objects belonging to the festival theme in the atlas, and read the basic attribute content of each node one by one, including the node number, the name of the festival event it belongs to, the activity time field, the regional classification label, and the functional behavior field. For the activity time field, uniformly convert it to the standard format of "year-month-day-period". For example, convert "the morning of the 5th day of the 2nd lunar month in 2024" to "2024-03-14-AM". Then, in combination with the regional classification field, map the administrative region, geographical location name, and local activity category one by one. For example, "Huizhou - mountain village - farming festival", and complete the operation of merging the attribution of the activity background. After the structure binding of the time and regional fields is completed, a standard combined field sequence is formed, with the construction form being the node number, time field + regional classification field. Then, using the time field as the sorting benchmark, number the nodes in chronological order. For example, if node A is marked with the time "2024-03-01-AM" and node B is marked with the time "2024-03-02-AM", then the sorting numbers are A = 1 and B = 2, forming the basic time process sequence. Subsequently, extract the functional behavior field and divide the described behavior types into stages, such as action expressions like "gather, spread, tidy up", corresponding to the behavior stages of "pre - process - end". On the basis of the time sorting result, add the behavior stage dimension. If the behavior of node A is "gather" and the behavior of node B is "tidy up", then their order in the festival process is A < B. Set the process sequence identifier of the node according to the time number and the behavior stage. For example, for node A, the sequence number = 1, and for node B, the sequence number = 2. Perform the number generation process for all festival nodes in this way. For example, if a certain paragraph in the festival atlas contains "In early March 2024, site preparation is carried out in the northern mountainous area", and another paragraph describes "Group sorting is carried out the next morning", then the two nodes should be assigned "time number 1, stage identifier pre" and "time number 2, stage identifier process" respectively, to obtain the festival node sequence annotation table.

[0026] S102: Based on the festival node sequence annotation table, extract the content of the ritual behavior field of the nodes, identify the process structure between the behavior fields, analyze the associated connection status of the action sequence and behavior type of adjacent nodes, and obtain the festival node process connection group; First, extract the ritual behavior field information corresponding to the sequential numbers in each node. The ritual behavior field is the activity action label recorded within the node. Common field examples are "setting up the stage, lining up, performing segments, exiting the stage". Arrange all the nodes in ascending order of the sequential numbers to construct a sequential behavior set. Classify the content of each behavior field according to the action function, and divide the actions into three functional stages: "preparation type", "execution type", and "tidying up type". For example, "setting up the stage, lining up" are classified as the preparation type, "performing segments, performing shows" are classified as the execution type, and "removing the cloth, exiting the stage" are classified as the tidying up type. After completing the classification, sequentially judge the corresponding relationship of the behavior function classification between adjacent nodes to determine whether the sequence conforms to the functional connection logic. If the previous node is of the preparation type and the next node is of the execution type, set it as the forward connection state. If the previous is of the execution type and the next is of the tidying up type, set it as the backward connection state. If the two nodes belong to the same function or the sequence is reversed, set it as an invalid connection. Then, extract the action sequence according to the nodes with valid connections, identify the arrangement method of the action content between the two nodes, and determine the sequence of the last action of the previous node and the first action of the next node. For example, if node A is "lining up - taking positions" and node B is "performing segments - taking a bright step", then there is no content repetition and the logic is continuous between the actions, which is determined as a sequential path. Assign a process label and classify all node groups that meet the conditions of "function classification matching and action sequence". Establish a mapping record for the node groups with valid process label values according to the sequential numbers, including node number pairs, connection directions, and behavior stage pairing categories. Integrate all eligible node combinations to obtain the festival node process connection group.

