Customer interaction data retrieval method based on dynamic data index

Abstract

The invention discloses a client interaction data retrieval method based on a dynamic data index, particularly relates to the technical field of data retrieval, and is used for solving the problem that long-term stable features and short-term burst features cannot be considered at the same time in the existing dynamic index. The method comprises the following steps: performing segmented aggregation analysis on historical customer interaction data in a long-term range, and extracting a stable feature baseline; a self-adaptive multi-scale burst feature detection algorithm is applied to recent client interaction data, and irregular short-term burst features are recognized; the short-term burst features are compared with the stable feature base line, feature items which do not conform to a preset quality threshold are removed, and composite feature weight distribution is formed; carrying out dynamic monitoring on the low-correlation feature group, evaluating potential risks of the low-correlation feature group and recording inducements of the low-correlation feature group; and finally, retrieving new input data based on the dynamic data index structure updated in real time. According to the method, comprehensive processing and efficient retrieval of complex feature modes are realized, and the capture capability of burst information and the adaptability of an index structure are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data retrieval, and more specifically, to a method for retrieval of customer interaction data based on dynamic data indexing. Background Art

[0002] As the scale and complexity of customer interaction data continue to grow, dynamic data indexing technology has been widely used in the organization and retrieval of real-time data. In practical applications, customer interaction data usually has significant time characteristics, including long-term stable characteristics that gradually change over time, and short-term burst characteristics that may suddenly occur due to specific events or behaviors; long-term stable characteristics refer to statistical characteristics in customer interaction data that remain stable over a long period of time, such as the average number of visits or the stable value of duration; short-term burst characteristics are abnormal changes caused by specific events or behaviors, such as a sharp increase in the number of visits or a sharp change in the length of stay in a short period of time. The two are independent of each other in terms of time characteristics but exist simultaneously.

[0003] Traditional dynamic indexing strategies mainly optimize a single feature and cannot take into account the coexistence of multiple feature types in the data. This limitation is particularly evident in data processing involving complex customer behavior patterns. In existing technologies, dynamic indexing is difficult to simultaneously process data with both long-term stable features and short-term burst features, which will lead to insufficient capture of burst information and affect the accuracy and real-time performance of data retrieval. Summary of the invention

[0004] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a customer interaction data retrieval method based on dynamic data indexing to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The customer interaction data retrieval method based on dynamic data indexing includes the following steps:

[0007] S1: Perform segmented aggregation analysis on historical customer interaction data over a long period of time, and extract stable feature values ​​from the long-term data as a stable feature baseline;

[0008] S2: Apply an adaptive multi-scale burst feature detection algorithm to recent customer interaction data to identify irregular short-term burst features by adjusting the time window and data scale;

[0009] S3: Compare the short-term burst features with the stable feature baseline, remove the feature items that do not meet the requirements according to the preset quality threshold, and form a composite feature weight distribution;

[0010] S4: Dynamically monitor the stability of low-correlation feature groups in the index, evaluate the potential risk of weakly correlated features triggering sudden behaviors in subsequent data based on the weak co-occurrence probability analysis between features, and record potential causes;

[0011] S5: The dynamic data index structure is updated in real time based on the composite feature weight distribution and potential inducements, and the newly input customer interaction data is retrieved based on the real-time updated dynamic data index structure.

[0012] In a preferred embodiment, segmented aggregation analysis is performed on historical customer interaction data within a long range to extract stable feature values ​​in the long-term data as a stable feature baseline, specifically including:

[0013] S101: Acquire historical customer interaction data within a long-term range, divide the historical customer interaction data into multiple continuous time periods according to a predetermined time interval, and mark the start and end time of each time period;

[0014] S102: extracting data features from the historical customer interaction data in each time period according to preset feature indicators, where the data features include the number and duration of customer interaction behaviors;

[0015] S103: Performing statistical processing on the data features in all time periods, analyzing the distribution of the data features in each time period, and calculating the time variation range of each data feature;

[0016] S104: According to the time variation range result of the statistical analysis, data features with a smaller time variation range are screened out and marked as stable feature candidate values;

[0017] S105: Perform cluster analysis on the stable feature candidate values, extract the central value of the clustering result as the stable feature baseline, and record it as the stable feature value of the long-term data.

[0018] In a preferred embodiment, an adaptive multi-scale burst feature detection algorithm is applied to recent customer interaction data to identify irregular short-term burst features by adjusting the time window and data scale, specifically including:

[0019] S201: Acquire recent customer interaction data, and divide the recent customer interaction data into multiple continuous time windows according to a preset time range;

[0020] S202: for the recent customer interaction data in each time window, calculate the recent feature value according to the multi-scale feature extraction rule, where the recent feature value includes the interaction frequency and the behavior persistence;

[0021] S203: dynamically adjusting the time window length and data scale parameters, respectively calculating the multi-scale feature values ​​under each time window, and forming a multi-scale feature matrix;

[0022] S204: Analyze the change pattern of the multi-scale feature matrix, screen and mark features with large mutation amplitudes as short-term burst features.

