Adaptive questionnaire optimization method based on answer path analysis

Abstract

The invention relates to the technical field of park wisdom management, and discloses a self-adaptive questionnaire optimization method based on answer path analysis, which comprises the following steps: acquiring initial answer path data, extracting path nodes, performing clustering analysis, acquiring an initial path mode, uniformly expressing the initial answer path data in a mode form, and obtaining an answer path model; constructing a path topological structure diagram according to the initial path mode, identifying a path long-tail mode and determining a sparse path based on the diagram, then correcting the path topological structure diagram according to the path long-tail mode and the sparse path, and executing path mode mapping on the corrected path topological structure diagram to obtain a corrected path mode; and finally, reconstructing the structure and content of the self-adaptive questionnaire according to the corrected path mode, outputting the optimized self-adaptive questionnaire, enabling the questionnaire behavior to follow the corrected path mode, and reducing the interference of the path long-tail mode and the sparse path from the source.

Description

Technical Field

[0001] This invention relates to the field of smart park management technology, and more specifically, to an adaptive questionnaire optimization method based on answer path analysis. Background Technology

[0002] As the application scale and functional complexity of smart management systems in diversified service scenarios increase, questionnaire surveys and path analysis targeting the service needs of enterprises in the park have gradually become important means of park management. Existing technologies typically use rule-based filtering, frequency statistics, threshold screening, and static path analysis to optimize the questionnaire structure and content. Such methods can achieve good optimization results under the premise of high data stability, uniform data distribution, and insignificant changes in path patterns.

[0003] However, in actual park services, the questionnaire path data exhibits a significant long-tail distribution, and the service needs of park enterprise users change frequently. These factors cause a decrease in the stability of optimization strategies under low-probability path patterns using traditional methods, thereby affecting the robustness of the park's intelligent management system. Therefore, the technical problem that needs to be solved is how to ensure the robustness of the park's intelligent management system while reducing the interference of long-tail distribution in questionnaire path data and improving the stability of low-probability path optimization strategies.

[0004] In view of this, the present invention proposes an adaptive questionnaire optimization method based on answer path analysis to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned shortcomings of existing technologies, this invention provides an adaptive questionnaire optimization method based on answer path analysis.

[0006] To achieve the above objectives, the present invention provides the following technical solution: Firstly, an adaptive questionnaire optimization method based on answer path analysis is provided, including: Obtain initial answer path data, extract path nodes based on the initial answer path data, and perform cluster analysis on the path nodes to obtain the initial path pattern; Construct a path topology graph based on the initial path pattern, identify long-tail patterns of paths and determine sparse paths based on the path topology graph; The path topology graph is corrected based on the long-tail pattern and sparse path, and the path pattern mapping is performed on the corrected path topology graph to obtain the corrected path pattern. Reconstruct the structure and content of the adaptive questionnaire based on the modified path pattern, and output the optimized adaptive questionnaire.

[0007] In one possible implementation, path nodes are extracted from the initial answer path data, and cluster analysis is performed on the path nodes to obtain the initial path pattern, including: Extract each answer node from the initial answer path data, record the position order of each answer node in the answer path, and generate a set of node order structures; Based on the node sequence structure set, determine the preorder or subsequent adjacency relationship between each answer node and other answer nodes to form a node adjacency relationship structure; Based on the node adjacency relationship structure, determine the set of nodes with the same adjacent node combinations, and form a node combination structure; The answer nodes are clustered according to the node combination structure, and the clustered node combinations are determined as the initial path pattern.

[0008] In one possible implementation, the set of nodes with the same adjacent node combinations is determined based on the node adjacency relationship structure, and a node combination structure is formed, including: Based on the node adjacency structure, extract the specific preceding and subsequent adjacent nodes of each answer node, and generate the node adjacency combination structure. The number of times adjacent node combinations appear in all answer paths is counted to form a node combination frequency table; Based on the node combination frequency table, determine the frequency distribution attribute of adjacent combinations of each node; The node combination structure is determined based on the adjacent combinations of nodes with frequency distribution attributes.

[0009] In one possible implementation, long-tail patterns of paths are identified and sparse paths are determined based on the path topology graph, including: Extract node connection segments consisting of consecutive answer nodes from the path topology graph and generate a set of node connection segments; The frequency of each node connection segment in the set of node connection segments in all answer paths is counted to form a node segment frequency table; The node fragment sparsity threshold is determined based on the node fragment frequency table; Based on the node fragment sparsity threshold, node connection fragments are filtered from the node fragment frequency table. The filtered node connection fragments are used as long-tail node connection fragments, and the long-tail node connection fragments are determined as the path long-tail pattern. Sparse paths are determined based on the node connection structure of the long-tail node connection segments.