[0027] S103: Call the node combinations in the festival node process connection group where the actions are continuous and the cultural elements are similar, and merge the combination content according to the process continuity to obtain the content of the festival cultural behavior set; First, identify the node groups with valid connection marks, call the previous and next node numbers in each combination, and extract the action fields and cultural description field contents recorded in the nodes. Arrange the action fields into a continuous sequence according to the node order, and identify the semantic relationship between adjacent action fields to determine whether they have the process connection attribute. For example, if the action of the previous node is "assembling" and the action of the next node is "lining up to enter", it can be regarded as a connection action and set as a positive sequential relationship. If the expression of the action content is very different, such as "entering" before and "closing summary" after, it is regarded as a stage interruption relationship and does not have the sequential attribute. Such a relationship will be marked as an invalid continuation. Then perform a comparison process on the cultural description field contents in the node combination. The cultural field mainly extracts three types of information: regional name, activity name category, and activity stage identifier. For example, if the cultural field of a node is "South of the City·Agricultural Exhibition·First Stage", and the cultural field of another node is "South of the City·Agricultural Exhibition·Second Stage", the region and category are consistent, and there is a progressive relationship between the stages. The cultural elements can be regarded as consistent, and such a combination is recorded. The state is "action connection + cultural alignment". Then, all combinations that meet the action continuity and whose cultural elements do not differ by more than one level are structurally merged. The node sequence numbering in each combination is determined. If there is no interruption between the numbers and the difference in continuous numbers is 1, the adjacent combinations are connected in series into a single combination segment. The action fields in the combination segment are extracted again and summarized to form a merged action list. For example, combination segment A contains "prepare materials and arrange the venue", and combination segment B contains "line up to enter the venue and start the ceremony". The merged order is "prepare materials, arrange the venue, line up to enter the venue and start the ceremony". The fields in this list are numbered in logical order, and each field records the original node to which it belongs for subsequent backtracking. After the list is formed, all cultural description fields of the merged paragraphs are extracted to generate a unified cultural label combination. The labels are checked for uniqueness. If there are multiple activity names or region fields, the one with the highest frequency of occurrence is output as the main label. The action field list and the cultural label combination are bound to an integrated structure output, and finally the content of the festival cultural behavior collection is obtained.

[0028] The specific steps of S2 are: S201: based on the same cultural behavior in the festival cultural behavior set content, extract the corresponding folk ritual type, classify the ritual type name, and classify the cultural behaviors of the same category into the same group to generate a ritual type merge set; First, extract the behavior record data in the set item by item, extract the action expression statements and environmental description phrases in each behavior field, perform semantic extraction on the verb structures and their accompanying functional terms in the text, and classify them according to the preset classification criteria of five behavior stages: "greeting", "preparation", "performance", "communication", and "ending". For each subject verb in the behavior field, perform a matching judgment. For example, the verb "arrange" in "venue arrangement" is classified into the preparation category, the "drink together" in "sit down and drink together" is classified into the communication category, and the "clean up" in "clean up the venue" is classified into the ending category. After extracting the verb, determine its behavior function according to the position of the verb in the sentence, the collocation object, the sentence structure, etc., and then group the fields with the same attribution according to the five-stage labels to construct a behavior field attribution dictionary table. Subsequently, count the number of behavior records in each group, and set the behavior group confirmation threshold to be greater than or equal to 3 records. For example, if the behavior fields "venue arrangement, seat arrangement, program rehearsal, team formation, prop handling, food distribution, environmental cleaning" are extracted from a regional cultural record, after extraction and classification, "arrange, prepare, rehearse" are classified into the preparation category, "distribute, clean up" are classified into the ending category, "formation" is classified into the greeting category, and "handling" is not classified temporarily due to lack of behavioral independence. Then, there are 3 records in the preparation category, 2 records in the ending category, and 1 record in the greeting category. Only the preparation category meets the formation conditions. The corresponding fields are uniformly grouped into the "preparation stage group", and the field group is bound to the "preparation" label to form a behavior merging structure, and finally a ritual type merging set is obtained.

[0029] S202: Call the data in the same group in the ritual type merging set, extract the culture-related fields in each group, compare the positions of the time series and event order in the fields, calculate the sequential offset amount between behaviors, and screen out the behavior combinations with continuous order and similar structures to obtain the sequential structure fitting group; The specific calculation formula for the sequential offset amount between behaviors is: ; Among them, represents the sequential offset amount between behaviors, represents the behavior sequence in the th behavior sequence in the th behavior sequence in the th behavior sequence in the th behavior sequence Specific calculation examples and operation derivation processes: Set the following parameters: , and the behavior is recorded by the event detection system The first item of occurs at time point 4 , and the behavior is recorded by the event detection system The first item of occurs at time point 6 After the cultural field vector of the behavior is processed by the structured text encoding system and merged, is obtained After the cultural field vector of the behavior is processed by the structured text encoding system and merged, is obtained Substitute the above values into the formula: ; ; ; This result shows that the behavior sequence and the behavior sequence have a shift degree of 19.65 in the order of the first behavior, indicating that there are obvious differences in the time sequence and the cultural field structure characteristics of the two behaviors. This shift index will be used as the input basis for measuring the continuity and structural similarity of the behavior combination, providing a basic judgment for the subsequent screening of the structural fitting group.

