Search

CN-121981244-A - Dynamic map visualization generation method and system based on relational tag graph

CN121981244ACN 121981244 ACN121981244 ACN 121981244ACN-121981244-A

Abstract

The application discloses a dynamic map visual generation method and a system based on a relational tag map, which relate to the technical field of dynamic map generation, the proposal carries out field association logic dynamic correction on a key field and an auxiliary field, acquires a corresponding correction result, judges whether to implement field association logic optimization based on the correction result, acquires a corresponding optimization result if the field association logic optimization is implemented, and does not implement the field association logic optimization, and carrying out dynamic layout evaluation on the map edges and acquiring corresponding evaluation results, carrying out node positioning analysis after acquiring the evaluation results, acquiring corresponding analysis results, carrying out dynamic rendering and visualization accurate correction after acquiring the analysis results, acquiring corresponding correction results, and judging whether the dynamic map visualization generation is qualified or not according to the correction results, thereby improving the accuracy of the dynamic map visualization generation.

Inventors

  • HUANG ZHIHUA

Assignees

  • 北京邮科科技有限公司

Dates

Publication Date
20260505
Application Date
20260123

Claims (10)

  1. 1. The dynamic map visualization generation method based on the relationship label graph is characterized by comprising the following steps of: Carrying out field association logic dynamic correction on the key field and the auxiliary field, acquiring a corresponding correction result, judging whether field association logic optimization is implemented or not based on the correction result, and acquiring a corresponding optimization result if the field association logic optimization is implemented; If field association logic optimization is not implemented, carrying out dynamic layout evaluation on the map edges, acquiring corresponding evaluation results, and carrying out node positioning analysis and acquiring corresponding analysis results after acquiring the evaluation results; After the analysis result is obtained, carrying out accurate correction on dynamic rendering and visualization, obtaining a corresponding correction result, and judging whether the visualization generation of the dynamic map is qualified or not according to the correction result.
  2. 2. The dynamic map visualization generation method based on the relational tag map as set forth in claim 1, wherein the specific process of dynamic correction of the field association logic is as follows: after the key fields are in one-to-one correspondence with the auxiliary fields, obtaining completely matched key fields and auxiliary fields, and expressing the number of the completely matched key fields and auxiliary fields as the number of associated matching fields for reflecting the association identification precision degree of the key fields and the auxiliary fields; Representing the result of the ratio of the number of the associated matching fields to the total number of the fields as a time sequence association degree; Judging whether the time sequence association degree is in a time sequence association degree reference range or not; If the time sequence association degree is larger than the maximum value in the time sequence association degree reference range, marking the corresponding key field and the auxiliary field as strong association fields, and directly extracting and analyzing the semantic meaning of the relation tag; if the time sequence association degree is within the time sequence association degree reference range, marking the corresponding key field and the corresponding auxiliary field as medium association fields, and carrying out field association optimization; And if the time sequence association degree is smaller than the minimum value in the time sequence association degree reference range, marking the corresponding key field and the auxiliary field as weak association fields, and performing time sequence association supplementary operation.
  3. 3. The dynamic map visualization generation method based on the relational tag map as set forth in claim 2, wherein the specific process of field association optimization is as follows: carrying out feature splitting treatment on the key fields based on a natural language treatment method, and splitting core feature subfields and extension feature subfields of the key fields; Performing time stamp alignment on the auxiliary field, splitting the time dimension of the original auxiliary field according to a preset refinement period, and obtaining a time sequence sub-field with a time stamp; the preset refinement period is set in advance by a preset person; After the key field and the auxiliary field are split, the core characteristic subfields of the key field and the time sequence subfields are subjected to random combination matching, a new key field-auxiliary field combination pair is generated, and then relation tag semantic extraction analysis is performed.
  4. 4. The dynamic map visualization generation method based on the relationship tag map as claimed in claim 2, wherein the specific process of the time sequence association supplementary operation is as follows: Based on a preset language matching vector method, vectorizing the weak correlation field and a core entity in a preset knowledge graph to respectively obtain a weak correlation field vector set and a preset graph vector space; Based on the weight of the field semantic importance preset by a preset person, carrying out weighted fusion on the weak association field vector set to obtain a query semantic vector; Based on the query semantic vector, performing approximate semantic retrieval in a preset knowledge graph to acquire an entity related to the weak association field and association information thereof; For each candidate entity, calculating cosine similarity between the candidate entity vector and the query semantic vector as association strength; and screening the fields with the association strength larger than a preset matching threshold, taking the corresponding association information as key association supplementary information, supplementing the key association supplementary information to the weak association fields, and carrying out semantic extraction analysis on the relationship labels.