[0010] In one possible implementation, node connection fragments are filtered from a node fragment frequency table based on a node fragment sparsity threshold, and the node connection fragments are determined to be long-tail node connection fragments, including: Sort the occurrence frequency of node connection segments to form a node segment sorting sequence; Determine the node segment sorting threshold in the node segment sorting sequence; Identify node connection segments in the sorted sequence that are located after the node segment sorting threshold to form candidate long-tail segments; Analyze candidate long-tail segments to determine the structural differences in node-connected segments; Based on structural differences, long-tailed node connection segments are selected from candidate long-tailed segments and identified.

[0011] In one possible implementation, the path topology graph is modified based on the long-tail pattern and sparse paths, including: The long-tail node set is determined based on the path long-tail pattern, and the long-tail node set is mapped to the corresponding node in the path topology graph to form a long-tail node structure. Determine the node order structure based on the node connection order of the sparse path; By comparing the long-tail node structure with the node sequence structure, the connection relationship of the nodes to be adjusted is determined, and the node adjustment structure is formed. Adjust the node connection relationships in the path topology diagram based on the node adjustment structure; The path topology diagram after the structural adjustment is completed is determined as the corrected path topology diagram.

[0012] In one possible implementation, the connection relationships of the nodes to be adjusted are determined to form a node adjustment structure, including: Extract all the long-tail node connections involved based on the long-tail node structure to form a node connection set; The sparse node structure is determined based on the connection order of nodes in the sparse path; The node connection set is structurally compared with the sparse node structure to determine the node connection relationships where there are differences in the connection order. The determined differences in the node connections form a node adjustment structure.

[0013] In one possible implementation, the node connection relationships in the path topology graph are structurally adjusted based on the node adjustment structure, including: Based on the node adjustment structure, determine the nodes that need to be adjusted and their associated node connections to form a node positioning structure; Based on the node positioning structure, the differences in node connection directions are identified to form a connection direction structure; Based on the connection direction structure, conflicting node connections in the path topology graph with conflicting directions are deleted to obtain the target node connection relationship. Based on the initial node connection relationships of the initial path pattern, the connection relationships of the target nodes are supplemented to obtain the path topology diagram after structural adjustment.

[0014] In one possible implementation, path pattern mapping is performed on the modified path topology graph to obtain modified path patterns, including: Extract the corrected node connection relationship of each answer node from the corrected path topology graph to generate a node connection mapping structure; Determine the transition relationships between nodes based on the node connection mapping structure, and generate the node transition structure; Identify node segments with repeated connections based on the node transition structure, and form a node segment merging structure; Node mapping relationships are constructed based on the node fragment merging structure, and corrected path patterns are generated based on the node mapping relationships.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention, after acquiring initial answer path data, extracts path nodes and performs cluster analysis to form initial path patterns. This merges similar paths within the same structure, weakening the impact of scattered noise on subsequent judgments. Based on the initial path patterns, a path topology graph is constructed, explicitly presenting the transition relationships between path nodes. Then, based on this graph, long-tail patterns are identified and sparse paths are determined, with low-frequency branches individually labeled at the structural level. Subsequently, the path topology graph is corrected based on the long-tail patterns and sparse paths, and path pattern mapping is performed on the corrected graph to obtain a corrected path pattern. This ensures that the main paths maintain stable representation, and the influence of sparse paths is limited and can be processed independently. Finally, the structure and content of the adaptive questionnaire are reconstructed based on the corrected path pattern, and an optimized adaptive questionnaire is output. This allows the questionnaire presentation and navigation to operate with the corrected path pattern as the framework, thereby reducing interference from the long-tail distribution of questionnaire path data, improving the stability of strategies under low-probability paths, and enhancing the overall robustness of the park's intelligent management system. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating an adaptive questionnaire optimization method based on answer path analysis in this invention. Figure 2 This is a schematic diagram of the structure of an adaptive questionnaire optimization system based on answer path analysis in this invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the embodiments of this application will be further described in detail below with reference to the accompanying drawings. In this application, the terms "first" and "second" are used to distinguish the same or similar items with essentially the same function. It should be understood that there is no logical or temporal dependency between "first" and "second", nor is there any limitation on the quantity or execution order. In this application, the term "at least one" means one or more, and "multiple" means two or more.