[0030] The specific steps of S3 are as follows: S301: Based on the completed node sequence in the folk culture relationship chain, extract the ritual behavior fields and cultural description contents of the nodes, classify and mark the ritual fields, and construct a field index table according to the node arrangement order to obtain the ritual field index structure; First, number the node sequence, confirm the arrangement order of the nodes in the graph structure, and sequentially extract the ritual behavior fields and cultural description fields corresponding to each node. The ritual behavior fields are mainly composed of phrases representing operation actions, such as "setting up the venue", "taking one's place", "passing objects", "seeing off", etc. The cultural description fields are used to reflect regional characteristics, festival backgrounds, or role attributes, such as "village entrance square", "February of the lunar calendar", "host", etc. After extraction, split the ritual behavior fields into action words and modifiers, perform type attribution operations on the action words, and classify them into stages according to the functional characteristics of the behaviors. The commonly used classification dimensions are set as "preparation type", "execution type", "ending type". For example, "setting up the venue" belongs to the preparation type, "passing objects" belongs to the execution type, and "seeing off" belongs to the ending type. After classification, add the category tags to the fields. For example, P01, E03, and F02 correspond to the index codes under the preparation, execution, and ending categories respectively. Subsequently, write the field marking results into the structure sequence according to the node order, establish a "node number + behavior mark" comparison table, and construct a field index structure according to the node numbers to generate a vertically segmented and horizontally classified field matrix structure. Each node corresponds to a set of behavior type indexes, and the field marks can appear repeatedly to represent the distribution of the same type of behavior in different nodes. For example, if the behaviors of node 5 are "setting up the venue, taking one's place", then its marks are P01, P02, and the corresponding index is 5 - P01 - P02. If the behaviors of node 6 are "passing objects, assisting", then it is 6 - E01 - E04. Finally, after completing the field extraction, behavior classification, and structure coding of all nodes, combine and summarize them to form an ordered behavior mark table to obtain the ritual field index structure.

[0031] S302: Call the field sequence in the ritual field index structure, record the cumulative occurrence times of each type of field in the node sequence, and calculate the repeated density value of the field distribution in the chain by combining the distribution ratio in the total number of nodes to obtain the ritual field repeated density measure; The specific calculation formula for the repeated density value of the field distribution in the chain is: ; Among them, represents the repeated density value of the field distribution in the chain, represents the total number of nodes in the node sequence, represents the cumulative occurrence times of the f - th type of field in the e - th node, represents the extreme difference value of the cumulative occurrence times of the f - th type of field in all nodes, represents the proportion of the occurrence times of the f - th type of field in the e - th node to the total number of fields in this node, represents the average value of the occurrence proportions of the f - th type of field in all nodes, represents the average value of the cumulative occurrence times of the f - th type of field in all nodes, It represents the variance of the number of occurrences of a field in the e-th node; Set the specific values as follows: The total number of nodes obtained by monitoring and collection ; Field The number of occurrences in each node is respectively , , ; ; The occurrence ratios of this field in the corresponding nodes are respectively , , ; Accordingly, the average ratio is: ; The average value of the number of occurrences is: ; The variances of the fields of each node collected by the monitoring system are: , , ; Substitute the above values into each item of the formula in sequence, and the operation expression is: ; This result indicates that the fluctuation value of the repetition density of the field among the nodes in the chain is 1.035, indicating that the field The combined difference degree of the occurrence frequency and ratio in the three nodes is relatively obvious. This value is obtained by mixing and weighting the normalized occurrence frequency of the field combined with its ratio deviation in each node and the within-node fluctuation variance, and then through standardization processing. This result is directly related to the ritual field repetition density measure and can be used to reveal the repetition pattern of the field in the chain structure, thereby further supporting the subsequent field redundancy control and distribution optimization analysis.

[0032] S303: Based on the node positions corresponding to the high-density fields in the ritual field repetition density measure, extract the ritual categories and the process numbers where the nodes are located. Set the nodes that meet the repetition density threshold and have the same category attribution as core nodes to obtain the folk core ritual nodes; First, select the fields with density values greater than the specified quantity threshold from the already constructed field density record table. The threshold setting refers to the average occurrence frequency of each field in the overall behavior fields. The average value is obtained by summing the occurrence frequencies of all behavior fields in the entire sequence and then dividing by the total number of fields. Then, select the fields greater than this average value as high-density items. For example, if the behavior fields appear 182 times in 72 nodes, the average density is 2.52, rounded up to 3 as the density screening threshold. All behavior fields with an occurrence count ≥ 3 are classified as high-density fields. For example, the fields "arrangement", "transfer", and "collection" appear in 5, 6, and 4 nodes respectively, meeting the conditions. Next, trace back to the corresponding node positions of these fields in the graph, extract their corresponding node numbers through the field-node mapping table, record the positions where each field appears in different nodes to form a field positioning matrix, and then read the "ritual category" field of these nodes. This field identifies the ritual paragraph classification to which the node belongs, such as "preparation category", "interaction category", "handover category", etc. Judge whether all the nodes corresponding to each field belong to the same ritual category. If a field appears in 5 nodes, 4 of which are in the "preparation category" and 1 is in the "ending category", it does not meet the condition of belonging to the same category. If the field all belongs to a single category, generate a candidate core node group through all the node numbers mapped by this field, and then extract the "process number" of each node. This number represents the sequential position of the node in the complete structure. For example, the process number of node number 10 is P10. Sort each group of candidate core nodes according to the process number, screen out the node groups with position jumps or process intervals greater than 2, and retain the set of continuously numbered processes. Mark each node in this set as a "core node" to obtain the folk custom core ritual nodes.

