Knowledge graph-based scientific research achievement clustering recommendation method and system

Abstract

The invention relates to the technical field of big data services, in particular to a scientific research achievement clustering recommendation method and system based on a knowledge graph, and the method comprises the steps: obtaining multi-dimensional information through a distributed data collection module, and constructing a scientific research achievement dynamic association graph; and combining the regional data optimization instruction set, the information alignment state table and the adjustment mark graph to realize information synchronization and deviation analysis, and outputting a monitoring region state evaluation value list. According to the method, the fluctuation expression precision of an information transmission path can be remarkably improved, the information synchronism control strength of a dense structure region is enhanced, the region needing to be adjusted is accurately positioned, local response compensation is realized, node state distribution is finally reconstructed, and the adaptability of a recommended application scene in a complex scientific research environment is expanded.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data services, and in particular to a scientific research achievement clustering recommendation method and system based on a knowledge graph. BACKGROUND

[0002] With the rapid growth of scientific research achievements and the increasing complexity of information retrieval needs, scientific research achievement clustering recommendation methods have gradually become an important means to improve the quality of scientific research information services. However, the existing technical solutions still have certain limitations in terms of accurate recommendation, data processing efficiency, and user query matching. According to the search, a scientific research achievement clustering recommendation method and system based on sub-graph representation learning (CN118626635B) is disclosed. This technology extracts keywords from the text of scientific papers and constructs a comprehensive knowledge graph, and combines sub-graph representation information to achieve clustering and recommendation of scientific research achievements. However, this solution mainly relies on keyword extraction from the text of scientific papers and fails to fully exploit multi-dimensional associated information (such as author collaboration networks and citation relationships) in scientific research achievements, which may affect the comprehensiveness and accuracy of the recommendation results. In addition, there is a lack of dynamic adjustment mechanism in the combination analysis process of user queries and sub-graph representation information, and the ability to adapt to diversified or fuzzy query requirements is insufficient.

[0003] In addition, a method for recommending English journals based on LSTM and knowledge graph (CN109299257B) is disclosed. This technology generates a candidate journal list through feature extraction, clustering analysis, and knowledge graph construction, thereby achieving fast recommendation for user input paper titles and abstracts. However, this solution mainly focuses on the English journal recommendation scenario, which has a limited application scenario. At the same time, its recommendation process relies on feature extraction of paper content and static knowledge graph, and lacks the ability to adapt to dynamically updated knowledge graph structures and real-time data changes, which may result in insufficient timeliness and flexibility of the recommendation results.

[0004] The above problems indicate that existing scientific research achievement clustering recommendation methods still have room for improvement in terms of multi-dimensional associated information mining, dynamic user query adaptation, and the universality of the recommendation scenario. Therefore, the present application aims to provide a scientific research achievement clustering recommendation method and system based on a knowledge graph, which introduces deep integration of multi-dimensional associated information, a dynamic query matching mechanism, and a universal recommendation framework, to further improve the accuracy, timeliness, and scope of application of the recommendation results, thereby better meeting the needs of users for efficient access to scientific research achievements. SUMMARY

[0005] To solve the technical problems existing in the prior art, the present application embodiment provides a scientific research achievement clustering recommendation method and system based on a knowledge graph. The technical solution is as follows:

[0006] A scientific research achievement clustering recommendation method based on a knowledge graph, comprising the following steps:

[0007] S1: Obtain multi-dimensional information of scientific research achievements through a distributed data acquisition module, record author cooperation network, citation relationship and keyword distribution characteristics, identify the initial mapping structure of key entities according to node correlation degree, and generate a scientific research achievement dynamic correlation graph;

[0008] S2: Based on the scientific research achievement dynamic correlation graph, extract the correlation strength between nodes and the regional abnormal distribution characteristics, combine the section correlation degree difference value and the preset threshold range to perform deviation analysis, filter the deviation region and adjust the information processing direction, and generate a regional data optimization instruction set;

[0009] S3: Call the regional data optimization instruction set to adjust the information acquisition strategy, synchronously record the time axis response value and peak value offset of the information acquisition module, compare the time axis and amplitude change before and after adjustment, extract the synchronous offset range of the corresponding region, and obtain a regional information alignment state table;

[0010] S4: Call the synchronous stable section information sequence in the regional information alignment state table, identify the information intensity change of the regional differentiated time window, perform offset analysis on the change threshold set by the reconstruction unit, mark the regions that need to be adjusted, and form a regional information adjustment marker map.

[0011] As a further scheme of the present application, the scientific research achievement dynamic correlation graph includes an author cooperation network distribution curve, a citation relationship characteristic parameter and a node classification identifier, the regional data optimization instruction set includes an information acquisition strategy parameter set, a correlation strength control value and a section information compensation factor, the regional information alignment state table includes a synchronous time offset, a peak value response difference value and an alignment state identifier code, and the regional information adjustment marker map includes a section position point that needs to be adjusted, an intensity offset judgment result and a structure change response identifier.

[0012] As a further scheme of the present application, the distributed data acquisition module adopts a multi-channel heterogeneous data fusion architecture, comprising:

[0013] An author cooperation network acquisition unit is configured to obtain an author cooperation strength value through an academic social network API, calculate the product of cooperation frequency and time decay factor as a node correlation degree benchmark value;

[0014] A citation relationship analysis unit is configured to extract the citation tree structure characteristics of the literature, calculate the node influence coefficient through the PageRank algorithm, and generate a dynamic weight value in combination with a citation time window;

[0015] The keyword distribution modeling unit is configured to analyze the abstract of the research achievement by using the LDA topic model, calculate the TF-IDF drift of the keyword in the time dimension, and generate a topic evolution feature vector.

[0016] As a further scheme of the present application, the step of obtaining the research achievement dynamic correlation graph is specifically:

[0017] S111: Obtain the multi-dimensional information of the research achievement through the distributed data acquisition module, record the response value of the author cooperation network in different research fields, and convert the cooperation strength index according to the calibration coefficient to form the operation state archive under the current research field and obtain the author cooperation network parameter set;

[0018] S112: Based on the author cooperation network parameter set, record the cooperation strength change of the author cooperation network under the differentiated research condition, analyze the mapping relationship between the research condition and the index fluctuation rate, reconstruct the information distribution characteristics in the research environment, evaluate the stability of the key node, and generate the research achievement dynamic correlation graph.