[0018] It should be noted that the information (including but not limited to user device information, user personal information, etc.), data (including but not limited to data used for analysis, data stored, data displayed, etc.) and signals involved in this application are all authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0019] Example 1 Please see Figure 1 As shown, this embodiment discloses an adaptive questionnaire optimization method based on answer path analysis, including: S10: Obtain initial answer path data, extract path nodes based on the initial answer path data, and perform cluster analysis on the path nodes to obtain the initial path pattern; In this embodiment, obtaining initial answer path data and extracting path nodes based on the initial answer path data refers to abstracting the historically accumulated questionnaire answer records in the park's smart management system into answer paths that are sequentially connected by multiple answer nodes. Each answer node can correspond to a specific questionnaire question or a combination of a question and an option. By uniformly extracting these answer nodes from all answer paths and recording their occurrence positions, the originally scattered answering behaviors can be transformed into a set of path nodes with unified coding rules. On this basis, cluster analysis is performed on the path nodes to obtain initial path patterns. Essentially, this involves grouping nodes with similar sequential connections and positional characteristics into one category, using a small number of representative path patterns to characterize the common structure of a large number of real paths, thereby providing a structured foundation for subsequent identification of long-tail path patterns and adjustment of the adaptive questionnaire structure.

[0020] For example, we can assume that there are three typical answer paths in a certain park, which can be abstracted into the following node sequence structure: the first path is A, B, C, D; the second path is A, B, E, D; and the third path is A, F, C, D. Here, A, B, C, D, E, and F represent different answer nodes. By extracting the above nodes from the initial answer path data and recording their position order in their respective paths, we can form a set of node sequence structures with three rows and four columns. The rows represent different answer paths, and the columns represent the positions of the answer nodes in the paths. Then, based on this sequence information, we can count the combinations of adjacent nodes before and after each node, and cluster the nodes accordingly. For example, nodes that are "preceded by A and followed by D" are considered as one class, thus forming several initial path patterns. Compared with the method of directly configuring rules on the original path set, this initial path pattern can more compactly reflect the structural characteristics of the mainstream answer paths.

[0021] Path nodes are extracted from the initial answer path data, and cluster analysis is performed on the path nodes to obtain the initial path pattern, including: Extract each answer node from the initial answer path data, record the position order of each answer node in the answer path, and generate a set of node order structures; Based on the node sequence structure set, determine the preorder or subsequent adjacency relationship between each answer node and other answer nodes to form a node adjacency relationship structure; Based on the node adjacency relationship structure, determine the set of nodes with the same adjacent node combinations, and form a node combination structure; The answer nodes are clustered according to the node combination structure, and the clustered node combinations are determined as the initial path pattern.

[0022] It should be noted that in the specific process of extracting path nodes from the initial answer path data and performing cluster analysis on the path nodes, the answer nodes can be extracted one by one from each answer path and the sequential position of the nodes in the path can be recorded to generate a node sequence structure set. In the node sequence structure set, each element contains at least the answer path identifier, the answer node identifier, and the position number of the node. Through this set, those skilled in the art can completely reconstruct the node sequence of any original answer path, and it can also be used for subsequent statistical analysis of the adjacency relationship between nodes.

[0023] Subsequently, based on the node sequence structure set, for each answer node, its preceding and subsequent adjacent nodes when it appears in different paths are determined, forming a node adjacency relationship structure. Based on the node adjacency relationship structure, the set of answer nodes with the same preceding and subsequent node combinations are identified, forming a node combination structure. Finally, the answer nodes are clustered according to the node combination structure, and the node combinations with the same or highly similar contextual connection relationships in each category are determined as the initial path pattern, thereby extracting the main structure of questionnaire answering behavior at the path pattern level.

[0024] Based on the node adjacency structure, determine the set of nodes with the same combination of adjacent nodes, and form a node combination structure, including: Based on the node adjacency structure, extract the specific preceding and subsequent adjacent nodes of each answer node, and generate the node adjacency combination structure. The number of times adjacent node combinations appear in all answer paths is counted to form a node combination frequency table; Based on the node combination frequency table, determine the frequency distribution attribute of adjacent combinations of each node; The node combination structure is determined based on the adjacent combinations of nodes with frequency distribution attributes.

[0025] In this embodiment, determining the set of nodes with the same adjacent node combinations based on the node adjacency relationship structure and forming a node combination structure is to characterize the context of the answer nodes at a finer granular level, thereby providing more explicit structural features for subsequent clustering analysis.

[0026] In practical implementation, the specific preceding and subsequent adjacent nodes of each answer node can be extracted based on the node adjacency relationship structure. The set of preceding and subsequent adjacent nodes is combined to form a node adjacency combination structure. Then, the occurrence frequency of the node adjacency combination structure in all answer paths is counted to form a node combination frequency table. Based on the node combination frequency table, the frequency distribution attribute of each node adjacency combination is determined, thereby identifying which adjacency combinations belong to high-frequency typical patterns and which belong to low-frequency exception patterns. Finally, a node combination structure is formed based on the node adjacency combinations with frequency distribution attributes, so that subsequent node clustering considers both structural similarity and occurrence frequency.