[0033] The specific steps of S4 are as follows: S401: Based on the folk custom core ritual nodes, extract the corresponding ritual category and cultural role information of the nodes, screen the paragraph content in the knowledge graph with the same ritual category and including the corresponding role labels, and generate a paragraph set corresponding to the nodes; It is necessary to extract a list of all numbers that have been marked as core nodes from the graph structure, and call the "ritual category" field and "cultural role" field under each node. The ritual category field is used to identify the process position of the behavior, such as preparation, interaction, and closing. The cultural role field indicates the type of people involved in the behavior or the function setting, such as "host", "assistant", "elder", "organizer", etc. First, extract the value of the "ritual category" field and establish a node classification mapping table. All core nodes are grouped and sorted according to the category field. The corresponding role set is recorded under each group, and a "node-role" two-way binding index structure is generated. Then enter the graph corpus block, extract all encoded paragraph contents from the text paragraph set, and identify the ritual category label of each paragraph. The label is the process attribute value bound to the paragraph. The extraction logic is to read the "ritual process type" content in the identification structure or structure annotation field at the front of the paragraph, and compare this field with the ritual category in the node. The field is matched for equality. If the node ritual category is "preparation category", only the text with the paragraph ritual label of "preparation category" is retained. After a successful match, the role matching process is entered, and the paragraph body content is decomposed to extract the entities corresponding to the "cultural role" in the semantic unit. For example, in the sentence patterns such as "the host gathers the participants" and "the organizer distributes the materials", the role phrases such as "host" and "organizer" are identified, and the set intersection matching is performed with the node-bound role. The paragraph that matches at least one co-occurring role is determined as a candidate paragraph. If the overlap between the number of roles in the paragraph and the number of node roles exceeds 50%, it is set as a strong matching segment. For example, if the node role is set to "host, assistant", "host leads" and "assistant arranges" appear in the paragraph, it is considered a complete match. If only one of them appears, it is recorded as a partial match. Finally, all paragraphs that meet the consistency of the ritual category and partial or complete match of the role field are screened and collected to obtain the node corresponding paragraph set.

[0034] S402: calling the paragraph text in the paragraph set corresponding to the node, extracting the cultural role names appearing in the paragraphs, matching them with the role labels in the nodes, marking the paragraph texts whose role names are all included in the node role labels, and obtaining a list of paragraphs with complete role matches; First, extract the main content of the text in order according to the paragraph number. Use a word extraction tool or a manually annotated role word list to extract all the words representing characters or performing duties in the text, and construct a paragraph role word set. Each word set corresponds to the paragraph number. For example, "host", "assistant" and "porter" are extracted from paragraph D007, forming the word set R (D007) = host, assistant, porter. Then call the established folk core ritual node information and query the role label set R (N) under the node. For example, node N015 sets the role label as host, assistant, organizer, and executes the set operation step to cross-check the paragraph word set with the node role set. The operation process is to compare each role word in the paragraph word set R (Di) one by one. If the word is included in the node role set, it is retained, otherwise it is discarded. When all entries in R (Di) appear in R (N), that is, R ( Di) is a subset of R(N), then the paragraph is considered to be a complete role match for the node and enters the next screening stage. If there is an entry that is not included in the node role set, the paragraph is directly skipped. This judgment logic is repeated for all paragraphs, and a list of paragraph numbers that meet the "complete inclusion" condition is recorded. For example, if paragraph D012 only contains "assistant" and "host", and the node role set is host, assistant, organizer, then D012 meets the complete match condition. If the role words in paragraph D014 are "foreign guest" and "explainer", and this label is not set in the node, it will not be included in the matching result. In actual processing, aliases and common names between role words need to be processed with equivalent meanings. For example, "person in charge" and "organizer" are unified and merged into one type of role word to avoid missing actual matching results. The paragraph text number set of all paragraphs that meet the node role label that completely covers the paragraph role word set is summarized to obtain a list of paragraphs with complete role matches.