[0023] In a preferred embodiment, the short-term burst feature is compared with the stable feature baseline, and the feature items that do not meet the requirements are eliminated according to the preset quality threshold to form a composite feature weight distribution, which specifically includes:

[0024] S301: Obtain a short-term burst feature set and a stable feature baseline set, and perform unified formatting processing;

[0025] S302: Pairing the short-term burst features with the stable feature baselines one by one according to the feature type and time window to form a comparable feature pair set;

[0026] S303: For each feature pair in the feature pair set, calculating the feature difference between the short-term burst feature and the stable feature baseline;

[0027] S304: Compare the feature difference with the preset quality threshold, remove the feature items whose difference exceeds the threshold range, and only retain the feature items that meet the requirements;

[0028] S305: Calculate weights for the retained feature items according to feature differences to generate a composite feature weight distribution.

[0029] In a preferred embodiment, dynamic stability monitoring is performed on the feature groups with low correlation in the index, and based on the weak co-occurrence probability analysis between the features, the potential risk of weakly correlated features inducing sudden behaviors in subsequent data is evaluated, and potential inducements are recorded, specifically including:

[0030] S401: extracting feature pairs with lower weights from the composite feature weight distribution to form a low-correlation feature group set;

[0031] S402: Calculate the co-occurrence probability between features based on the feature items in the low-correlation feature group set to obtain a co-occurrence matrix between each pair of features;

[0032] S403: Analyze the potential correlation between low-correlation features based on the weak co-occurrence probability in the co-occurrence matrix and identify potential causes;

[0033] S404: Dynamically monitor potential triggers in low-correlation feature groups, track trends of slight changes between features, and assess the potential risk of weakly correlated features triggering sudden behaviors in subsequent data.

[0034] In a preferred embodiment, the co-occurrence probability between features is calculated based on the feature items in the low correlation feature group set to obtain a co-occurrence matrix between each pair of features, specifically:

[0035] Count the number of co-occurrences of the two features in the feature pair in all historical time windows, and the total number of occurrences of each individual feature in the feature pair;

[0036] The co-occurrence probability is calculated using the co-occurrence frequency and the total number of occurrences. The formula is: Among them, P q，co represents the co-occurrence probability of a feature pair, N1 represents the total number of occurrences of the first feature in the feature pair, N2 represents the total number of occurrences of the second feature in the feature pair, Min(N1, N2) represents the smaller value of the total number of occurrences of the two features in the feature pair, and C q Indicates the number of times the two features in a feature pair co-occur in historical data;

[0037] The calculation results are expressed in the form of a co-occurrence matrix: M = {P ab，co |a, b∈G}; where M represents the co-occurrence matrix of low-correlation feature groups, P ab，co represents the co-occurrence probability of the feature pair (a, b), a and b represent any two features in the low-correlation feature group, and G represents the set of low-correlation feature groups.

[0038] In a preferred embodiment, the potential associations between low-correlation features are analyzed based on the weak co-occurrence probabilities in the co-occurrence matrix to identify potential causes, specifically:

[0039] Set a co-occurrence probability threshold, extract feature pairs that satisfy the co-occurrence probability less than the co-occurrence probability threshold and mark them as low co-occurrence probability feature pairs;

[0040] For each set of low co-occurrence probability feature pairs, combine the feature value fluctuations within the time window to determine whether there is a potential cause;

[0041] The specific potential cause identification rule is: if the co-occurrence probability of a feature pair is less than the co-occurrence probability threshold, and the fluctuation patterns of the two features in the feature pair show a related trend, then it is marked as a potential cause.

[0042] In a preferred embodiment, the dynamic data index structure is updated in real time based on the composite feature weight distribution and the potential inducement, and the newly input customer interaction data is retrieved based on the real-time updated dynamic data index structure, specifically including:

[0043] S501: extracting the weight of each feature pair and the potential risk probability of the sudden behavior from the composite feature weight distribution to generate an updated data set;

[0044] S502: adjusting the priority of the feature node according to the weight of the feature pair in the updated data set, and redefining the correlation strength between the feature pairs in combination with the potential risk probability of the sudden behavior to optimize the index structure;

[0045] S503: cleaning up low-weight feature nodes in the dynamic data index structure, and sorting high-weight feature nodes to improve retrieval efficiency and structure compactness;

[0046] S504: Use the real-time updated dynamic data index structure to search for newly input customer interaction data, match high-weight feature pairs associated with the input data, and return search results.

[0047] Technical effects and advantages of the customer interaction data retrieval method based on dynamic data indexing of the present invention:

[0048] 1. Through segmented aggregation analysis, the stable feature baseline of long-term data is extracted, and the short-term burst features are accurately identified by combining the adaptive multi-scale burst feature detection algorithm, which realizes the effective separation and comprehensive processing of different time features in customer interaction data. It can identify, compare and fuse short-term burst features and long-term stable features respectively, overcoming the problem that traditional dynamic indexing strategies cannot take into account multiple feature patterns, and provides an efficient feature basis for the subsequent optimization of index structure.