[0027] For example, in this instance, we can count the adjacent combinations of each node in the aforementioned node sequence structure set. If node B exhibits the adjacent characteristic of "preceding node is A and subsequent node is C or E" in multiple paths, then "A→B→C" and "A→B→E" can be recorded as two types of adjacent combinations in the node adjacent combination structure, and the number of times these two combinations appear in all answer paths can be counted.

[0028] For example, if A→B→C appears twice and A→B→E appears once, the following node combination frequency table fragment is formed: the first row is A→B→C, which appears twice; the second row is A→B→E, which appears once. Based on this, combinations with an occurrence frequency of 2 or more can be marked as mainstream adjacent combinations, and combinations with a lower occurrence frequency but similar structure to mainstream combinations can be marked as secondary combinations. Then, these node adjacent combinations with frequency distribution attributes are merged into node combination structures to guide the subsequent classification of B and nodes with similar contexts into the same initial path pattern, thereby highlighting the structural stability of high-frequency paths without losing long-tail information.

[0029] S20: Construct a path topology graph based on the initial path pattern, identify long-tail path patterns based on the path topology graph, and determine sparse paths; In this embodiment, it can be understood that constructing a path topology graph based on the initial path pattern means expressing the initial path pattern obtained by clustering as a directed graph, using answer nodes as nodes in the graph, and using the sequential connection relationship between different answer nodes as directed edges, thereby obtaining a path topology graph that can cover the mainstream answer path structure. Based on this, identifying long-tail patterns and determining sparse paths in the path topology graph means extracting node connection segments composed of consecutive answer nodes in the directed graph, counting the occurrence frequency of these node connection segments in all answer paths and determining the sparsity threshold, identifying node connection segments with low occurrence frequency and distinctive structure as long-tail node connection segments, and determining the corresponding sparse paths accordingly.

[0030] This method can identify long-tail patterns in questionnaire answering paths at the path level, thus providing a basis for subsequent path structure correction and adaptive questionnaire optimization, thereby reducing the interference of long-tail path data on the overall path pattern.

[0031] For example, we can assume that the path topology diagram constructed based on the initial path pattern contains the following node connection segments consisting of consecutive answer nodes: A→B→C, A→B→D, B→E→F, C→G→H. By counting the occurrences of these node connection segments across all answer paths, we can obtain the following three-row, two-column data structure: A→B→C, occurrences 50; A→B→D, occurrences 5; C→G→H, occurrences 2. Analyzing this node segment frequency table, we can consider A→B→D and C→G→H, which have significantly lower occurrences, as sparse segments. Furthermore, we can combine these sparse segments into sparse paths. This ensures that the mainstream path A→B→C remains the primary structure while enabling the system to explicitly identify and separately process long-tail paths containing A→B→D or C→G→H, thus avoiding completely ignoring low-probability but business-significant path patterns during questionnaire optimization.

[0032] Identifying long-tail patterns and determining sparse paths based on path topology graphs includes: Extract node connection segments consisting of consecutive answer nodes from the path topology graph and generate a set of node connection segments; The frequency of each node connection segment in the set of node connection segments in all answer paths is counted to form a node segment frequency table; The node fragment sparsity threshold is determined based on the node fragment frequency table; Based on the node fragment sparsity threshold, node connection fragments are filtered from the node fragment frequency table. The filtered node connection fragments are used as long-tail node connection fragments, and the long-tail node connection fragments are determined as the path long-tail pattern. Sparse paths are determined based on the node connection structure of the long-tail node connection segments.

[0033] It should be noted that in the specific process of identifying long-tail patterns and determining sparse paths based on path topology graphs, we can first extract node connection segments composed of consecutive answer nodes from the path topology graph. A node sequence with a fixed length or within a preset range is taken as a connection segment element, and all connection segments are aggregated to form a node connection segment set. Then, for each node connection segment in the node connection segment set, its occurrence frequency is counted in all initial answer paths to form a node segment frequency table. Next, based on the overall distribution of the occurrence frequency of each connection segment in the node segment frequency table, a node segment sparsity threshold is determined. Subsequently, based on the node segment sparsity threshold, node connection segments with occurrence frequencies below the threshold are selected from the node segment frequency table. These segments are considered long-tail node connection segments and determined as long-tail patterns. Finally, based on the connection structure of nodes in the long-tail node connection segments, one or more continuous sparse paths are combined along the connectable directed relationships between segments, thus characterizing the long-tail portion of the questionnaire answer path at the structural level.