[0035] S403: based on the paragraphs marked in the role coverage marking results, extract the original position order in the graph structure, compare the node position with the paragraph sequence and rearrange the structure position to align the paragraph structure with the node sequence, and obtain the ritual-driven folk culture paragraph; First, obtain all the paragraph numbers marked as "exact match", and sequentially extract their paragraph sequence numbers in the original knowledge graph text sequence. This sequence number is the position number of the paragraph in the corpus, usually represented by the sequence field in the paragraph structure table. For example, the position of paragraph D006 is sequence number 32, and D008 is 35. After extracting the sequence numbers of all paragraphs, bind them to their respective core node numbers. The core node number is the structure index number held by the "folk custom core ritual node" in the graph, such as N009, N011, etc. Then, obtain the arrangement order of each core node in the graph node structure, denoted as the node sequence number, to form a one-to-one correspondence table, that is, the combined structure of paragraph sequence number - node number - node sequence number. Sort this structure, arranging it in ascending order with the node sequence number as the primary key, and retrieve whether there are records where the paragraph sequence is inconsistent with the node sequence in the sorted result. For example, if node N007 is in the 3rd position, but its corresponding paragraph is D011 with a position of 9th, then it is determined that the paragraph position is inconsistent with the node, and enter the position rearrangement process. Move D011 to the corresponding position sequence 3 of N007, and so on, perform paragraph position adjustment operations on all inconsistent records. The position adjustment operation should maintain the text integrity between paragraphs, replace the original position with numbers and structure pointers, and avoid semantic breaks caused by content movement. After completing the paragraph movement, generate a new paragraph sequence structure table, recording the new position, bound node number, and original number of each paragraph. Check whether there are duplicate paragraph mapping situations in this structure table. If a paragraph is mapped to two nodes, execute the first-occurrence retention strategy, that is, preferentially retain the position corresponding to the earlier-occurring node, and delete the other mapping. This strategy prevents redundant paragraphs from appearing repeatedly. For example, if paragraph D015 is mapped to nodes N004 and N005, and N004 is in the earlier position, then D015 is retained for N004, and the mapping to N005 is released. Finally, complete the consistent alignment of the three-party structure of paragraph number - position - node sequence to obtain the ritual-driven folk culture paragraphs.

[0036] The specific steps of S5 are as follows: S501: Based on the paragraphs associated with the ritual-driven folk culture paragraphs, extract the folk ritual content, local cultural characteristics, and cultural role fields in the paragraphs, classify and group the fields, and construct the element set corresponding to the paragraphs to generate the paragraph attribute set; First, obtain all the paragraph texts that have been marked as alignment completed, and sequentially extract the semantic field content in each paragraph according to the paragraph number. Focus on identifying three types of semantic units: one is the phrase describing the action, which is identified as "folk ritual content"; the second is the elements related to regions, natural environments, dialect features, etc., which are identified as "local cultural characteristics"; the third is the person's title participating in the execution or organization of the action, which is identified as "cultural role field". The extraction process is carried out by using an artificial recognition form or matching the keywords in the corpus. For example, for the paragraph "On the village square, the host called the villagers together to form a formation", then "call" and "form a formation" are identified as ritual content, "village square" is a cultural characteristic, and "host" and "villagers" are role fields. Perform a classification and summary operation on various fields, and classify the extraction results into the corresponding categories according to the field attributes to form a paragraph field structure set. Subsequently, number and identify the three types of fields in each paragraph respectively. The numbering basis is the position order of the field's first appearance in the paragraph, and the coding format is used, such as: I1 (ritual field 1), L2 (regional characteristic 2), R1 (role field 1), etc., to ensure the traceability of the fields in the semantic order. After the field classification is completed, organize the combination methods of the three types of fields to construct a ternary element structure set. Each paragraph forms a three-dimensional structure of a ritual field set, a cultural characteristic set, and a role set, and at the same time records the paragraph number to which the set belongs and establishes a paragraph index pointer. For example, after extracting paragraph D014, it generates: I = "set the position", "take a seat", L = "inside the ancestral hall", R = "host", "young people", and is marked as the attribute set number A014. Archive and organize the ternary field structures corresponding to all paragraphs to obtain a paragraph attribute set.