[0049] 2. By dynamically monitoring the stability of low-correlation feature groups and combining the weak co-occurrence probability analysis between features, it is possible to assess in advance the risk of sudden behaviors that may be caused by weakly correlated features. Furthermore, through the real-time update and efficient retrieval of the dynamic data index structure, it ensures the rapid capture and processing of sudden information and potential inducements in dynamic data. This index optimization strategy based on feature weights and risk probabilities significantly improves the accuracy and real-time performance of retrieval, effectively making up for the shortcomings of traditional methods in processing complex customer interaction data, and is suitable for scenarios with high-frequency changes and complex features. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a schematic diagram of the customer interaction data retrieval method based on dynamic data indexing of the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Example: Figure 1 The present invention provides a method for retrieving customer interaction data based on dynamic data index, which comprises the following steps:

[0053] S1: Perform segmented aggregation analysis on historical customer interaction data over a long period of time and extract stable feature values ​​from the long-term data as a stable feature baseline.

[0054] S2: Apply an adaptive multi-scale burst feature detection algorithm to recent customer interaction data to identify irregular short-term burst features by adjusting the time window and data scale.

[0055] S3: Compare the short-term burst features with the stable feature baseline, eliminate the feature items that do not meet the requirements according to the preset quality threshold, and form a composite feature weight distribution.

[0056] S4: Dynamically monitor the stability of low-correlation feature groups in the index. Based on the weak co-occurrence probability analysis between features, evaluate the potential risk of weakly correlated features causing sudden behaviors in subsequent data and record potential causes.

[0057] S5: The dynamic data index structure is updated in real time based on the composite feature weight distribution and potential inducements, and the newly input customer interaction data is retrieved based on the real-time updated dynamic data index structure.

[0058] Perform segmented aggregation analysis on historical customer interaction data over a long period of time, and extract stable feature values ​​from the long-term data as a stable feature baseline, including:

[0059] S101: Acquire historical customer interaction data within a long period, divide the historical customer interaction data into multiple continuous time periods according to a predetermined time interval, and mark the start and end time of each time period:

[0060] For example, if the time interval is one day, the daily data corresponds to one time period; if the time interval is one week, the weekly data corresponds to one time period. The time period division rules should ensure that the time periods are continuous and non-overlapping.

[0061] S102: extracting data features from the historical customer interaction data in each time period according to preset feature indicators, where the data features include the number and duration of customer interaction behaviors:

[0062] After completing the time period division, perform feature extraction on the historical customer interaction data in each time period. Feature extraction needs to be completed based on preset indicators, including:

[0063] Number of customer interactions: Calculate the total number of all customer interactions within a period of time as a basic indicator for measuring customer activity.

[0064] Duration of customer interaction behavior: The total duration of all customer interaction behaviors within a period of time is counted as an indicator to measure the depth of customer interaction participation.

[0065] By analyzing the specific content of each interactive data, the characteristic value of each time period is gradually calculated.

[0066] S103: Perform statistical processing on the data features in all time periods, analyze the distribution of the data features in each time period, and calculate the time variation range of each data feature:

[0067] The eigenvalues ​​extracted from each time period are statistically processed, the distribution of eigenvalues ​​in all time periods is analyzed, and the time variation range of each feature is calculated. The specific analysis includes the following steps: statistically analyzing the distribution of eigenvalues ​​in each time period to determine the central tendency and dispersion of eigenvalues; calculating the fluctuation range of each eigenvalue in different time periods to evaluate the stability of the feature in the time dimension; and comprehensively evaluating the eigenvalues ​​in all time periods to record the variation range of each feature.

[0068] S104: According to the time variation range results of the statistical analysis, data features with a smaller time variation range are screened out and marked as stable feature candidate values:

[0069] According to the results of statistical analysis, the data features are screened. The screening rules need to be set in combination with the range of changes in the feature values. For example, only feature values ​​with a small time range are screened out and marked as stable feature candidate values; feature values ​​with a large time range are excluded to ensure the stability of the candidate features.

[0070] Among them, the specific screening conditions of the time change range should be set according to business needs and data characteristics, such as by setting a predefined fluctuation threshold.

[0071] S105: Perform cluster analysis on the stable feature candidate values, extract the central value of the clustering result as the stable feature baseline, and record it as the stable feature value of the long-term data:

[0072] The cluster analysis process includes the following steps:

[0073] The stationary feature candidate values ​​are divided into several categories, and the cluster center is calculated according to the distribution of each category of feature values;

[0074] The cluster center is the stable characteristic baseline, which is used to represent the stable characteristic value of data in the long term.

[0075] The specific method of cluster analysis can adopt an algorithm suitable for business needs, such as the average clustering method based on the eigenvalue distribution or other clustering tools with efficient computing capabilities. Finally, the stable feature baseline is recorded as the stable eigenvalue of the long-term data.