[0019] As a further scheme of the present application, the step of obtaining the region data optimization instruction set is specifically:

[0020] S211: Based on the research achievement dynamic correlation graph, identify the correlation strength and abnormal distribution characteristics between nodes, extract the information change rate per unit time in the continuous node, and classify the interval according to the correlation gradient value, identify the response interval of the information change on the node, and obtain the correlation response change interval;

[0021] S212: Call the correlation response change interval, according to the node segment information difference value and the upper and lower limit information ratio in the path, scan the difference fluctuation amplitude and error trend of the node segment, compare the matching degree of real-time information and node distribution trend, calculate the information offset degree value, judge the abnormal distribution area in the node, extract the path position group that needs to be adjusted, and generate the region data optimization instruction set.

[0022] As a further scheme of the present application, the step of obtaining the region information alignment state table is specifically:

[0023] S311: Call the region data optimization instruction set to adjust the information acquisition strategy, compare the current information value according to the adjustment node number and the target information configuration, execute the high-low priority information difference adjustment, and record the response starting time, peak time and peak amplitude, obtain the adjustment path response time sequence group.

[0024] S312: According to the adjustment path response time series group, extract the start time, peak time and amplitude of the nodes before and after the adjustment, identify the offset difference sequence, calculate the regional synchronization offset strength value, map the strength value with the node distribution, filter the node group within the synchronization range and arrange the information timing sequence to obtain the regional information alignment status table.

[0025] As a further solution of the present invention, the step of obtaining the region information adjustment mark map is specifically as follows:

[0026] S411: calling the synchronization stable segment information sequence in the region information alignment state table, extracting the information intensity per unit time of the synchronization stable segment, and comparing it with the intensity sequence of the corresponding position of the adjacent region information segment, identifying the time deviation of the information intensity, and obtaining the region information offset value;

[0027] S412: Based on the region information offset value and the change threshold set by the reconstruction unit, determine the difference between the value of each region information offset and the threshold, filter the offset regions exceeding the threshold, and obtain the block change determination coefficient;

[0028] S413: Based on the block change judgment coefficient, detect the information offset trend in the area and the continuity between the offset positions, mark the areas where the offset trend is stably rising or falling and spatially continuous, calculate the regional offset mark value, combine the spatial range of the offset area, identify the connected identification partition number and fill it into the mark map to form the regional information adjustment mark map.

[0029] As a further embodiment of the present invention, the method further comprises:

[0030] S5: calling the positioning correction segment information in the regional information adjustment mark map, identifying the adjusted high and low priority information combination value, reconstructing the regional node response state according to the associated feature relationship corresponding to the combination information, and outputting a monitoring area state assessment value list;

[0031] The monitoring area status assessment value list includes a high-priority and low-priority information fluctuation comparison value, a regional response difference factor, and a response status level after reconstruction.

[0032] As a further solution of the present invention, the steps for obtaining the monitoring area status assessment value list are specifically as follows:

[0033] S511: Calling the positioning correction segment information in the regional information adjustment mark map, extracting high-priority and low-priority information pairs of the positioning segments, matching the node coordinates with the path numbers, and grouping the combination information values ​​in spatial sequence to generate a dual-path positioning combination value set;

[0034] S512: Based on the dual-path positioning combination value set, the fluctuation difference of the combination value is distinguished, the information pair of the distinguishable degree is extracted, the regional response group of the fluctuation characteristic is screened according to the set dual-priority response threshold, and the dual-priority characteristic fluctuation rate sequence is generated.

[0035] S513: According to the dual-priority characteristic fluctuation rate sequence, the mapping relationship between the fluctuation parameter and the node state is analyzed, the dual-priority state inversion of the regional node is carried out, the node state of the positioning section is identified, and the monitoring area state evaluation value list is output.

[0036] A scientific research achievement clustering recommendation system based on a knowledge graph, the system comprising:

[0037] The distributed data acquisition module obtains multi-dimensional information of the scientific research achievements through a multi-source academic feature acquisition matrix, compares the node state information in the difference research field horizontally, identifies the information response difference according to the path grouping, integrates the amplitude change section of the whole path information, and constructs a scientific research achievement dynamic correlation graph.

[0038] The correlation positioning module extracts a regional dual-priority information ratio sequence based on the scientific research achievement dynamic correlation graph, identifies the response difference between paths and screens the deviation mutation section, identifies the correlation trend intersection point, determines the region range that needs to be adjusted, and generates a correlation section positioning set.

[0039] The information synchronization module adjusts the high-low priority information acquisition proportion in the indicated region based on the correlation section positioning set, records the data acquisition matrix time axis response value and peak value offset amplitude sequence in the corresponding section, compares the response difference between paths before and after adjustment, integrates the stable synchronization point group, and establishes an information alignment synchronization state table.

[0040] The correlation structure identification module extracts the amplitude change quantity in the synchronization section information based on the information alignment synchronization state table, compares the adjacent path response floating interval according to the time window, maps the offset out-of-limit position to a two-dimensional imaging surface, marks the local correlation strength mutation region, and outputs a correlation structure change layer.

[0041] The state parameter evaluation module extracts the corresponding high-low priority information combination value based on the marked region in the correlation structure change layer, combines the set reference fluctuation rate value table, carries out unit area fluctuation response calculation, and outputs a monitoring area state evaluation value list.

[0042] The technical scheme provided by the embodiment of the application has at least the following beneficial effects:

[0043] In the present application, the multidimensional information of scientific research achievements is acquired by the distributed data acquisition module, and a complete dynamic correlation graph of scientific research achievements is constructed, the information fluctuation characteristics under different research fields are clarified, and the fluctuation expression accuracy of the information transmission path is significantly improved. By using the linkage analysis mechanism of the correlation trend and the information upper and lower limit difference, the information deviation section is effectively identified and the acquisition strategy is adjusted, so that the information acquisition path has higher resolution and dynamic regulation ability in different priority sections. The response time and peak change are extracted synchronously in the adjustment process, the regional alignment state information is constructed, the control strength of the information synchronization in the dense structure region is enhanced, and the influence of the time axis drift on the subsequent identification accuracy is avoided. Based on the alignment state, the difference between the regional intensity change and the preset threshold is further judged, the information stable section difference is identified, the region needing adjustment is accurately positioned, and local response compensation under high sensitivity is realized. Finally, the node state distribution is reconstructed by combining the information fluctuation characteristics, the response level of high and low priority information in multiple research scenarios is established, the state evaluation value is output, the identification and state quantitative judgment accuracy of the multi-layer structure details are enhanced, and the adaptability of the recommended application scenarios in complex scientific research environments is significantly expanded. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The method flowchart of the present application is shown in Figure 1.