[0037] S502: Invoke the ritual and role fields in the paragraph attribute set, analyze the consistency of the content order and role combination among paragraphs, and identify the paragraph groups with continuous structure and semantic connection features to obtain the paragraph logical sequence structure; First, extract the two items of "ritual field set" and "role field set" in each paragraph structure. The ritual field set is used to characterize the attributes of the paragraph behavior stage, and the role field set is used to identify the identity types of the execution unit and the behavior participants. Number all the paragraph structures in the original paragraph order to form a sequential array. Compare the ritual field structures between two adjacent paragraphs, and extract the last field and the first field of the next paragraph for action logic matching. For example, if the ritual fields of paragraph D015 are venue arrangement and seating, and those of D016 are action instructions and role announcement, then extract "seating - action instructions" as the cross-paragraph connection group. Perform action logic judgment on the connection group to identify whether it is a combination with process continuity in the ritual stage sequence. Combinations such as "venue arrangement - taking one's seat" or "greeting - passing" can be regarded as continuous operations, and such combinations are determined as structural connection items. Then perform role field comparison to determine whether there is a role intersection or a leading role continuation relationship between the role field sets of the two paragraphs. For example, if the role fields of D015 are the host and villagers, and those of D016 are the host and the organizer, then there is a leading role of "the host" running through the behavior chain, meeting the role combination connection condition. Set all the paragraph pairs that simultaneously meet the action logic continuity and the role leading consistency as "continuable structure groups". Then group the paragraphs that continuously meet the above characteristics in multiple adjacent paragraphs. For example, if paragraphs D013 - D014 - D015 all meet the connection conditions, they are combined into "logical sequence paragraph group A", and record the starting and ending paragraph numbers, the core behavior field sequence, and the role continuity field chain within the group. Check the behavior order within all the paragraph groups and eliminate the group pairs with action conflicts or sequence contradictions. For example, "removal - arrangement" belongs to a logical conflict. After elimination and grouping, integrate all the paragraph combinations that meet the action stage continuation + role leading consistency into a sequential structure unit to obtain the paragraph logical sequence structure.

[0038] S503: Based on the paragraph groups in the paragraph logical sequence structure, extract the original order and logical position of the paragraphs, adjust the sequential relationship of their structures in the knowledge graph, and combine and relocate them to obtain the reconstructed ontology of the local folk culture knowledge graph; First, obtain the paragraph numbers contained in each paragraph group and the sequential positions of the corresponding paragraphs in the original knowledge graph, and extract the arrangement order of each paragraph group in the logical structure to form an "original position-logical order" mapping table. This mapping table is used to record the position differences between the paragraphs in the initial structure and semantic structure of the text. For example, paragraph D012 is located at serial number 34 in the graph, but its order in logical group B is 2. Then, the paragraph needs to be adjusted from position 34 to the position of logical order 2. Perform position comparison on all paragraphs to identify paragraphs with offsets between the original order and the logical order. These paragraphs are structurally rearranged. When rearranging, adjust the pointer numbers of the paragraphs in the graph based on the logical order, retain the original paragraph content numbers, and mark the "migration identifier" field in the rearrangement list to indicate that the paragraph has been migrated from the original order to the new structural position. Then, embed the rearranged paragraphs into the original graph paragraph structure in logical order to form an adjusted paragraph structure. The rows are repaired synchronously, and the "forward link" and "backward link" fields of each paragraph are updated to ensure that the structural logic chain is not broken. For example, if paragraph D010 was originally after D009 and before D011, and is now moved after D013, the backward link of D013 will be pointed to D010, and the backward link of D010 will be updated to D014. After all the link relationships are repaired, a paragraph sequence with dual alignment of logic and structure is formed. On this basis, according to the "ritual behavior field", "cultural role field", "regional context field" and other information of each paragraph, the attribute binding relationship between the paragraph and the node in the graph is reconstructed, and a synchronous mapping between the paragraph content pointer and the node in the ritual flow chart is established. The paragraph number is bidirectionally bound to the core node number of the knowledge graph. For example, paragraph D010 is bound to node N013, and D011 is bound to N014. The node control table is rebuilt to complete the homing pairing relationship between the paragraph and the node, and finally the reconstructed ontology of the local folk culture knowledge graph is obtained.

[0039] See also Figure 2 , a local folk culture reconstruction system based on knowledge graph, including: The node collection module extracts the time field, region field and ritual behavior field based on the festival event nodes in the knowledge graph, numbers the time in sequence, classifies and codes the region, analyzes the node set with consistent time continuity and region affiliation, combines the process stage content of the behavior field, identifies the action connection relationship between adjacent nodes, and screens the node group that can be connected to obtain the content of the festival cultural behavior set; The type recognition module extracts ritual keywords and scene element phrases based on the cultural behaviors in the festival cultural behavior collection content, classifies the keywords, performs semantic fitting analysis on the places, objects used, and action methods, combines behaviors with consistent attribution into type sequences, identifies paragraphs with associated levels, and obtains the folk culture relationship chain; Based on the node sequence in the folk culture relationship chain, the behavior extraction module extracts the behavior keywords of the nodes, counts their occurrence frequencies in the paragraphs, records the keywords that are continuously stable and positionally overlapping, extracts the node positions and the upper and lower text fields, sorts them according to the frequency and coverage range, marks the high-density aggregation areas, and obtains the folk core ritual nodes; Based on the high-frequency behavior fields and numbers in the folk core ritual nodes, the paragraph corresponding module extracts the paragraphs in the graph that include the same behavior fields and similar role contents, identifies the intersection of the role and action fields, filters the paragraph contents with high overlap, and rearranges the paragraph list according to the action sequence to obtain the folk culture paragraphs driven by the ritual; Based on the folk culture paragraphs driven by the ritual, the structure reorganization module extracts the action connection fields, scene names, and role phrases within the paragraphs, analyzes the stage continuation relationship of the action fields, establishes connection chains for the paragraph groups with connection characteristics, splices them in order and maps them into the graph, and records the semantic paths between the paragraphs and the role phrases to obtain the reconstructed ontology of the local folk culture knowledge graph.