[0076] An adaptive multi-scale burst feature detection algorithm is applied to recent customer interaction data to identify irregular short-term burst features by adjusting the time window and data scale, including:

[0077] S201: Acquire recent customer interaction data, and divide the recent customer interaction data into multiple continuous time windows according to a preset time range:

[0078] Among them, each time window has a clear start time and end time, and the length of the time window is determined by the actual analysis requirements.

[0079] S202: For the recent customer interaction data in each time window, calculate the recent feature value according to the multi-scale feature extraction rule. The recent feature value includes the interaction frequency and behavior persistence:

[0080] The multi-scale feature extraction rules are as follows:

[0081] First, different time scale parameters are selected, such as small time periods (e.g., 10 minutes) and large time periods (e.g., 1 hour), to extract features at different time scales.

[0082] Then, the average characteristic value and the change characteristic value within the time scale are calculated to describe the distribution law of customer behavior in the short-term and long-term scales.

[0083] Finally, data at different scales are combined to form a comprehensive multi-scale feature representation.

[0084] The expression of interaction frequency is: The expression of behavior persistence is: Among them, F i，k represents the interaction frequency of the i-th time window under the k-th scale parameter, D i，k represents the behavior persistence of the i-th time window under the k-th scale parameter, n i，k represents the total number of interactive behaviors in the i-th time window under the k-th scale parameter, S k represents the kth scale parameter, j represents the number of the jth interaction behavior in a certain time window, and e j represents the end time of the jth interaction behavior, s j Indicates the start time of the jth interaction behavior.

[0085] The 1 in the expression of the interaction frequency indicates that each interaction behavior contributes one count, which is used to count the frequency of the interaction behavior.

[0086] The comprehensive multi-scale feature representation includes two dimensions: interaction frequency and behavior persistence. In each time window, a feature set is generated through multi-scale feature extraction, which is represented as an interaction frequency feature set and a behavior persistence feature set. The interaction frequency feature set contains interaction frequency values ​​at different scales, while the behavior persistence feature set contains behavior duration values ​​at different scales. These two sets together constitute the comprehensive feature representation of each time window, which is used to describe the complete feature distribution of customer interaction behavior at multiple scales.

[0087] S203: Dynamically adjust the time window length and data scale parameters, calculate the multi-scale feature values ​​under each time window respectively, and form a multi-scale feature matrix:

[0088] Time window length adjustment: The time window length is controlled by the parameter sequence, and data features of different time granularities are generated by setting different time intervals.

[0089] Data scale parameter adjustment: Generate eigenvalue matrix columns at different scales by changing the data scale parameters.

[0090] Calculate the eigenvalue of the i-th time window under the k-th scale parameter, the expression is as follows: M ik =f(W i , S k ), where M ik represents the eigenvalue of the i-th time window under the k-th scale parameter, W i represents the i-th time window, f(W i , S k ) represents the feature extraction function (determined by the time window and data scale).

[0091] S204: Analyze the change pattern of the multi-scale feature matrix, screen and mark the features with large mutation amplitude as short-term burst features:

[0092] Calculate the change rate of the characteristic values ​​in adjacent time windows to determine whether there is a sudden change; filter the characteristic values ​​whose change rate exceeds the corresponding preset threshold, and mark the characteristic values ​​that meet the conditions as short-term burst characteristics.

[0093] Among them, the preset threshold corresponding to the change rate is a fixed value set according to specific business needs and data characteristics, which is used to determine whether the rate of change of the feature value is abnormal. Each threshold corresponds to a specific time window and scale parameter, which indicates the minimum amplitude of the feature change considered significant under this condition, and is used to screen the mutation feature value.

[0094] The short-term burst features that have been screened out are recorded, and their corresponding time windows and related behavior data are marked. For example, the time window number, change rate, and associated behavior features of the burst features are recorded for subsequent use.

[0095] Compare the short-term burst features with the stable feature baseline, remove the feature items that do not meet the requirements according to the preset quality threshold, and form a composite feature weight distribution, including:

[0096] S301: Obtain a short-term burst feature set and a stable feature baseline set, and perform unified formatting to ensure that the feature data can be directly compared:

[0097] The short-term burst feature set and the stable feature baseline set are obtained respectively. The feature set includes different types of data features (such as interaction frequency, behavior persistence, etc.). In order to ensure the comparability between feature sets, the feature sets need to be formatted. The formatting process includes:

[0098] Normalize all feature values ​​to the same range, for example, to [0, 1].

[0099] The feature set is classified and sorted according to feature type to ensure that features of the same type can be directly compared.

[0100] The feature set is time-aligned according to the time window to ensure that the short-term burst features and the stable feature baseline have a corresponding relationship at the same time point.

[0101] S302: According to the feature type and time window, the short-term burst feature is paired with the stable feature baseline item by item to form a comparable feature pair set:

[0102] If the type of the short-term burst feature is consistent with the type of the stable feature baseline and they belong to the same time window, they are paired; if the short-term burst feature meets the conditions with multiple stable feature baselines, the baseline with the closest time distance is selected for pairing.