[0045] Figure 2 The acquisition flowchart of the dynamic correlation graph of scientific research achievements of the present application is shown in Figure 2.

[0046] Figure 3 The acquisition flowchart of the regional data optimization instruction set of the present application is shown in Figure 3.

[0047] Figure 4 The acquisition flowchart of the regional information alignment state table of the present application is shown in Figure 4.

[0048] Figure 5 The acquisition flowchart of the regional information adjustment marker map of the present application is shown in Figure 5.

[0049] Figure 6 The acquisition flowchart of the monitoring regional state evaluation value list of the present application is shown in Figure 6. DETAILED DESCRIPTION

[0050] The technical solutions in the present application will be described below with reference to the accompanying drawings.

[0051] In the embodiments of the present application, the words such as "example", "for example" are used to represent an example, illustration or explanation. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. In fact, the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0052] In the embodiments of the present application, "image" and "picture" can be used interchangeably, and it should be pointed out that the meanings expressed are consistent when the distinction is not emphasized. "Of", "corresponding" and "corresponding" can be used interchangeably, and it should be pointed out that the meanings expressed are consistent when the distinction is not emphasized.

[0053] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1, and the meanings expressed are consistent when the distinction is not emphasized.

[0054] In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail in conjunction with the drawings and specific embodiments.

[0055] Please refer to Figure 1 The present application provides a technical solution: a scientific research achievement clustering recommendation method based on a knowledge graph, comprising the following steps:

[0056] S1: Obtain multi-dimensional information of scientific research achievements through a distributed data acquisition module, record author cooperation network, citation relationship and keyword distribution characteristics, identify the initial mapping structure of key entities according to node correlation degree, and generate a scientific research achievement dynamic correlation graph;

[0057] The distributed data acquisition module adopts a multi-channel heterogeneous data fusion architecture, including:

[0058] The author cooperation network acquisition unit is configured to obtain the author cooperation strength value through the academic social network API, calculate the product of the cooperation frequency and the time decay factor as the node correlation degree benchmark value;

[0059] The citation relationship analysis unit is configured to extract the citation tree structure characteristics of the literature, calculate the node influence coefficient through the PageRank algorithm, and generate dynamic weight values combined with the citation time window;

[0060] The keyword distribution modeling unit is configured to analyze the achievement abstract by using the LDA topic model, calculate the TF-IDF drift of the keywords in the time dimension, and generate a topic evolution feature vector;

[0061] S2: Based on the scientific research achievement dynamic correlation graph, extract the correlation strength and regional abnormal distribution characteristics between nodes, combine the section correlation difference value and the preset threshold range for deviation analysis, filter the deviation area and adjust the information processing direction, and generate a regional data optimization instruction set;

[0062] S3: Call the regional data optimization instruction set adjustment information collection strategy, record the time axis response value and peak value offset of the information collection module, and extract the corresponding regional synchronization offset range by comparing the time axis and amplitude changes before and after adjustment. Get the regional information alignment state table;

[0063] S4: Call the synchronization stable segment information sequence in the regional information alignment state table, identify the information intensity change of the regional differentiated time window, and analyze the offset with the change threshold set by the reconstruction unit. Mark the regions that need to be adjusted to form a regional information adjustment marker map;

[0064] S5: Call the positioning correction segment information in the regional information adjustment marker map, identify the high and low priority information combination value after adjustment, and reconstruct the regional node response state according to the corresponding associated feature relationship of the combination information. Output the monitoring region state evaluation value list.

[0065] The scientific research achievement dynamic correlation graph includes author cooperation network distribution curve, reference relationship characteristic parameter, node classification identifier, regional data optimization instruction set includes information collection strategy parameter set, correlation strength control value, section information compensation factor, regional information alignment state table includes synchronization time offset, peak value response difference value, alignment state identifier code, regional information adjustment marker map includes adjustment section position point, intensity offset judgment result, structure change response identifier, monitoring region state evaluation value list includes high and low priority information fluctuation comparison value, regional response difference factor, reconstructed response state level, etc.

[0066] Please refer to Figure 2 , the acquisition steps of the scientific research achievement dynamic correlation graph are as follows:

[0067] S111: Obtain multi-dimensional information of scientific research achievements through a distributed data collection module, record the response value of the author cooperation network in different research fields, and convert the cooperation intensity index according to the calibration coefficient to form the running state archive in the current research field and obtain the author cooperation network parameter set;

[0068] Assuming that for the two research fields of "graphene-based supercapacitors" and "carbon nanotube sensors", the author "Dr. Li" collects the original signals of his academic activities from the fourth quarter of 2024 to the first quarter of 2025 using the academic social network API, and obtains a response value of 520 in the field of "graphene-based supercapacitors" and a response value of 310 in the field of "carbon nanotube sensors". Then, set the calibration coefficient, which is set by referring to the weighted average of the average journal impact factor and the funding intensity of the National Natural Science Foundation project in the past five years. Through regression analysis of historical data of 15 mainstream research fields, the calibration coefficient of the "graphene-based supercapacitors" field is set to 1.25 due to its high heat and high investment, and the calibration coefficient of the "carbon nanotube sensor" field is set to 0.95. The setting process has undergone 3 rounds of cross-validation experiments, each round using 10 fields of historical data as the training set and 5 as the validation set, and the final coefficient error rate is controlled within 4%. According to this, the cooperation intensity index is converted, and the cooperation intensity index of "Dr. Li" in the field of "graphene-based supercapacitors" is calculated as 520 × 1.25 = 650, and in the field of "carbon nanotube sensors" is 310 × 0.95 = 294.5. Such information, together with the cooperation intensity index values of other collaborators such as "Dr. Wang", "Professor Zhang" and others in different research fields, is integrated to form the running state file of the current research field. See Table 1 below for specific data. Finally, structured data is extracted from the file to obtain the author cooperation network parameter set.

[0069] Table 1: Example of author cooperation network running state file

[0070]

[0071]

[0072] As shown in Table 1, the table lists the academic activity response data of some researchers in different research fields and their standardized cooperation intensity index. This running state file provides basic data input for subsequent analysis.