  5. 5. The dynamic map visualization generation method based on the relationship tag map as claimed in claim 4, wherein the specific process of semantic extraction and analysis of the relationship tag is as follows: Carrying out semantic analysis on the associated field obtained after the dynamic correction of the field association logic; Counting the number of relation labels with consistent semantics as the number of effective labels; The ratio of the effective label number to the total extracted label number is expressed as the semantic consistency rate of the relation labels; If the semantic consistency ratio of the relation labels is larger than the semantic consistency ratio of the preset relation labels, marking the corresponding set of the relation labels as a qualified label set, generating a relation label graph, and directly executing graph edge dynamic layout assessment; And otherwise, marking the corresponding relation tag set as a tag set to be optimized, and carrying out relation tag semantic correction.
  6. 6. The dynamic map visualization generation method based on the relationship tag map as claimed in claim 5, wherein the specific process of semantic modification of the relationship tag is as follows: Performing preliminary correction on the relation labels in the label set to be optimized based on a preset semantic mapping dictionary, specifically, extracting core vocabularies of the relation labels one by one; matching the relation labels with relation labels in a preset semantic mapping dictionary one by one; If the number of the words in the relation tag which is the same as the number of words in the relation tag in the preset semantic mapping dictionary is larger than or equal to the threshold number of the preset words, the corresponding words are expressed as core words, and a prompt is sent to a preset person to extract the core words in the relation tag, otherwise, the corresponding words are expressed as common words, and screening is carried out; Based on the common relation labels in the pre-recorded natural language processing task in the pre-set semantic mapping dictionary, extracting core words of the relation labels one by one, and carrying out accurate matching with the core words of the relation labels in the pre-set semantic mapping dictionary, if the matching is successful, namely the core words of the relation labels and the core words of the pre-set semantic mapping dictionary are successfully matched, directly calling corresponding standard expressions in the dictionary; If the relation tag semantic consistency rate is not matched with the core vocabulary and is partially matched with the core vocabulary, re-matching preset candidate correction rules in a preset mapping dictionary, and if the relation tag semantic consistency rate lifting quantity is still smaller than a preset lifting threshold, sending a relation tag semantic matching exception prompt to preset personnel; The method comprises the steps of calling context association strength to verify the relation labels, specifically, marking the fields, corresponding to the preset semantic mapping dictionary, of each relation label one by one as normalized fields, and extracting core entities in the normalized fields; Counting co-occurrence frequency of core vocabularies corresponding to the relation labels obtained after preliminary correction and core vocabularies corresponding to the extracted core entities in the candidate entities, and counting total word frequency of the core vocabularies; the ratio of the co-occurrence frequency to the total word frequency is expressed as the context association strength of the relation tag; If the context association strength is greater than a preset context association strength threshold and the relation tag semantic consistency rate lifting amount is greater than a preset lifting threshold, confirming that correction is effective, reserving the corrected relation tag, supplementing the corrected relation tag to a tag set to be optimized, and executing map edge dynamic layout assessment; Otherwise, a relation tag semantic matching exception prompt is sent to a preset person.
  7. 7. The dynamic map visualization generation method based on the relationship label graph as claimed in claim 6, wherein the specific process of map edge dynamic layout assessment is as follows: acquiring actual semantics of the map edges at the initial time of the dynamic layout evaluation of the map edges and the corresponding time after a preset period; the ratio of the corresponding quantity of the actual semantics of the two is expressed as the edge definition accuracy; the specific acquisition process of the actual semantics of the map edge is as follows: Inputting a relationship label graph of a pre-defined edge semantic into a preset knowledge graph embedding model; extracting a context description from the original interaction records of each edge in the relationship tag map; using preset semantic categories in the relation tag graph as relation identifications, and training a classifier by using the text semantic vectors on the edges and the contexts and the corresponding preset semantic tags; classifying each side based on a classifier in a dynamic map corresponding to the relationship label graph to obtain the actual semantics of the side; if the edge