[0103] S303: For each feature pair in the feature pair set, calculate the feature difference between the short-term burst feature and the stable feature baseline:

[0104] The difference between the short-term burst feature and the stable feature baseline in the feature pair is calculated and the absolute value is taken to obtain the feature difference.

[0105] The calculation of feature difference needs to be performed on all feature pairs in the feature pair set one by one to generate a complete difference set.

[0106] S304: Compare the feature difference with the preset quality threshold, remove the feature items whose difference exceeds the threshold range, and only retain the feature items that meet the requirements:

[0107] If the feature difference is less than or equal to the preset quality threshold, the feature pair is retained; if the feature difference is greater than the preset quality threshold, the feature pair is discarded.

[0108] Among them, the preset quality threshold is a fixed value set according to specific business needs and feature attributes. It is used to determine whether the difference between short-term burst features and stable feature baselines meets the requirements. Each feature type can correspond to a different quality threshold to measure the acceptable range of feature differences and ensure the rationality and accuracy of the screening results.

[0109] S305: Calculate the weights of the retained feature items according to the feature differences to generate a composite feature weight distribution:

[0110] The weight of a feature pair is the inverse of the feature difference corresponding to the feature pair.

[0111] The weights of all feature pairs are integrated to form a composite feature weight distribution, which includes the weights of all feature pairs.

[0112] Dynamic stability monitoring is performed on low-correlation feature groups in the index. Based on the weak co-occurrence probability analysis between features, the potential risk of weakly correlated features causing sudden behaviors in subsequent data is evaluated, and potential causes are recorded, including:

[0113] S401: Extracting feature pairs with lower weights from the composite feature weight distribution to form a low-correlation feature group set:

[0114] In the composite feature weight distribution, each feature pair has a corresponding weight value. The weight value is used to reflect the strength of the association between the feature pairs. A lower weight value indicates a weaker association between the two features. First, feature pairs with weight values ​​lower than the preset threshold are extracted from the composite feature weight distribution to form a low-association feature group set.

[0115] The extraction rules are as follows:

[0116] Set the weight threshold and extract the K q <T K The set of feature pairs.

[0117] The extraction result is expressed as: G = {P q |K q <T K}; where G represents a set of low-correlation feature groups, P q represents the qth feature pair, K q represents the weight of the qth feature pair, T K Represents the weight threshold.

[0118] The above rules ensure that only feature pairs with low weights are included in the subsequent analysis.

[0119] Among them, T KIt is a fixed value set according to the specific business scenario and feature correlation strength. It is used to distinguish low-correlation features from high-correlation features. When the weight of a feature pair is lower than the threshold, it is considered that its correlation is low and needs to be included in the low-correlation feature group for further analysis and monitoring.

[0120] S402: Calculate the co-occurrence probability between features based on the feature items in the low-correlation feature group set, and obtain the co-occurrence matrix between each pair of features:

[0121] In the low-correlation feature group set, the historical data of each feature pair is counted, and the co-occurrence probability of the feature pair in the historical data is calculated. The co-occurrence probability is used to quantify the frequency of two features appearing at the same time. The specific process is as follows:

[0122] Co-occurrence frequency statistics: Count the number of co-occurrences of two features in a feature pair in all historical time windows.

[0123] Total occurrence count: Counts the total occurrence of each individual feature in a feature pair.

[0124] Calculate the co-occurrence probability: Use the co-occurrence frequency and the total number of occurrences to calculate the co-occurrence probability. The formula is: Among them, P q，co Represents the co-occurrence probability of a feature pair, which is used to quantify the frequency of two features appearing simultaneously in a feature pair; N1 represents the total number of occurrences of the first feature in the feature pair; N2 represents the total number of occurrences of the second feature in the feature pair; Min(N1, N2) represents the smaller value of the total number of occurrences of the two features in the feature pair, which is used to standardize the co-occurrence probability; C q Indicates the number of times the two features in a feature pair co-occur in historical data.

[0125] The calculation results are expressed in the form of a co-occurrence matrix: M = {P ab，co |a, b∈G}; where M represents the co-occurrence matrix of low-correlation feature groups; P ab，co It represents the co-occurrence probability of the feature pair (a, b), which is used to quantify the co-occurrence frequency of feature a and feature b in historical data; a and b represent any two features in the low-correlation feature group.

[0126] S403: Analyze the potential correlation between low-correlation features based on the weak co-occurrence probability in the co-occurrence matrix and identify potential causes:

[0127] Analyze each co-occurrence probability in the co-occurrence matrix, combine the historical distribution of feature pairs, and identify potential associations between low-correlation features. The steps of potential association analysis are as follows:

[0128] A co-occurrence probability threshold is set, and feature pairs whose co-occurrence probability is less than the co-occurrence probability threshold are extracted and marked as low co-occurrence probability feature pairs.

[0129] For each set of low co-occurrence probability feature pairs, the fluctuation of feature values ​​within the time window is combined to determine whether there is a potential cause.