[0073] S112: Based on the author cooperation network parameter set, record the changes in cooperation intensity of the author cooperation network under different research conditions, analyze the mapping relationship between research conditions and index volatility, reconstruct the information distribution characteristics in the scientific research environment, evaluate the stability of key nodes, and generate a dynamic correlation graph of scientific research results.

[0074] Based on the obtained author collaboration network parameter set, for example, the cooperation intensity index of "Dr. Li" in the field of "graphene-based supercapacitors" is 650, record the change of cooperation intensity under different research conditions, here the different research conditions specifically refer to the experimental environment is switched from "aqueous electrolyte" to "ionic liquid electrolyte", after switching, the cooperation intensity index collected again is 578.5, and then analyze the mapping relationship between research conditions and index fluctuation rate, first calculate the index change rate, the calculation process is the difference between the two cooperation intensity indexes divided by the original index, that is, (650-578.5) / 650=0.11, here 0.11 is the index fluctuation rate, which is mapped with the research condition of "electrolyte environment change" as a key-value pair and stored as "{ 'condition': 'electrolyte change', 'fluctuation rate': 0.11}", then, according to this mapping relationship, the information distribution characteristics in the scientific research environment are reconstructed, and the stability of the key nodes is evaluated, a key node, such as the scientific research achievement (a core paper) numbered P1 here, its stability is inversely proportional to the index fluctuation rate of the core author "Dr. Li", the specific calculation method of stability is 1 minus the maximum index fluctuation rate of the core author, that is, 1-0.11=0.89, in order to quantitatively classify the stability, set the stability threshold interval: the stability value in the interval [0.95, 1.0] is defined as "high stability", the value in the interval [0.85, 0.95) is defined as "medium stability", and the value less than 0.85 is defined as "low stability", these intervals are based on the statistical analysis of the survival status of the past 500 key nodes after the change of research conditions, therefore, the stability of node P1 is evaluated as "medium stability", and the stability of node P1 and other nodes (such as P2, P3) (for example, 0.96 and 0.82 respectively) are integrated to generate a dynamic correlation graph of scientific research achievements;

[0075] Based on the dynamic correlation graph, weighted feature similarity analysis is performed:

[0076] A scientific research achievement feature vector space is constructed, and the dimension weight is defined: cooperation network density (weight 0.4), calculated as the ratio of actual collaborators to potential collaborators; cross-domain citation rate (weight 0.3), taking the number of times cited by other field literature as a percentage; achievement timeliness coefficient (weight 0.2), calculated as e -0.05t (t is the number of years of publication); theme similarity (weight 0.1), based on the KL divergence conversion value of the theme distribution generated by the LDA model;

[0077] Weighted cosine similarity is used for calculation:

[0078]

[0079] Taking two achievements in the field of materials science as an example, suppose:

[0080] Achievement A feature values: (0.85, 0.42, 0.78, 0.91)

[0081] Achievement B feature values: (0.67, 0.35, 0.63, 0.82)

[0082] Calculate the weighted inner product:

[0083]

[0084] Calculate the normalization factor:

[0085]

[0086] Similarity:

[0087] In order to eliminate the numerical expansion caused by the weight distribution deviation, compensate for the basic similarity difference in different research fields, and adapt to the requirement of the recommendation system for similarity value sensitivity, the similarity is normalized:

[0088]

[0089] Where, μ = 0.6, σ = 0.2 (based on historical data statistics), and for the modified similarity 1.995, it is beyond the reasonable range, so the Sigmoid function is used for secondary calibration:

[0090]

[0091] Take k = 2,

[0092] The final similarity = 0.93 x 0.9 (theme similarity weight compensation coefficient) = 0.837;

[0093] Set the similarity threshold value 0.65 (verified by ROC curve), and select the top 20% as the recommended candidate set. The results show that the two scientific research achievements have high correlation in the weighted feature space (more than the threshold value 0.65), and the system will establish a recommended correlation relationship. This similarity value will be used as the core basis for generating the recommended candidate set.

[0094] Please refer to Figure 3 The acquisition step of the region data optimization instruction set is specifically:

[0095] S211: Based on the dynamic correlation graph of scientific research achievements, identify the correlation strength and abnormal distribution characteristics between nodes, extract the unit time information change rate in continuous nodes, and classify the interval based on the correlation gradient value, identify the response interval of information change on the node, and obtain the correlation response change interval;

[0096] The scientific research achievement dynamic correlation graph contains a series of nodes and their stability values, for example, a node sequence is [P1: 0.89, P2: 0.96, P3: 0.82, P4: 0.80, P5: 0.93], first, the correlation strength between nodes and the abnormal distribution characteristics are identified, it is found through observation that the stability of node P2 to P3 has a sharp decline from 0.96 to 0.82, which constitutes an abnormal distribution point, then, the information change rate per unit time in the continuous nodes is extracted, it is assumed that each node interval represents a one-month observation period, then the information change rate between P2 and P3 is (0.82-0.96) / 1=-0.14, and the rate between P3 and P4 is (0.80-0.82) / 1=-0.02, then, the interval classification is carried out combined with the correlation gradient value, the correlation gradient value is defined as the absolute value of the information change rate, so the correlation gradient value between P2 and P3 is 0.14, and the correlation gradient value between P3 and P4 is 0.02, the classification threshold is set: the gradient value greater than 0.1 is classified into the “high fluctuation interval”, the value in the range of [0.05, 0.1] is classified into the “moderate fluctuation interval”, and the value less than 0.05 is classified into the “stable interval”, the threshold is determined by analyzing the fluctuation of the historical project data in the database, and selecting the fluctuation values of the 80% and 30% quantile points as the division basis, according to this, the connection segment between P2 and P3 is classified into the “high fluctuation interval”, the response interval of the information change on the node is further identified, and finally the continuous node segment [P2, P3] with high fluctuation characteristics is identified, and the correlation response change interval is obtained.