definition accuracy is greater than the preset edge definition accuracy, performing node positioning analysis; If the edge definition accuracy is smaller than or equal to the preset edge definition accuracy, identifying an edge semantic overlapping region of the dynamic map based on a preset edge semantic conflict detection algorithm, and performing node positioning analysis on nodes in the dynamic map after the edge semantic overlapping region is screened out; The side semantic overlapping area refers to an area where the semantic description of the relationship label atlas is inconsistent or repeated when the relationship label atlas is obtained by comparing the atlas side dynamic layout evaluation initial time with the relationship label atlas obtained by the atlas side dynamic layout evaluation final time, and the positions of the sides are overlapped or covered in the relationship label atlas space layout; the specific process of the node positioning analysis is as follows: the coincidence quantity of the nodes in the actual dynamic map and the nodes obtained after the dynamic layout evaluation of the map edges is expressed as the node position coincidence ratio; Judging whether the node position coincidence degree is larger than or equal to a node position coincidence standard value; if the node position coincidence degree is larger than or equal to the node position coincidence standard value, executing dynamic rendering and visual accurate correction; If the node position coincidence degree is smaller than the node position coincidence standard value, correcting the actual layout position of the node based on a preset position correction algorithm; and judging whether the node position coincidence degree obtained again after correction is larger than or equal to a node position coincidence standard value, if so, executing dynamic rendering and visual accurate correction, otherwise, sending a node layout correction abnormality prompt to a preset person.
  8. 8. The dynamic map visualization generation method based on the relationship label graph as claimed in claim 6, wherein the specific process of map edge dynamic layout assessment is as follows: In the preset dynamic spectrum, taking the nodes of the preset dynamic spectrum as reference points, acquiring the coordinates of the group of reference points in the current dynamic spectrum and the corresponding node coordinates of the current dynamic spectrum, and representing a set of one-to-one correspondence of the node coordinates of the group of reference points and the corresponding node coordinates of the current dynamic spectrum as a point pair set; calculating coordinate transformation parameters of the point pair set based on a spatial registration algorithm; converting coordinates of the node coordinates corresponding to the current dynamic map in a one-to-one correspondence manner based on the obtained coordinate conversion parameters; Acquiring the node position precision after transformation, and if the node position precision is greater than or equal to a node position precision reference value, executing dynamic rendering and visual accurate correction; if the node position accuracy is smaller than the node position accuracy reference value, sending a node position accuracy abnormality prompt to a preset person.
  9. 9. The dynamic map visualization generation method based on the relational tag map as set forth in claim 8, wherein the specific process of dynamic rendering and visualization precision correction is as follows: the ratio of the number of the accurate identification tags to the total key relationship tags is expressed as relationship tag accuracy for evaluating the visual transmission accuracy of the relationship tags; the ratio of the visual presentation frame number to the total rendering frame number in the dynamic map is expressed as the map dynamic rendering accuracy; Judging whether the accuracy of the relation tag is greater than or equal to the accuracy of a preset relation tag and judging whether the dynamic rendering accuracy of the map is greater than or equal to the dynamic rendering accuracy of the preset map; if the relation label accuracy is greater than or equal to the preset relation label accuracy and the map dynamic rendering accuracy is greater than or equal to the preset map dynamic rendering accuracy, sending a dynamic map visualization generation qualification prompt to preset personnel; Otherwise, sending an abnormal prompt generated by the dynamic map visualization to a preset person.
  10. 10. The dynamic map visualization generation system based on the relationship label graph is characterized by comprising a field association logic dynamic correction module, a map edge dynamic layout evaluation module and a dynamic rendering and visualization accurate correction module; The field association logic dynamic correction module is used for carrying out field association logic dynamic correction on the key field and the auxiliary field, acquiring a corresponding correction result, judging whether field association logic optimization is implemented or not based on the correction result, and acquiring a corresponding optimization result if the field association logic optimization is implemented; The map side dynamic layout evaluation module is used for carrying out map side dynamic layout evaluation when field association logic optimization is not implemented, acquiring corresponding evaluation results, carrying out node positioning analysis after the evaluation results are acquired, and acquiring corresponding analysis results; The dynamic rendering and visualization accurate correction module is used for carrying out dynamic rendering and visualization accurate correction after the analysis result is obtained, obtaining a corresponding correction result, and judging whether the dynamic map visualization generation is qualified or not according to the correction result.