[0130] The specific potential cause identification rule is: if the co-occurrence probability of a feature pair is less than the co-occurrence probability threshold, and the fluctuation patterns of the two features in the feature pair show a related trend, then it is marked as a potential cause.

[0131] Among them, the co-occurrence probability threshold is an important criterion for screening feature pairs. It is set according to the historical co-occurrence characteristics of the feature pairs to distinguish significant co-occurrence from weak co-occurrence. When the co-occurrence probability of a feature pair is lower than the threshold, its co-occurrence correlation is considered to be weak and requires further analysis.

[0132] The relative trend of the eigenvalues ​​is calculated using the following formula: Among them, D q represents the correlation trend coefficient of the qth feature pair, which is used to quantify the degree of linear correlation between two feature values; X t and Y t Respectively represent the eigenvalues ​​of the two features in the feature pair at time t; and represent the means of the two eigenvalues ​​respectively.

[0133] The feature value is used to quantify the performance of the feature at time t, usually including interaction frequency, behavior duration, etc.

[0134] According to the correlation trend coefficient, the range of its value is used to determine whether the fluctuation patterns of the two features in the feature pair show a correlation trend. When the correlation trend coefficient is greater than 0, the fluctuations of the two features are positively correlated, that is, the feature values ​​rise or fall synchronously over time; when the correlation trend coefficient is less than 0, the fluctuations of the two features are negatively correlated, that is, when one feature value rises, the other feature value falls; when the correlation trend coefficient is equal to 0, the two features have no correlation, and the fluctuation patterns do not affect each other.

[0135] Set the positive correlation threshold and negative correlation threshold, which are important indicators for judging the correlation trend of feature pairs. The positive correlation threshold represents the lowest coefficient value of the significant positive correlation between two features, and the negative correlation threshold represents the highest coefficient value of the significant negative correlation between two features. Both are used together to screen feature pairs with significant correlation trends.

[0136] If the correlation trend coefficient is greater than or equal to the positive correlation threshold or the correlation trend coefficient is less than or equal to the negative correlation threshold, it is considered that the fluctuation patterns of the two features in the feature pair show a correlation trend. The judgment result of the correlation trend can be used to further analyze the potential feature correlation or risk assessment.

[0137] S404: Dynamically monitor potential triggers in low-correlation feature groups, track trends of slight changes between features, and assess the potential risk of weakly correlated features triggering sudden behaviors in subsequent data:

[0138] For the identified potential triggers, dynamic monitoring is implemented to track the changing trends between characteristics, and the potential risk probability of triggering sudden behavior in the future is calculated based on the characteristic fluctuations within the time window.

[0139] For each set of potential causal feature pairs, the changes in their feature values ​​are continuously tracked, and the fluctuation amplitudes in different time windows are recorded.

[0140] Based on the fluctuation amplitude and time weight, the potential risk probability of sudden behavior is calculated. The potential risk probability of sudden behavior is the product of the co-occurrence probability of the feature pair and the time weight.

[0141] Among them, the time weight is a dynamic parameter set according to the timeliness of the time window. It is used to reflect the impact of the latest data on feature analysis. The weight is usually calculated through a predefined decay function. The closer the window is to the current time, the higher the weight. Common methods include exponential decay or linear decline models.

[0142] The greater the potential risk probability of sudden behavior, the greater the potential risk of weakly correlated features causing sudden behavior in subsequent data, indicating that the co-occurrence relationship and time trend between weakly correlated features are more likely to cause abnormal fluctuations, which shows that in subsequent data, these features are more likely to cause sudden behavior and need to be monitored and evaluated.

[0143] Based on the composite feature weight distribution and potential inducements, the dynamic data index structure is updated in real time, and the newly input customer interaction data is retrieved based on the real-time updated dynamic data index structure, including:

[0144] S501: Extract the weight of each feature pair and the potential risk probability of the sudden behavior from the composite feature weight distribution to generate an updated data set:

[0145] From the composite feature weight distribution, the weight of each feature pair is extracted one by one, along with the potential risk probability of the sudden behavior obtained from the potential cause analysis. The feature weight value represents the association strength of the feature pair, while the risk probability is used to quantify the risk that the feature pair may cause the sudden behavior.

[0146] The extracted feature weight values ​​and risk probabilities are combined according to feature pairs to form an updated data set for adjusting the dynamic data index structure.

[0147] S502: Adjust the priority of the feature node according to the weight of the feature pair in the updated data set, and redefine the correlation strength between the feature pairs in combination with the potential risk probability of the sudden behavior to optimize the index structure:

[0148] According to the weight of each feature pair, the priority of the corresponding feature node in the index structure is dynamically adjusted. The feature node with a higher weight value has a higher priority in the index structure.

[0149] Adjustment rules: Increase the priority index level of high-weight feature nodes in the index structure to ensure that they participate in retrieval matching faster; lower the priority of low-weight feature nodes to reduce the resource usage of the index structure.