[0097] S212: Call the correlation response change interval, according to the node segment information difference value and the upper and lower limit information ratio in the path, scan the difference fluctuation amplitude and error trend of the node segment, compare the matching degree of real-time information and node distribution trend, calculate the information offset degree value, judge the abnormal distribution area in the node, extract the path position group that needs to be adjusted, and generate the regional data optimization instruction set;

[0098] Call the node segment [P2, P3] of the associated response change interval, and scan the node segment difference fluctuation amplitude and error trend according to the node segment information difference and the upper and lower limit information ratio within the path. First, calculate the information difference, that is, the absolute value of the stability difference between nodes P2 and P3 |0.96-0.82|=0.14, and then calculate the upper and lower limit information ratio within the path. The path refers to the entire sequence from P1 to P5. Its maximum stability is 0.96 of P2 and the minimum is 0.96 of P4. 0.80, the information ratio is 0.80 / 0.96=0.833. Next, compare the matching degree between real-time information and node distribution trend. Here, real-time information refers to the difference of 0.14 in the [P2, P3] segment, while the node distribution trend is represented by the average inter-node difference of the entire path [P1-P5]. The average difference is (|0.96-0.89|+|0.82-0.96|+|0.80-0.82|+|0.93-0.80|).

[0099] / 4=0.09. Subsequently, the information deviation degree value is calculated, which is determined by the ratio of the real-time information difference to the trend average difference, that is, 0.14 / 0.09=1.556. The abnormal distribution area in the node is judged, and the deviation degree threshold is set to 1.5. When the calculated deviation degree value exceeds this threshold, the area is judged to be an abnormal distribution area. This threshold is established by simulation tests on 1000 groups of mature scientific research paths and selecting the minimum threshold that can cover 90% of the known mutation points. Because 1.556 is greater than 1.5, [P2, P3] is confirmed to be an abnormal distribution area. Finally, the path position group that needs to be adjusted, that is, the identifiers of nodes P2 and P3, is extracted to generate a regional data optimization instruction set.

[0100] See also Figure 4 , the steps for obtaining the regional information alignment status table are as follows:

[0101] S311: Calling the regional data optimization instruction set to adjust the information collection strategy, comparing the current information value according to the adjustment node number and the target information configuration, performing high- and low-priority information difference adjustment, and recording the response start time, peak time, and peak amplitude to obtain the adjustment path response time series group;

[0102] The regional data optimization instruction set specifies that the association information between nodes P2 and P3 needs to be adjusted in strategy. According to the comparison between the target information configuration and the current information value of the adjustment node numbers P2 and P3, the stability difference of adjacent nodes should not exceed 0.05, while the difference between P2 (0.96) and P3 (0.82) is 0.14, far exceeding the target. According to this, high and low priority information difference adjustment is performed. The specific operation is to increase the data collection weight of the research related to node P3 (for example, supplement the literature citing its previous achievements, or strengthen the cooperation network of its authors), and increase the weight coefficient from 1.0 to 1.5, while reducing the collection weight of the peripheral information associated with node P2 with low degree, from 1.0 to 0.7. After executing the adjustment strategy, the response starting time, peak time and peak amplitude are recorded synchronously, for example, starting from the time when the strategy adjustment instruction is issued (T0 = 15:30:00), the stability value of node P3 starts to change, and at T1 = 15:35:10, its stability value reaches the peak value 0.88, and this peak value is the peak amplitude. The complete record of this process (node number P3, starting time T0, peak time T1, peak amplitude 0.88) is captured, and similar records of other affected nodes (such as the stability of P2 decreases to 0.95) together constitute the adjustment path response time sequence group.

[0103] S312: According to the adjustment path response time sequence group, extract the starting time, peak time and amplitude of the nodes before and after adjustment, identify the offset difference value sequence, calculate the regional synchronization offset strength value, map the strength value and node distribution, screen the node group within the synchronization range and arrange the information time sequence, and obtain the regional information alignment state table;

[0104] The adjustment path response time sequence group data contains the response record of node P3 {start time: 15:30:00, peak time: 15:35:10, amplitude: 0.88} and the response record of node P2 {start time: 15:30:05, peak time: 15:34:50, amplitude: 0.95}. First, the start time, peak time and amplitude of the nodes before and after adjustment are extracted, the offset difference sequence is identified, the time offset is calculated, the start time offset of P3 and P2 is 15:30:05-15:30:00=5 seconds, the peak time offset is 15:35:10-15:34:50=20 seconds, then the regional synchronization offset intensity value is calculated, which integrates the time and amplitude synchronization, and the calculation method is the product of the inverse of the time offset and the amplitude correlation. Here, it is simplified to the quantification of the time offset. A reference synchronization time difference of 10 seconds is set. Through the analysis of 50 successful adjustment cases in history, it is found that the region with a response time difference less than 10 seconds has high synchronization. The synchronization offset intensity is calculated as 1-(|start time offset-reference|+|peak time offset-reference|) / (2*reference)=1-(|5-10|+|20-10|) / (2*10)=1-15 / 20=0.25. Then the intensity value and node distribution are numbered and mapped. The intensity value 0.25 is mapped to the node pair [P2, P3]. The synchronized node group in the range is selected and the information time sequence is arranged. The synchronization intensity threshold is set to 0.7, that is, the node group with an intensity value greater than 0.7 is considered to be synchronized. Since 0.25 is less than 0.7, the node group [P2, P3] does not reach a synchronized stable state after this round of adjustment, so the synchronized stable segment information sequence is empty, triggering the next round of adjustment or marking as asynchronous. Assuming that in the adjustment of another region [P5, P6], the calculated synchronization offset intensity value is 0.85, then the node group is selected, and the information sequence after adjustment is arranged to obtain the regional information alignment state table.

[0105] Please refer to Figure 5 The acquisition steps of the regional information adjustment marker diagram are as follows:

[0106] S411: Call the synchronized stable segment information sequence in the regional information alignment state table, extract the information intensity of each unit time of the synchronized stable segment, and compare it with the intensity sequence of the corresponding position of the adjacent regional information segment to identify the time deviation of the information intensity and obtain the regional information offset value.

[0107] The calling area information alignment state table synchronously stable segment information sequence is called, for example, the information sequence of the node segment [P5(0.93), P6(0.91)] that has reached synchronous stability is selected, and the information intensity of each unit time of the synchronously stable segment is extracted. It is assumed that in a unit time (for example, 1 day), the information collection intensity (which can be quantified as the number of collected data points) of the P5 and P6 areas is [105, 108, 110, 107, 112] and [98, 102, 103, 101, 105], respectively. Then, the intensity sequence [80, 75, 120, 115, 85] of the corresponding position of the adjacent area information segment, which is the non-synchronous area [P2(0.95), P3(0.88)] in this case, is compared, the time deviation of the information intensity is identified, and the specific operation is to calculate the absolute value of the difference between the two sequences at each time point to form a deviation sequence. For example, at the first time point, the deviation is |105-80|=25, and at the second time point, the deviation is |108-75|=33. In this way, the deviation sequence [25, 33, 7, 8, 27] is obtained. Finally, the average value of the deviation sequence is calculated as the area information offset value, that is, (25+33+7+8+27) / 5=20. The area information offset value is obtained as 20.