[0034] Based on the node fragment sparsity threshold, node connection fragments are filtered from the node fragment frequency table, and those fragments are identified as long-tail node connection fragments, including: Sort the occurrence frequency of node connection segments to form a node segment sorting sequence; Determine the node segment sorting threshold in the node segment sorting sequence; Identify node connection segments in the sorted sequence that are located after the node segment sorting threshold to form candidate long-tail segments; Analyze candidate long-tail segments to determine the structural differences in node-connected segments; Based on structural differences, long-tailed node connection segments are selected from candidate long-tailed segments and identified.

[0035] In this embodiment, when filtering node connection segments from the node segment frequency table and determining long-tail node connection segments based on the node segment sparsity threshold, the occurrence frequency of the node segment frequency table can be sorted from high to low to form a node segment sorting sequence. Then, a node segment sorting threshold is determined in the sorting sequence based on a preset ratio or inflection point feature. The low-probability node connection segments that are located after the sorting threshold and whose occurrence frequency is at the tail end are taken as candidate long-tail segments. Then, the candidate long-tail segments are further analyzed to determine their structural difference features.

[0036] For example, it can be determined whether these segments deviate significantly from the mainstream path segments in terms of the number of nodes, node type, or connection direction. When some candidate long-tail segments are only minor variations of the mainstream segments in terms of structure, they can be eliminated. Meanwhile, those candidate long-tail segments that have representative differences in structure and whose occurrence frequency is consistently in the low range are identified as long-tail node connection segments. This way, the interference of pure noise long tails is weakened while the truly valuable long-tail path structures are preserved.

[0037] For example, the node connection segments in the aforementioned node segment frequency table can be sorted by their frequency of occurrence, resulting in the following node segment sorting sequence: A→B→C, 50 occurrences; B→C→D, 40 occurrences; A→B→D, 5 occurrences; C→G→H, 2 occurrences. In this sorting sequence, the top two segments can be considered as mainstream segments, and the sorting threshold is set between the second and third positions. A→B→D and C→G→H, which are located after the sorting threshold, are considered as candidate long-tail segments. Then, structural difference analysis of these two candidate long-tail segments reveals that A→B→D differs from the mainstream segment A→B→C only in the last node, while C→G→H introduces new nodes G and H. If the system determines that it needs to prioritize abnormal paths that introduce new node combinations, then only C→G→H can be identified as a long-tail node connection segment. This forms a set of long-tail node connection segments that considers both frequency of occurrence and structural difference characteristics, providing a concise and representative input for subsequent path topology correction based on long-tail patterns.

[0038] S30: Correct the path topology graph based on the long-tail pattern and sparse path, and perform path pattern mapping on the corrected path topology graph to obtain the corrected path pattern. Furthermore, modifying the path topology diagram based on long-tail patterns and sparse paths involves, after identifying which node connection segments belong to long-tail patterns and the sparse paths formed by combinations of these long-tail segments, introducing this low-probability but representative structural information into the path topology diagram. This involves locating and adjusting node connections with directional conflicts, missing connections, or unreasonable structures. The modified path topology diagram retains the stable backbone structure of the mainstream paths while also reasonably absorbing effective connections from long-tail paths. Based on this, path pattern mapping is performed on the modified path topology diagram to regenerate a modified path pattern reflecting the latest structural adjustments. This provides a more robust path pattern foundation for the subsequent structural and content reconstruction of the adaptive questionnaire.

[0039] The path topology graph is revised based on the long-tail pattern and sparse path, including: The long-tail node set is determined based on the path long-tail pattern, and the long-tail node set is mapped to the corresponding node in the path topology graph to form a long-tail node structure. Determine the node order structure based on the node connection order of the sparse path; By comparing the long-tail node structure with the node sequence structure, the connection relationship of the nodes to be adjusted is determined, and the node adjustment structure is formed. Adjust the node connection relationships in the path topology diagram based on the node adjustment structure; The path topology diagram after the structural adjustment is completed is determined as the corrected path topology diagram.

[0040] It should be understood that in the process of correcting the path topology diagram based on the long-tail pattern and sparse path, the first step is to determine the complete set of long-tail nodes involved based on the long-tail node connection segments contained in the long-tail pattern, and map this set of long-tail nodes to the corresponding node positions in the path topology diagram to form a long-tail node structure that identifies the distribution of long-tail nodes. At the same time, one or more node sequence structures with long-tail nodes as the main branches are extracted based on the node connection order in the sparse path. The long-tail node structure is then compared with the node sequence structure to identify node connection relationships in the topology diagram that are inconsistent with or missing from the connection order of the sparse path. These node connection relationships to be adjusted are abstracted into node adjustment structures. Subsequently, the relevant connections in the path topology diagram are corrected in direction, redundantly deleted, or supplemented as necessary based on the node adjustment structures. Finally, the path topology diagram with the adjusted structure is obtained and determined as the corrected path topology diagram. This ensures that the corrected topology respects the connection order of the real sparse path while avoiding excessive disturbance to the overall topology caused by long-tail noise.