[0150] Based on the potential risk probability of the sudden behavior of each feature pair, the association strength between feature nodes is redefined. For feature pairs with higher potential risk probability of sudden behavior, their association strength in the index structure is strengthened so that they can be monitored and processed first.

[0151] S503: Clean up the low-weight feature nodes in the dynamic data index structure, and sort the high-weight feature nodes to improve retrieval efficiency and structure compactness:

[0152] Perform cleaning operations on feature nodes whose weight values ​​are lower than the preset threshold in the dynamic data index structure: if the weight of a feature pair is less than the preset cleaning threshold, remove the feature node from the index structure; retain feature nodes whose feature pair weights are greater than or equal to the preset cleaning threshold, ensuring that only important features are retained in the index structure.

[0153] For the cleaned index structure, the remaining feature nodes are sorted in descending order using the weights of the feature pairs in the updated data set. The sorting rules are as follows: sort from high to low according to the weights of the feature pairs; feature nodes with high priority occupy high positions in the index structure to improve index retrieval efficiency.

[0154] After the cleaning and sorting operations, the resource positions in the index structure are reallocated to ensure that high-weight feature nodes occupy key positions while reducing the resource consumption of low-priority nodes to form a more compact index structure.

[0155] S504: Use the real-time updated dynamic data index structure to search for the newly input customer interaction data, match the high-weight feature pairs associated with the input data, and return the search results:

[0156] Preprocess the newly input customer interaction data and extract feature values ​​according to the feature rules in the dynamic data index structure; for example, the newly input data may contain user operation records (such as click frequency, visit duration, etc.). By parsing these records, feature values ​​such as "number of visits is 5 times" and "stay time is 120 seconds" can be extracted.

[0157] The extracted feature values ​​are matched one by one with the sorted high-weight feature nodes in the dynamic data index structure, and the matching rules are based on the correlation between the feature values ​​and the high-priority features in the index.

[0158] For example, a high-weight feature node in the index structure may record the feature of "the number of clicks is greater than 3 times and the dwell time is more than 100 seconds"; if the newly input data contains the feature values ​​of "the number of clicks is 5 times" and "the dwell time is 120 seconds", it is matched successfully with the node.

[0159] All successfully matched feature nodes are sorted in descending order according to their weight values, and the most relevant feature nodes are screened out and returned as part of the retrieval results.

[0160] For example, if the new input data matches two feature nodes at the same time (such as node A and node B), where the weight of node A is 0.8 and the weight of node B is 0.6, node A is returned to the search results first; the search results finally returned may include the matching feature node number and its associated contextual information (such as feature description, time range, etc.).

[0161] The returned search results are provided to subsequent analysis modules to support user behavior pattern analysis, anomaly detection or other business logic. The format of the search results may include information such as the matching feature node number, related description, and matching feature value.

[0162] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0163] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0164] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0165] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0166] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0167] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0168] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0169] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0170] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0171] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A customer interaction data retrieval method based on dynamic data indexing, characterized in that: The steps include: S1: Perform segmented aggregation analysis on historical customer interaction data over a long period of time, and extract stable feature values ​​from the long-term data as a stable feature baseline; S2: Apply an adaptive multi-scale burst feature detection algorithm to recent customer interaction data to identify irregular short-term burst features by adjusting the time window and data scale; S3: Compare the short-term burst features with the stable feature baseline, remove the feature items that do not meet the requirements according to the preset quality threshold, and form a composite feature weight distribution; S4: Dynamically monitor the stability of low-correlation feature groups in the index, evaluate the potential risk of weakly correlated features triggering sudden behaviors in subsequent data based on the weak co-occurrence probability analysis between features, and record potential causes; S5: The dynamic data index structure is updated in real time based on the composite feature weight distribution and potential inducements, and the newly input customer interaction data is retrieved based on the real-time updated dynamic data index structure.

2. The customer interaction data retrieval method based on dynamic data indexing according to claim 1 is characterized in that: Perform segmented aggregation analysis on historical customer interaction data over a long period of time, and extract stable feature values ​​from the long-term data as a stable feature baseline, including: S101: Acquire historical customer interaction data within a long-term range, divide the historical customer interaction data into multiple continuous time periods according to a predetermined time interval, and mark the start and end time of each time period; S102: extracting data features from the historical customer interaction data in each time period according to preset feature indicators, where the data features include the number and duration of customer interaction behaviors; S103: Performing statistical processing on the data features in all time periods, analyzing the distribution of the data features in each time period, and calculating the time variation range of each data feature; S104: According to the time variation range result of the statistical analysis, data features with a smaller time variation range are screened out and marked as stable feature candidate values; S105: Perform cluster analysis on the stable feature candidate values, extract the central value of the clustering result as the stable feature baseline, and record it as the stable feature value of the long-term data.