[0108] S412: According to the area information offset value, the numerical difference between each area information offset and the threshold limit set by the reconstruction unit is judged, the offset area exceeding the threshold limit is screened out, and the block change judgment coefficient is obtained.

[0109] According to the obtained area information offset value 20, the numerical difference is judged in combination with the change threshold limit set by the reconstruction unit. The change threshold limit of the reconstruction unit here is set based on the statistical distribution of the natural fluctuation of the information collection intensity between different areas when the system is normally running. Through analysis of the data under the condition of no instruction intervention for 30 consecutive days, it is determined that the upper limit of the 95% confidence interval is 15 data points. Therefore, the change threshold limit is set to 15. Then, the numerical difference between each area information offset and the threshold limit is judged, and the calculation process is 20-15=5. Since the difference value 5 obtained by calculation is positive, it indicates that the area information offset exceeds the threshold limit. Therefore, the comparison area between [P2, P3] and [P5, P6] is screened out as the offset area, and the block change judgment coefficient is obtained. This coefficient is a Boolean value or a quantitative value. Here, since it exceeds the threshold limit, it can be recorded as 1, and if it does not exceed the threshold limit, it can be recorded as 0. Finally, the block change judgment coefficient is obtained as 1.

[0110] S413: Based on the block change judgment coefficient, the information offset trend in the area and the continuity between the offset positions are detected, the area with a stably rising or falling offset trend and spatial continuity is marked, the area offset marker value is calculated, the connected marker partition number is identified in combination with the spatial range of the offset area, and the marker map is filled in to form an area information adjustment marker map.

[0111] Based on the obtained block change judgment coefficient 1, this coefficient indicates that there is a significant deviation between regions, and further detection of the information deviation trend within the region and the continuity of the deviation position is required. The region with deviation, i.e., the [P2, P3] region, is retrieved, and the internal information acquisition intensity sequence of the region is [80, 75, 120, 115, 85]. By calculating the change trend between adjacent points, the trend sequence [-, -, +, -, -] is obtained. The change from 75 to 120 is “+” (upward), and the others are mainly “-” (downward), which does not constitute a stable one-way trend, so the region is not marked. Assuming that the intensity sequence of another deviation region [P7, P8] is [90, 95, 102, 108, 115], the trend sequence is [+, +, +, +], which shows a stable upward trend and spatial continuity (because it is adjacent nodes), so the region is marked, and the region deviation marking value is calculated. The calculation method of this value is the continuous length of the trend multiplied by the average change rate, i.e., 4 x ((115-90) / 4) = 25. Then, combined with the spatial range of the deviation region, the connected identification partition number is identified, and the region [P7, P8] is taken as a connected partition, which is given a unique number, such as “ADJ-001”, and this number and related information are filled into the marking map to form a region information adjustment marking map.

[0112] Table 2: Adjustment marking example of region information

[0113]

[0114] As shown in Table 2, the table shows the adjustment marking formed after judging the information deviation trend and continuity. The partition ADJ-001 is marked due to its stable deviation trend, and subsequent positioning correction is required.

[0115] Please refer to Figure 6 The acquisition steps of the region state evaluation value list are as follows:

[0116] S511: Call the positioning correction segment information in the region information adjustment marking map, extract the high priority and low priority information pairs of the positioning segment, match the node coordinates and path numbers, and collect and combine the information values in spatial sequence to generate a double-path positioning combination value set;

[0117] The positioning correction segment information in the marked graph is adjusted, that is, the region marked as "ADJ-001", which involves nodes P7 and P8, in which the high-priority and low-priority information pairs are extracted, the high-priority information is defined as the keywords directly related to the core theme of the region (such as "electrode material aperture"), and the data acquisition intensity sequence is [90, 95, 102, 108, 115], and the low-priority information is defined as the auxiliary information with lower correlation (such as "experimental environment humidity"), and the data acquisition intensity sequence is [50, 52, 51, 53, 50], and the nodes coordinates P7, P8 and the path number ADJ-001 are matched, and the combined information values are collected in the spatial sequence, and at the first time point, the combined value is (90, 50), and at the second time point, the combined value is (95, 52), and so on, to form a sequence of 5 information pairs [(90, 50), (95, 52), (102, 51), (108, 53), (115, 50)], which is the generated double-path positioning combined value set.

[0118] S512: Based on the double-path positioning combined value set, the fluctuation difference of the combined value is distinguished, the distinguishable information pair is extracted, the region response group of the fluctuation feature is screened according to the set double-priority response threshold, and a double-priority feature fluctuation rate sequence is generated.

[0119] Based on the generated dual-path positioning combination value set [(90, 50), (95, 52), (102, 51), (108, 53), (115, 50)], the fluctuation difference of the combination value is distinguished, first, the difference value of each information pair is calculated, and the difference value sequence [40, 43, 51, 55, 65] is obtained, then the distinguishable information pair is extracted, the distinguishable degree is the variation coefficient of the difference value sequence, the variation coefficient is the standard deviation divided by the average value of the difference value sequence, the average value of the difference value sequence is (40+43+51+55+65) / 5=50.8, the standard deviation is about 9.26, and the variation coefficient is 9.26 / 50.8=0.182, the variation coefficient threshold is set to 0.15, which is established by calculating 30 groups of known effective and invalid information pair sample data, and it is found that the variation coefficient of the effective information pair is greater than 0.15, because 0.182>0.15, it indicates that the information pair has sufficient distinguishable degree, next, according to the set dual-priority response threshold, the area response group of fluctuation characteristics is screened, the high-priority response threshold is set to continuous growth, and the low-priority response threshold is set to fluctuation range not more than 5% (i.e. the difference between the maximum value and the minimum value divided by the average value), the high-priority sequence [90, 95, 102, 108, 115] meets the continuous growth, and the average value of the low-priority sequence [50, 52, 51, 53, 50] is 51.2, the fluctuation is (53-50) / 51.2=5.86, which is slightly higher than 5%, but considering the overall trend, it is still regarded as a candidate area response group, and the dual-priority characteristic fluctuation rate sequence is generated;