[0041] Determine the connection relationships of the nodes to be adjusted to form the node adjustment structure, including: Extract all the long-tail node connections involved based on the long-tail node structure to form a node connection set; The sparse node structure is determined based on the connection order of nodes in the sparse path; The node connection set is structurally compared with the sparse node structure to determine the node connection relationships where there are differences in the connection order. The determined differences in the node connections form a node adjustment structure.

[0042] In this embodiment, when determining the connection relationships of the nodes to be adjusted and forming the node adjustment structure, all long-tail nodes and their connections can be extracted based on the long-tail node structure to form a node connection set. This node connection set reflects the existing connection methods of long-tail nodes in the current path topology graph. At the same time, a sparse node structure is constructed based on the connection order of nodes in the sparse path. This sparse node structure reflects the ideal connection order that long-tail nodes should present on the real low-probability path. By comparing the node connection set with the sparse node structure one by one, it is possible to identify which node pairs have the opposite connection direction in the topology graph to the direction in the sparse path, or which are missing in the topology graph but exist in the sparse path. These node connection relationships with different connection orders are collected in a unified manner to form a node adjustment structure, so that the subsequent structure adjustment steps can modify these different nodes in a targeted manner without affecting the stable connection relationships between the mainstream path nodes.

[0043] For example, we can assume that there is a node connection segment A→B→C in the path topology graph, and the connection order observed in a certain sparse path is A→C→B. Based on the long-tail node structure, we can extract the two long-tail node connection relationships A→B and B→C to form a node connection set. Based on the sparse path, we can obtain the sparse node structures A→C and C→B. When comparing the two structures, we can find that there is a difference in the connection direction between B→C and C→B, and the A→C connection relationship is missing in the topology graph. Therefore, the difference in the node connection relationships corresponding to B→C and A→C can be included in the node adjustment structure, providing a precise adjustment target for subsequent correction of connection direction and missing edges.

[0044] The structural adjustment of node connections in the path topology graph is performed based on node adjustments, including: Based on the node adjustment structure, determine the nodes that need to be adjusted and their associated node connections to form a node positioning structure; Based on the node positioning structure, the differences in node connection directions are identified to form a connection direction structure; Based on the connection direction structure, conflicting node connections in the path topology graph with conflicting directions are deleted to obtain the target node connection relationship. Based on the initial node connection relationships of the initial path pattern, the connection relationships of the target nodes are supplemented to obtain the path topology diagram after structural adjustment.

[0045] It should be noted that when adjusting the node connection relationships in the path topology graph based on the node adjustment structure, the nodes to be adjusted and all their associated connections can be locked first based on the node adjustment structure to form a node positioning structure. This allows for precise location of the local subgraphs that need modification in complex topologies. Then, based on the node positioning structure, the differences between the directions of each connection edge and the ideal connection directions in the sparse node structure are analyzed to generate a connection direction structure. In the connection direction structure, node connection relationships with conflicting directions are identified and deleted. This results in target node connection relationships that retain only those with consistent directions or those that are not yet complete. Finally, referring to the initial node connection relationships in the initial path pattern, the target node connection relationships are supplemented without destroying the overall topology of the mainstream path. For example, directed edges are added to connections that are missing but exist in the sparse path. This results in a path topology graph after structural adjustment, where the corrected graph absorbs the actual connection order in the sparse path while retaining the backbone structure of the mainstream path.

[0046] For example, for the aforementioned differing connection relationships B→C and C→B, the nodes involved can be identified as B and C in the node positioning structure, and B→C can be marked as a directional conflict connection in the connection direction structure. Subsequently, during the structural adjustment process, the directed edge B→C can be deleted, while C→B can be retained or added. At the same time, if the connection relationship A→C exists in the initial path pattern but is not yet reflected in the current topology graph, the connection relationship of the target node can be supplemented according to the initial path pattern by adding the connection edge A→C. Ultimately, the adjusted path topology graph is reflected as A→C→B in the local structure, which is consistent with the actual connection order observed in the sparse path, thereby improving the adaptability of the path topology structure to long-tail paths.

[0047] Perform path pattern mapping on the corrected path topology graph to obtain the corrected path patterns, including: Extract the corrected node connection relationship of each answer node from the corrected path topology graph to generate a node connection mapping structure; Determine the transition relationships between nodes based on the node connection mapping structure, and generate the node transition structure; Identify node segments with repeated connections based on the node transition structure, and form a node segment merging structure; Node mapping relationships are constructed based on the node fragment merging structure, and corrected path patterns are generated based on the node mapping relationships.