3. The method for retrieving customer interaction data based on dynamic data index according to claim 2, characterized in that: An adaptive multi-scale burst feature detection algorithm is applied to recent customer interaction data to identify irregular short-term burst features by adjusting the time window and data scale, including: S201: Acquire recent customer interaction data, and divide the recent customer interaction data into multiple continuous time windows according to a preset time range; S202: for the recent customer interaction data in each time window, calculate the recent feature value according to the multi-scale feature extraction rule, where the recent feature value includes the interaction frequency and the behavior persistence; S203: dynamically adjusting the time window length and data scale parameters, respectively calculating the multi-scale feature values ​​under each time window, and forming a multi-scale feature matrix; S204: Analyze the change pattern of the multi-scale feature matrix, screen and mark features with large mutation amplitudes as short-term burst features.

4. The method for retrieving customer interaction data based on dynamic data index according to claim 3, characterized in that: Compare the short-term burst features with the stable feature baseline, remove the feature items that do not meet the requirements according to the preset quality threshold, and form a composite feature weight distribution, including: S301: Obtain a short-term burst feature set and a stable feature baseline set, and perform unified formatting processing; S302: Pairing the short-term burst features with the stable feature baselines one by one according to the feature type and time window to form a comparable feature pair set; S303: For each feature pair in the feature pair set, calculating the feature difference between the short-term burst feature and the stable feature baseline; S304: Compare the feature difference with the preset quality threshold, remove the feature items whose difference exceeds the threshold range, and only retain the feature items that meet the requirements; S305: Calculate weights for the retained feature items according to feature differences to generate a composite feature weight distribution.

5. The method for retrieving customer interaction data based on dynamic data index according to claim 4, characterized in that: Dynamic stability monitoring is performed on low-correlation feature groups in the index. Based on the weak co-occurrence probability analysis between features, the potential risk of weakly correlated features causing sudden behaviors in subsequent data is evaluated, and potential causes are recorded, including: S401: extracting feature pairs with lower weights from the composite feature weight distribution to form a low-correlation feature group set; S402: Calculate the co-occurrence probability between features based on the feature items in the low-correlation feature group set to obtain a co-occurrence matrix between each pair of features; S403: Analyze the potential correlation between low-correlation features based on the weak co-occurrence probability in the co-occurrence matrix and identify potential causes; S404: Dynamically monitor potential triggers in low-correlation feature groups, track trends of slight changes between features, and assess the potential risk of weakly correlated features triggering sudden behaviors in subsequent data.

6. The method for retrieving customer interaction data based on dynamic data index according to claim 5, characterized in that: Based on the feature items in the low-correlation feature group set, the co-occurrence probability between features is calculated, and the co-occurrence matrix between each pair of features is obtained, which is specifically: Count the number of co-occurrences of the two features in the feature pair in all historical time windows, and the total number of occurrences of each individual feature in the feature pair; The co-occurrence probability is calculated using the co-occurrence frequency and the total number of occurrences. The formula is: Among them, P q，co represents the co-occurrence probability of a feature pair, N1 represents the total number of occurrences of the first feature in the feature pair, N2 represents the total number of occurrences of the second feature in the feature pair, Min(N1, N2) represents the smaller value of the total number of occurrences of the two features in the feature pair, and C q Indicates the number of times the two features in a feature pair co-occur in historical data; The calculation results are expressed in the form of a co-occurrence matrix: M = {P ab，co |a, b∈G}; where M represents the co-occurrence matrix of low-correlation feature groups, P ab，co represents the co-occurrence probability of the feature pair (a, b), a and b represent any two features in the low-correlation feature group, and G represents the set of low-correlation feature groups.

7. The method for retrieving customer interaction data based on dynamic data index according to claim 6, characterized in that: According to the weak co-occurrence probability in the co-occurrence matrix, the potential correlation between low-correlation features is analyzed to identify potential causes, specifically: Set a co-occurrence probability threshold, extract feature pairs that satisfy the co-occurrence probability less than the co-occurrence probability threshold and mark them as low co-occurrence probability feature pairs; For each set of low co-occurrence probability feature pairs, combine the feature value fluctuations within the time window to determine whether there is a potential cause; The specific potential cause identification rule is: if the co-occurrence probability of a feature pair is less than the co-occurrence probability threshold, and the fluctuation patterns of the two features in the feature pair show a related trend, then it is marked as a potential cause.

8. The method for retrieving customer interaction data based on dynamic data index according to claim 7, characterized in that: Based on the composite feature weight distribution and potential inducements, the dynamic data index structure is updated in real time, and the newly input customer interaction data is retrieved based on the real-time updated dynamic data index structure, including: S501: extracting the weight of each feature pair and the potential risk probability of the sudden behavior from the composite feature weight distribution to generate an updated data set; S502: adjusting the priority of the feature node according to the weight of the feature pair in the updated data set, and redefining the correlation strength between the feature pairs in combination with the potential risk probability of the sudden behavior to optimize the index structure; S503: cleaning up low-weight feature nodes in the dynamic data index structure, and sorting high-weight feature nodes to improve retrieval efficiency and structure compactness; S504: Use the real-time updated dynamic data index structure to search for newly input customer interaction data, match high-weight feature pairs associated with the input data, and return search results.

Images