Description

Dynamic map visualization generation method and system based on relational tag graph Technical Field The invention relates to the technical field of dynamic map generation, in particular to a dynamic map visualization generation method and system based on a relationship label map. Background With the rapid development and wide application of big data technology, the data scale grows exponentially, the data types become more complex, the data containing massive entities and the association relation between the entities becomes a core asset of the information age, in this context, the relation tag graph becomes a core data model for representing the association structure of the entities, the technical advantages of structural description of the complex relation, multi-dimensional tag semantic bearing and clear tracing of association paths are fully exerted, the time-sequence evolution characteristics of the entity relation can be intuitively presented, the method becomes a core technical means in the fields of knowledge graph construction and the like, the dynamic graph visualization can accurately mine core value information such as the thermal change of the area, and the method has important significance for enterprise data driving decision and market refined operation. The traditional map visualization relies on manual drawing or static visualization, and although the requirement of visual display of simple relations can be met, short plates with low relation label matching precision, untimely dynamic evolution characteristic capture, poor suitability of large-scale data and difficult realization of relation evolution process tracing exist, and the requirements of entity relation refinement and dynamic monitoring in complex data in a new period cannot be met, so that the precision of building the relation label map is a key premise for guaranteeing the dynamic map visualization precision. In order to achieve the purpose of accurately generating the dynamic map visualization, the prior art generally directly collects real-time multi-source data (such as commodity ID (Identifier) of sales data, purchasing user and transaction time) by calling a platform open API (OpenApplication Programming Interface ) and the like, after collection is completed, cleans and screens the original multi-source data, eliminates unqualified data such as serious multi-source data (such as key field deletion) and irregular format (such as date format confusion), and based on the obtained multi-source data, carries out structural definition on the entity such as the attribute of the commodity, including commodity ID, class label, specification parameter, brand attribution and the like, further converts the original multi-source data into a relation label with definite semantics, namely the original multi-source data is used as a basis, converts discrete and uncorrelated numerical value type data or class type data in the multi-source data into a structured label with definite logic association, and comprises a relation time period, a relation type and the like, wherein the relation label takes the entity as a node, takes time-sequential association as an edge, and extracts key fields (such as commodity classification, attribution a hierarchy, a service attribute map and the like to generate a relation label. In order to intuitively present the graph of the relationship tag graph, a self-adaptive visual layout algorithm (such as a force guiding algorithm) is generally adopted to calculate and optimize the spatial arrangement of nodes and edges in the relationship tag graph, so that the internal structure of a network is clearly disclosed to obtain a static graph, and finally, according to the requirements of a specific application scene on visual precision and instantaneity, a rendering engine drives smooth transition and evolution of the graph of the relationship tag graph among different time slices, wherein the rendering engine drives layout data of a time-series graph to a special visual component or a graph library of a smooth transition animation, such as a D3.js frame and a WebGL frame, so that the generation of the time-series evolution animation is completed, and finally, a qualified and standardized dynamic graph closely fitting the analysis requirement is output, and further, the visual accurate generation of the dynamic graph is realized. In the process of generating a relationship label based on multi-source data fusion, in the prior art, key fields (such as commodity classification, hierarchy attribution, service association attribute and the like in sales data) in the multi-source data are subjected to labeling extraction, when a relationship label graph is generated by using a structured table (such as an order table and a commodity table in the sales data) as a main storage carrier, time sequence information (such as order generation time in the sales data) is often used as an auxiliary field, because attribute field