[0048] Understandably, after adjusting the path topology graph, it is necessary to re-extract the connection relationships between nodes from the revised path topology graph and generate a node connection mapping structure. This node connection mapping structure can be regarded as the set of incoming and outgoing edges of each answer node in the revised topology. Based on this, the transition relationships between nodes are deduced according to the node connection mapping structure, and a node transition structure is generated to describe the reachability and path length information from one node to another. Subsequently, node segments that appear repeatedly in different paths are identified based on the node transition structure, and these repeated segments are summarized into a node segment merging structure. Node mapping relationships are constructed through the node segment merging structure, mapping structurally equivalent or highly similar segments to the same type of path pattern. Finally, a revised path pattern is generated based on the node mapping relationship, so that the revised path pattern can accurately reflect the updated path structure characteristics after considering long-tail patterns and sparse paths. Compared with the initial path pattern, it more comprehensively covers the structural trade-offs between mainstream paths and typical long-tail paths, providing a more stable pattern basis for subsequent adaptive questionnaire optimization.

[0049] S40: Reconstruct the structure and content of the adaptive questionnaire based on the modified path pattern, and output the optimized adaptive questionnaire.

[0050] Furthermore, reconstructing the structure and content of the adaptive questionnaire based on the modified path pattern means that after obtaining the modified path pattern, the order of questions, branching paths, and key answer nodes on high-frequency and typical long-tail paths reflected in the modified path pattern are used as the design basis to adjust and trim the question arrangement, branching logic, and question set of the original adaptive questionnaire. This preserves and strengthens the path segments that are more likely to be accessed in the actual answering process, while weakening or removing structurally redundant or abnormal paths caused only by noisy long tails. In this example, the starting questions commonly included in the modified path pattern can be used as the questionnaire entry point, questions that consistently appear in core positions in most paths can be used as mandatory questions, and questions that only appear in specific long-tail paths but have clear structural characteristics can be used as conditional trigger questions, thus forming a new adaptive questionnaire structure. Finally, the question text, option content, and jump rules corresponding to this structure are output as the optimized adaptive questionnaire.

[0051] Compared to questionnaire designs that do not consider the correction results of long-tail paths, this approach can improve the coverage stability of low-probability paths while reducing interference from long-tail paths, thereby enhancing the robustness of the park's smart management system under changing service demands.

[0052] Example 2 Please see Figure 2As shown, based on the same inventive concept, this embodiment discloses an adaptive questionnaire optimization system based on answer path analysis. For details not covered in this embodiment, please refer to the relevant sections of Embodiment 1. The system includes: Path clustering module: Used to obtain initial answer path data, extract path nodes based on the initial answer path data, and perform cluster analysis on the path nodes to obtain the initial path pattern; Topology recognition module: used to construct a path topology graph based on the initial path pattern, identify long-tail patterns of paths based on the path topology graph, and determine sparse paths; Pattern correction module: used to correct the path topology graph based on the path long tail pattern and sparse path, and perform path pattern mapping on the corrected path topology graph to obtain the corrected path pattern. Questionnaire Reconstruction Module: Used to reconstruct the structure and content of the adaptive questionnaire based on the modified path pattern, and output the optimized adaptive questionnaire.

[0053] The detailed description above, in conjunction with the accompanying drawings, describes examples but does not represent all examples that can be implemented or fall within the scope of the claims. The terms “example” and “exemplary” are used in this specification to mean “serving as an example, instance or illustration” and do not mean “superior to or better than other examples”.

[0054] Throughout this specification, the phrase "an embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment of the invention. Therefore, the use of these phrases may refer to more than one embodiment. Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0055] It should also be noted that these embodiments may be described as processes depicted as flowcharts, structural diagrams, or block diagrams. Although a flowchart may describe the operations as sequential processes, many of these operations can be performed in parallel or concurrently, and the order of these operations may be rearranged.

Claims

1. An adaptive questionnaire optimization method based on answer path analysis, characterized in that, include: Obtain initial answer path data, extract path nodes based on the initial answer path data, and perform cluster analysis on the path nodes to obtain the initial path pattern; Construct a path topology graph based on the initial path pattern, identify long-tail patterns of paths and determine sparse paths based on the path topology graph; The path topology graph is corrected based on the long-tail pattern and sparse path, and the path pattern mapping is performed on the corrected path topology graph to obtain the corrected path pattern. Reconstruct the structure and content of the adaptive questionnaire based on the modified path pattern, and output the optimized adaptive questionnaire.