[0120] When the dynamic time warping (DTW) algorithm is used for sequence matching, the high-priority sequence H={h The low-priority sequence L={l τ} τ＝1 Tau, construct the cumulative distance matrix:

[0121] D(t,τ)=|h t -l τ |+min{D(t-1,τ),D(t,τ-1),D(t-1,τ-1)};

[0122] The optimal path selection needs to meet:

[0123] Boundary conditions: starting from (1, 1) and ending at (T, T); Monotonicity constraint: the path strictly increases with time; Continuity constraint: the distance between adjacent points is less than or equal to 1;

[0124] Suppose the obtained material characterization experimental data is:

[0125] High-priority sequence: [90, 95, 102, 108, 115] (electrode material performance index);

[0126] Low priority sequence: [50, 52, 51, 53, 50] (ambient humidity data);

[0127] Compute the regularized path:

[0128]

[0129] Unit distance:

[0130] This value is significantly higher than the preset unit distance threshold 12.5, indicating that there is a morphological difference between the high priority sequence (electrode material performance) and the low priority sequence (ambient humidity), and this discrimination result will trigger the feature fluctuation rate sequence reconstruction process to provide input parameters for subsequent state inversion.

[0131] S513: According to the double-priority feature fluctuation rate sequence, analyze the mapping relationship between fluctuation parameters and node states, perform double-priority state inversion of regional nodes, identify and locate the node state, and output a list of monitoring area state evaluation values;

[0132] According to the double-priority feature fluctuation rate sequence, analyze the mapping relationship between fluctuation parameters and node states, and establish a mapping rule: when the high-priority information intensity is steadily rising (fluctuation rate is positive and stable) and the low-priority information intensity is stable (fluctuation rate is close to zero), the node state is mapped to "healthy evolution"; when both are rising synchronously, the state is "potential overheating"; when the high-priority is declining and the low-priority is stable, the state is "development stagnation". These mapping relationships are constructed based on rule extraction from domain expert knowledge base and machine learning classification results of historical cases. Next, perform double-priority state inversion of regional nodes. For the region "ADJ-001", the high-priority information shows continuous growth, while the low-priority information fluctuates in a small range. This state best fits the "healthy evolution" in the mapping rule, so the identified and located node state is "healthy evolution". Finally, this evaluation result and the state evaluation results of all other monitoring areas (such as P2, P3 areas being evaluated as "fluctuation adjustment") are summarized and output as a list of monitoring area state evaluation values.

[0133] A scientific research achievement clustering recommendation system based on a knowledge graph, the system comprising:

[0134] The distributed data acquisition module obtains multi-dimensional information of scientific research achievements through a multi-source academic feature acquisition matrix, compares the node state information in different research fields horizontally, identifies information response differences according to path grouping, integrates the amplitude change section of the whole path information, and constructs a dynamic correlation graph of scientific research achievements;

[0135] The correlation positioning module extracts a regional double-priority information ratio sequence based on a scientific research achievement dynamic correlation graph, identifies path response difference and filters deviation mutation sections, identifies correlation trend intersection points, determines a region range to be adjusted, and generates a correlation section positioning set;

[0136] The information synchronization module adjusts high and low priority information collection ratios in the indicated region based on the correlation section positioning set, records data collection matrix time axis response values and peak offset amplitude sequences in the corresponding sections, compares path response differences before and after adjustment, integrates stable synchronization point groups, and establishes an information alignment and synchronization state table;

[0137] The correlation structure identification module extracts amplitude change amounts in the synchronization section information, compares adjacent path response floating intervals according to time windows, maps offset out-of-limit positions to a two-dimensional imaging surface, marks local correlation strength mutation regions, and outputs a correlation structure change layer;

[0138] The state parameter evaluation module extracts corresponding high and low priority information combination values based on the marked regions in the correlation structure change layer, combines a set reference fluctuation rate value table, calculates unit region fluctuation responses, and outputs a monitoring region state evaluation value list.

[0139] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A scientific research results clustering recommendation method based on knowledge graph, characterized in that: The following steps are involved: S1: Obtain multi-dimensional information on scientific research results through a distributed data acquisition module, record author collaboration networks, citation relationships, and keyword distribution characteristics, identify the initial mapping structure of key entities based on node association, and generate a dynamic association graph of scientific research results; S2: Based on the dynamic association graph of the scientific research results, extract the association strength between nodes and the regional abnormal distribution characteristics, combine the segment association difference with the preset threshold range to perform deviation analysis, filter out the deviation area and adjust the information processing direction, and generate the regional data optimization instruction set; S3: calling the regional data optimization instruction set to adjust the information collection strategy, synchronously recording the time axis response value and peak offset of the information collection module, and extracting the synchronization offset range of the corresponding region by comparing the time axis and amplitude changes before and after the adjustment to obtain the regional information alignment status table; S4: calling the synchronization stable segment information sequence in the regional information alignment state table, identifying the information intensity change in the regional differentiation time window, performing offset analysis with the change threshold set by the reconstruction unit, marking the area to be adjusted, and forming a regional information adjustment marking map.

2. The method for clustering and recommending scientific research results based on knowledge graph according to claim 1, characterized in that: The dynamic association diagram of scientific research results includes the author cooperation network distribution curve, citation relationship characteristic parameters, and node classification identification; the regional data optimization instruction set includes the information collection strategy parameter set, association strength control value, and segment information compensation factor; the regional information alignment status table includes the synchronization time offset, peak response difference value, and alignment status identification code; the regional information adjustment mark diagram includes the segment position point to be adjusted, the intensity offset judgment result, and the structural change response identification.

3. The scientific research achievement clustering recommendation method based on knowledge graph according to claim 1 is characterized in that: The distributed data acquisition module adopts a multi-channel heterogeneous data fusion architecture, including: Author collaboration network acquisition unit: configured to obtain author collaboration intensity values ​​through the academic social network API, and calculate the product of collaboration frequency and time decay factor as the node association benchmark value; Citation relationship analysis unit: configured to extract the citation tree structure characteristics of the document, calculate the node influence coefficient through the PageRank algorithm, and generate a dynamic weight value based on the citation time window; Keyword distribution modeling unit: This unit is configured to use the LDA topic model to analyze the summary of the results, calculate the TF-IDF drift of the keywords in the time dimension, and generate a topic evolution feature vector.