[0040] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claimed rights.

Claims

1. A method for reconstructing local folk culture based on knowledge graph, characterized in that: The following steps are involved: S1: Obtain festival node information in the knowledge graph, compare the cultural element type and folk ritual behavior field corresponding to the node in terms of time and structure, and obtain the content of the festival cultural behavior set; S2: Based on the cultural behaviors of the festival cultural behavior set, extract the folk ritual types and cultural association characteristics, connect them in logical order and content fitting degree, and obtain the folk culture relationship chain; S3: Based on the node sequence in the folk culture relationship chain, the ritual behavior fields and cultural descriptions in the nodes are extracted, and the repeated manifestations of the behavior fields in the sequence are marked to obtain the folk core ritual nodes; S4: Based on the core nodes in the folk core ritual nodes, obtain the cultural paragraphs in the atlas that match the ritual category and role characteristics, reorder the role participation scope and behavior description sequence, and obtain the ritual-driven folk culture paragraphs; S5: Based on the ritual-driven folk culture paragraph association segments, the ritual content, local characteristics and role interaction relationships in the paragraphs are extracted, the semantic connection between paragraphs is analyzed, the combined content is mapped into the graph to update the cultural ontology, and the local folk culture knowledge graph is obtained to reconstruct the ontology.

2. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized by: The content of the festival cultural behavior set includes node time sequence numbering, regional identification field, and behavior process segment annotation; the folk culture relationship chain includes ritual type classification labels, cultural related attribute groups, and paragraph connection structure tables; the folk core ritual nodes include high-frequency behavior field lists, node location information groups, and ritual category attribution identifiers; the ritual-driven folk culture paragraphs include role matching paragraph sets, behavior field mapping tables, and paragraph order rearrangement lists; the local folk culture knowledge graph reconstruction ontology includes semantic structure fragments, cultural role interaction nodes, and mapping link items in the knowledge graph.

3. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized in that: The specific steps of S1 are: S101: Obtain festival node information in the knowledge graph, collect the cultural element type, ritual behavior field and time annotation of the node, combine and analyze the time field and regional attributes, analyze the sequential position of the node in the festival process, and generate a festival node sequence annotation table; S102: Based on the festival node sequence annotation table, extract the ritual behavior field content of the node, identify the process structure between the behavior fields, analyze the association and connection status of the action sequence and behavior type of adjacent nodes, and obtain the festival node process connection group; S103: calling a node combination with continuous actions and similar cultural elements in the festival node process connection group, merging the combination content according to process continuity, and obtaining a festival cultural behavior collection content.

4. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized in that: The specific steps of S2 are: S201: based on the same cultural behavior in the festival cultural behavior set, extract the corresponding folk ritual type, classify the ritual type name, and classify the cultural behaviors of the same category into the same group to generate a ritual type merged set; S202: calling the same group of data in the ritual type merged set, extracting the culturally related fields in each group, comparing the positions of the time series and event sequence in the fields, calculating the sequence offsets between behaviors, screening the behavior combinations with continuous sequence and similar structure, and obtaining the sequence structure fitting group; S203: Based on the behavior combination in the sequential structure fitting group, extract the starting node and the connecting node in the combination, analyze the connection relationship of the nodes in the process sequence, construct the sequentially connected and semantically connected combination into a linear structure, and obtain the folk culture relationship chain.

5. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized in that: The calculation formula of the sequence offset between the behaviors is specifically: ； in, Represents the sequence offset between behaviors. Representing a behavior sequence Middle The temporal position value of the item, Representing a behavior sequence Middle The temporal position value of the item, Representing a behavior sequence Middle The cultural field sequence structure score value of the item, Representing a behavior sequence Middle The cultural field sequence structure score value of the item, Represents the total number of comparable behavior items.

6. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized in that: The specific steps of S3 are: S301: Based on the node sequence connected in the folk culture relationship chain, extract the ritual behavior field and cultural description content of the node, classify and mark the ritual field, and construct a field index table according to the node arrangement order to obtain the ritual field index structure; S302: calling the field sequence in the ritual field index structure, recording the cumulative number of occurrences of each type of field in the node sequence, and calculating the repetition density value of the field distribution in the chain in combination with the distribution ratio in the total number of nodes, to obtain the ritual field repetition density value; S303: Based on the node position corresponding to the high-density field in the ritual field repetition density, the ritual category and process number of the node are extracted, and the nodes that meet the repetition density threshold and have the same category are set as core nodes to obtain the folk core ritual nodes.

7. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized in that: The calculation formula for the repetition density value of the field distributed in the chain is specifically: ； in, Represents the repetition density value of the field in the chain. Represents the total number of nodes in the node sequence, Represents the cumulative number of occurrences of the f-th type field in the e-th node, Represents the range value of the cumulative number of occurrences of the f-th field in all nodes, Represents the ratio of the number of occurrences of the f-th type of field in the e-th node to the total number of fields in the node. Represents the average value of the proportion of the f-th type of field in all nodes, Represents the average value of the cumulative occurrence of the f-th field in all nodes. Represents the variance of the number of occurrences of the field in the e-th node.

8. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized in that: The specific steps of S4 are: S401: Based on the folk core ritual node, extract the ritual category and cultural role information corresponding to the node, filter the paragraph content with the same ritual category and corresponding role label in the knowledge graph, and generate a paragraph set corresponding to the node; S402: calling the paragraph text in the paragraph set corresponding to the node, extracting the cultural role names appearing in the paragraphs, matching them with the role labels in the nodes, marking the paragraph texts whose role names are all included in the node role labels, and obtaining a list of paragraphs with complete role matches; S403: Based on the paragraphs marked in the role coverage marking result, extract the original position order in the graph structure, compare the node position with the paragraph sequence and rearrange the structural position to align the paragraph structure with the node sequence, and obtain the ritual-driven folk culture paragraph.

9. The method for reconstructing local folk culture based on knowledge graph according to claim 1 is characterized in that: The specific steps of S5 are: S501: Based on the paragraphs associated with the ritual-driven folk culture paragraph, extract the folk ritual content, local cultural characteristics and cultural role fields in the paragraphs, classify and group the fields, and construct a feature set corresponding to the paragraph to generate a paragraph attribute set; S502: calling the ritual and role fields in the paragraph attribute set, analyzing the consistency of the content sequence and role combination between paragraphs, identifying paragraph groups with continuous structure and semantic connection characteristics, and obtaining the paragraph logical sequence structure; S503: Based on the paragraph groups in the paragraph logical sequence structure, extract the original order and logical position of the paragraphs, adjust the sequence and combine and reposition the structural relationship in the graph, and obtain the reconstructed ontology of the local folk culture knowledge graph.

10. The local folk culture reconstruction system based on knowledge graph is characterized by: The system is used to implement the method for reconstructing local folk culture based on knowledge graph according to any one of claims 1 to 9, and the system includes: The node collection module extracts the time field, region field and ritual behavior field based on the festival event nodes in the knowledge graph, numbers the time in sequence, classifies and codes the region, analyzes the node set with consistent time continuity and region affiliation, combines the process stage content of the behavior field, identifies the action connection relationship between adjacent nodes, and screens the node group that can be connected to obtain the content of the festival cultural behavior set; The type recognition module extracts ritual keywords and scene element phrases based on the cultural behaviors in the festival cultural behavior set content, classifies the keywords, performs semantic fitting analysis on the places, objects used, and action methods, combines behaviors with consistent attribution into type sequences, identifies paragraphs with associated levels, and obtains folk culture relationship chains; The behavior extraction module extracts the behavior keywords of the nodes based on the node sequence in the folk culture relationship chain, counts the frequency of occurrence in the paragraph, records the keywords that are continuous, stable and overlapped, extracts the node positions and context fields, sorts them by frequency and coverage, marks the high-density aggregation areas, and obtains the folk core ritual nodes; The paragraph correspondence module extracts paragraphs with the same behavior fields and similar role contents in the graph based on the high-frequency behavior fields and numbers in the folk core ritual nodes, identifies the intersection of the role and action fields, selects the paragraph contents with high overlap, and rearranges the paragraph list according to the action sequence to obtain the ritual-driven folk culture paragraphs; The structural reorganization module is based on the ritual-driven folk culture paragraphs, extracts the action connection fields, scene names and role phrases in the paragraphs, analyzes the stage continuation relationship of the action fields, establishes connection chains for paragraph groups with connection features, splices them in sequence and maps them into the graph, records the semantic paths between paragraphs and role phrases, and obtains the reconstruction ontology of the local folk culture knowledge graph.

Images