2. The adaptive questionnaire optimization method based on answer path analysis according to claim 1, characterized in that, Path nodes are extracted from the initial answer path data, and cluster analysis is performed on the path nodes to obtain the initial path pattern, including: Extract each answer node from the initial answer path data, record the position order of each answer node in the answer path, and generate a set of node order structures; Based on the node sequence structure set, determine the preorder or subsequent adjacency relationship between each answer node and other answer nodes to form a node adjacency relationship structure; Based on the node adjacency relationship structure, determine the set of nodes with the same adjacent node combinations, and form a node combination structure; The answer nodes are clustered according to the node combination structure, and the clustered node combinations are determined as the initial path pattern.

3. The adaptive questionnaire optimization method based on answer path analysis according to claim 2, characterized in that, Based on the node adjacency structure, determine the set of nodes with the same combination of adjacent nodes, and form a node combination structure, including: Based on the node adjacency structure, extract the specific preceding and subsequent adjacent nodes of each answer node, and generate the node adjacency combination structure. The number of times adjacent node combinations appear in all answer paths is counted to form a node combination frequency table; Based on the node combination frequency table, determine the frequency distribution attribute of adjacent combinations of each node; The node combination structure is determined based on the adjacent combinations of nodes with frequency distribution attributes.

4. The adaptive questionnaire optimization method based on answer path analysis according to claim 3, characterized in that, Identifying long-tail patterns and determining sparse paths based on path topology graphs includes: Extract node connection segments consisting of consecutive answer nodes from the path topology graph and generate a set of node connection segments; The frequency of each node connection segment in the set of node connection segments in all answer paths is counted to form a node segment frequency table; The node fragment sparsity threshold is determined based on the node fragment frequency table; Based on the node fragment sparsity threshold, node connection fragments are filtered from the node fragment frequency table. The filtered node connection fragments are used as long-tail node connection fragments, and the long-tail node connection fragments are determined as the path long-tail pattern. Sparse paths are determined based on the node connection structure of the long-tail node connection segments.

5. The adaptive questionnaire optimization method based on answer path analysis according to claim 4, characterized in that, Based on the node fragment sparsity threshold, node connection fragments are filtered from the node fragment frequency table, and those fragments are identified as long-tail node connection fragments, including: Sort the occurrence frequency of node connection segments to form a node segment sorting sequence; Determine the node segment sorting threshold in the node segment sorting sequence; Identify node connection segments in the sorted sequence that are located after the node segment sorting threshold to form candidate long-tail segments; Analyze candidate long-tail segments to determine the structural differences in node-connected segments; Based on structural differences, long-tailed node connection segments are selected from candidate long-tailed segments and determined.

6. The adaptive questionnaire optimization method based on answer path analysis according to claim 1, characterized in that, The path topology graph is revised based on the long-tail pattern and sparse path, including: The long-tail node set is determined based on the path long-tail pattern, and the long-tail node set is mapped to the corresponding node in the path topology graph to form a long-tail node structure. Determine the node order structure based on the node connection order of the sparse path; By comparing the long-tail node structure with the node sequence structure, the connection relationship of the nodes to be adjusted is determined, and the node adjustment structure is formed. Adjust the node connection relationships in the path topology diagram based on the node adjustment structure; The path topology diagram after the structural adjustment is completed is determined as the corrected path topology diagram.

7. The adaptive questionnaire optimization method based on answer path analysis according to claim 6, characterized in that, Determine the connection relationships of the nodes to be adjusted to form the node adjustment structure, including: Extract all the long-tail node connections involved based on the long-tail node structure to form a node connection set; The sparse node structure is determined based on the connection order of nodes in the sparse path; The node connection set is structurally compared with the sparse node structure to determine the node connection relationships where there are differences in the connection order. The determined differences in the node connections form a node adjustment structure.

8. The adaptive questionnaire optimization method based on answer path analysis according to claim 7, characterized in that, The structural adjustment of node connections in the path topology graph is performed based on node adjustments, including: Based on the node adjustment structure, determine the nodes that need to be adjusted and their associated node connections to form a node positioning structure; Based on the node positioning structure, the differences in node connection directions are identified to form a connection direction structure; Based on the connection direction structure, conflicting node connections in the path topology graph with conflicting directions are deleted to obtain the target node connection relationship. Based on the initial node connection relationships of the initial path pattern, the connection relationships of the target nodes are supplemented to obtain the path topology diagram after structural adjustment.

9. The adaptive questionnaire optimization method based on answer path analysis according to claim 1, characterized in that, Perform path pattern mapping on the corrected path topology graph to obtain the corrected path patterns, including: Extract the corrected node connection relationship of each answer node from the corrected path topology graph to generate a node connection mapping structure; Determine the transition relationships between nodes based on the node connection mapping structure, and generate the node transition structure; Identify node segments with repeated connections based on the node transition structure, and form a node segment merging structure; Node mapping relationships are constructed based on the node fragment merging structure, and corrected path patterns are generated based on the node mapping relationships.

Images