4. The scientific research achievement clustering recommendation method based on knowledge graph according to claim 1 is characterized in that: The steps for obtaining the dynamic association diagram of scientific research results are specifically as follows: S111: Obtain multi-dimensional information on scientific research results through the distributed data acquisition module, record the response values ​​of the author cooperation network in different research fields, and convert the cooperation intensity index based on the calibration coefficient to form an operating status file in the current research field and obtain the author cooperation network parameter set; S112: Based on the author collaboration network parameter set, record the changes in collaboration intensity of the author collaboration network under differentiated research conditions, analyze the mapping relationship between research conditions and indicator volatility, reconstruct the information distribution characteristics in the scientific research environment, evaluate the stability of key nodes, and generate a dynamic correlation diagram of scientific research results.

5. The scientific research achievement clustering recommendation method based on knowledge graph according to claim 1 is characterized in that: The steps for obtaining the regional data optimization instruction set are specifically as follows: S211: Based on the dynamic association graph of scientific research results, identify the association strength and abnormal distribution characteristics between nodes, extract the information change rate per unit time in continuous nodes, and perform interval classification based on the association gradient value to identify the response interval of the information change on the node, and obtain the association response change interval; S212: Call the associated response change interval, scan the difference fluctuation amplitude and error trend of the node segment according to the ratio of the node segment information difference to the upper and lower limit information in the path, compare the matching degree of real-time information with the node distribution trend, calculate the information offset degree value, judge the abnormal distribution area in the node, extract the path position group that needs to be adjusted, and generate the regional data optimization instruction set.

6. The method for clustering and recommending scientific research results based on knowledge graph according to claim 1, characterized in that: The steps for obtaining the region information alignment status table are specifically as follows: S311: Calling the regional data optimization instruction set to adjust the information collection strategy, comparing the current information value according to the adjustment node number and the target information configuration, performing high- and low-priority information difference adjustment, and recording the response start time, peak time, and peak amplitude to obtain an adjustment path response time series group; S312: According to the adjustment path response time series group, extract the start time, peak time and amplitude of the nodes before and after the adjustment, identify the offset difference sequence, calculate the regional synchronization offset strength value, map the strength value with the node distribution, filter the node group within the synchronization range and arrange the information timing sequence to obtain the regional information alignment status table.

7. The method for clustering and recommending scientific research results based on knowledge graph according to claim 1, characterized in that: The steps for obtaining the region information adjustment mark map are specifically as follows: S411: calling the synchronization stable segment information sequence in the region information alignment state table, extracting the information intensity per unit time of the synchronization stable segment, and comparing it with the intensity sequence of the corresponding position of the adjacent region information segment, identifying the time deviation of the information intensity, and obtaining the region information offset value; S412: Based on the region information offset value and the change threshold set by the reconstruction unit, determine the difference between the value of each region information offset and the threshold, filter the offset regions exceeding the threshold, and obtain the block change determination coefficient; S413: Based on the block change judgment coefficient, detect the information offset trend in the area and the continuity between the offset positions, mark the areas where the offset trend is stably rising or falling and spatially continuous, calculate the regional offset mark value, combine the spatial range of the offset area, identify the connected identification partition number and fill it into the mark map to form the regional information adjustment mark map.

8. The method for clustering and recommending scientific research results based on knowledge graph according to claim 1, characterized in that: The method further comprises: S5: calling the positioning correction segment information in the regional information adjustment mark map, identifying the adjusted high and low priority information combination value, reconstructing the regional node response state according to the associated feature relationship corresponding to the combination information, and outputting a monitoring area state assessment value list; The monitoring area status assessment value list includes a high-priority and low-priority information fluctuation comparison value, a regional response difference factor, and a response status level after reconstruction.

9. The method for clustering and recommending scientific research results based on knowledge graph according to claim 8, characterized in that: The steps for obtaining the monitoring area status assessment value list are specifically as follows: S511: Calling the positioning correction segment information in the regional information adjustment mark map, extracting high-priority and low-priority information pairs of the positioning segments, matching the node coordinates with the path numbers, and grouping the combination information values ​​in spatial sequence to generate a dual-path positioning combination value set; S512: Based on the dual-path positioning combination value set, performing fluctuation difference discrimination on the combination value, extracting discernible information pairs, screening regional response groups of fluctuation characteristics according to the set dual-priority response threshold, and generating a dual-priority characteristic fluctuation rate sequence; S513: Analyze the mapping relationship between fluctuation parameters and node states based on the dual-priority characteristic fluctuation rate sequence, perform dual-priority state inversion of regional nodes, identify the node states of the positioning segments, and output a list of monitoring area state assessment values.

10. A scientific research results clustering recommendation system based on knowledge graph, characterized in that: The system is used for the scientific research achievement clustering recommendation method based on knowledge graph according to any one of claims 1 to 9, and the system includes: The distributed data acquisition module acquires multi-dimensional information of scientific research results through a multi-source academic feature acquisition matrix, performs a horizontal comparison of node status information in different research fields, identifies information response differences by path grouping, and integrates the information amplitude change sections of the entire path to construct a dynamic correlation diagram of scientific research results. Based on the dynamic correlation graph of the scientific research results, the correlation positioning module extracts the regional dual-priority information ratio sequence, identifies the response difference between the paths and screens the deviation mutation segments, identifies the intersection points of the correlation trends, determines the scope of the area to be adjusted, and generates the correlation segment positioning set; The information synchronization module adjusts the ratio of high-priority and low-priority information collection within the indicated area based on the associated segment positioning set, records the time axis response value and peak offset amplitude sequence of the data collection matrix within the corresponding segment, compares the response differences between the paths before and after the adjustment, integrates the stable synchronization point group, and establishes an information alignment synchronization status table; The correlation structure identification module aligns the synchronization state table based on the information, extracts the amplitude change in the synchronization segment information, compares the floating interval of the adjacent path response according to the time window, maps the offset exceeding position to the two-dimensional imaging surface, marks the local correlation strength mutation area, and outputs the correlation structure change layer; The state parameter assessment module extracts the corresponding high and low priority information combination values ​​based on the marked areas in the associated structure change layer, calculates the unit area fluctuation response in combination with the set reference volatility value table, and outputs a list of monitoring area state assessment values.

Images