A Personal Password Trustworthiness Assessment Method Integrating Biometrics and Rule Engine

Abstract

This invention discloses a method for evaluating the credibility of personal passwords by integrating biometrics and a rule engine, relating to the field of password evaluation technology. The method includes the following steps: acquiring the user's biometric data and password input data, and parsing the password input data to obtain password structure switching points; isolating and identifying the biometric data based on the password structure switching points to obtain a first isolation zone; extracting biometric change feature sequences from the first isolation zone, and recombining the judgment rules in a preset rule engine based on the biometric change feature sequences to obtain a password structure rule group; evaluating the credibility of the personal password based on the password structure rule group, and outputting the password credibility evaluation result. This solves the problem that existing technologies fail to correlate password structure changes with the natural fluctuations of biometrics during the input process, leading to reduced accuracy in evaluating the credibility of the password input process.

Description

Technical Field

[0001] This invention relates to the field of cryptographic evaluation technology, and more specifically, to a method for evaluating the credibility of personal cryptography that integrates biometrics and a rule engine. Background Technology

[0002] With the ever-increasing security requirements of information systems, password-based authentication remains one of the most prevalent authentication methods in various terminal devices, network services, and application systems. Existing password authentication mechanisms typically focus on verifying the password itself, such as determining access permission by comparing it to a preset password or checking its length and complexity. To enhance password authentication security, some existing technologies are beginning to incorporate user behavior characteristics or biometric information, such as analyzing a user's keystroke rhythm, input sequence, or touch behavior to assist in identifying the user. On the other hand, rule engines, as a technology for decision-making and reasoning based on preset logical rules, have been applied in areas such as access control and risk assessment.

[0003] However, in the password input process, existing technologies often treat the biometric data generated by the user as a continuous and homogeneous temporal sequence for overall analysis, focusing on statistical overall rhythm, average pressure, or global stability indicators, while lacking a detailed characterization and explicit expression of the internal logical structure of the password. In particular, they fail to differentiate and model the switching relationships between different character categories, password structure nodes, and the resulting changes in input behavior. In actual input scenarios, when users switch from letters to numbers, from ordinary characters to special characters, or when the case of characters changes, they often experience short-term, localized, and structurally oriented fluctuations in biometric features. These fluctuations are natural input phenomena resulting from the complexity of the password structure. However, because existing methods do not correlate these structural factors with changes in biometric features, these localized changes are easily submerged in the overall temporal sequence characteristics, making them difficult to explain reasonably. Consequently, the consistency of behavior and the degree of structural matching during the password input process cannot be accurately characterized, ultimately affecting the objective assessment and precise judgment of the credibility of the password input process. To address these problems, this invention proposes a solution. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a personal password credibility assessment method that integrates biometrics and rule engines. By jointly analyzing the temporal changes of biometrics with rule-based decision logic, this method addresses the problem that the prior art fails to correlate and model the changes in password structure with the natural fluctuations of biometrics during the input process, leading to a decrease in the accuracy of the credibility assessment of the password input process.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A method for assessing the credibility of a personal password by integrating biometrics and a rule engine includes the following steps: acquiring the user's biometric data and password input data, and parsing the password input data to obtain password structure switching points; isolating and identifying the biometric data based on the password structure switching points to obtain a first isolation zone; extracting biometric change feature sequences from the first isolation zone, and recombining the judgment rules in a preset rule engine based on the biometric change feature sequences to obtain a password structure rule group; assessing the credibility of the personal password based on the password structure rule group, and outputting the password credibility assessment result.

[0007] In a preferred embodiment, the step of acquiring the user's biometric data and password input data, and parsing the password input data to obtain password structure switching points, specifically involves: collecting the user's continuous biometric signal stream during password input to form biometric data; synchronously recording the key sequence and timing information during password input to form password input data; parsing the key sequence and constructing a directed graph of password character state transitions; identifying the location where each password character changes based on the directed graph of password character state transitions and marking it as a first candidate switching point; combining the timing information to calculate the time interval between adjacent input operations and identifying the input operation positions where the time interval is greater than a preset first threshold to obtain a second candidate switching point; performing a union of the first and second candidate switching points and topologically sorting the merged set according to the input order to obtain the password structure switching point.

[0008] In a preferred embodiment, the step of parsing the key sequence and constructing a directed graph of state transitions for the password characters specifically involves: parsing the key sequence to obtain character category labels; constructing transition edges for character categories based on temporal information and counting the transition frequencies between adjacent character categories to form a transition frequency matrix; and constructing a directed graph of state transitions based on the transition frequency matrix, wherein the nodes of the directed graph of state transitions are character categories and the edge weights are transition frequencies.

[0009] In a preferred embodiment, the isolation and identification of biometric data based on cryptographic structure switching points to obtain a first isolation region specifically involves: expanding a buffer of a preset time length forward and backward to both sides of each cryptographic structure switching point to obtain several biometric segments; performing multi-scale segmentation on the biometric signal sequence within each biometric segment to obtain several biometric signal sub-segments; extracting finger movement response parameters within each biometric signal sub-segment and performing feature fusion on the finger movement response parameters to obtain a biometric feature vector within each biometric signal sub-segment; performing unsupervised clustering on the biometric feature vectors within all biometric signal sub-segments to obtain several biometric response clusters; and selecting regions that overlap with the cryptographic structure switching points in time and whose number of samples within the biometric response clusters exceeds a preset second threshold to obtain the first isolation region.

[0010] In a preferred embodiment, the step of extracting the biological change feature sequence from the first isolation zone specifically involves: performing differential processing on the biological feature data in the first isolation zone along the time axis to obtain the pressure change gradient and rhythm interval derivative; identifying positive and negative jump points exceeding a preset amplitude from the pressure change gradient and recording the occurrence time and duration of the jump points; extracting monotonic segments from the rhythm interval derivative and calculating the time span and slope mean of the monotonic segments; and fusing the occurrence time and duration of the jump points, the time span and slope mean of the monotonic segments to obtain the biological change feature sequence.

[0011] In a preferred embodiment, the step of recombining the judgment rules in the preset rule engine based on the biological change feature sequence to obtain a cryptographic structure rule group specifically involves: obtaining a judgment rule library in the preset rule engine, the judgment rule library containing rule condition statements and execution actions; mapping the biological change feature sequence to the variable space in the rule condition statements to obtain several rule subsets; performing dependency analysis on the several rule subsets and constructing a rule dependency graph; reordering and merging the nodes in the rule dependency graph based on the biological change feature sequence to obtain a recombined rule set; and grouping the recombined rule set according to preset execution priority and logical correlation to obtain a cryptographic structure rule group.

[0012] In a preferred embodiment, the step of performing dependency analysis on several rule subsets and constructing a rule dependency graph specifically involves: extracting the condition variables and action variables of each rule in the rule subsets and constructing a variable association matrix; identifying the input-output dependencies between rules based on the variable association matrix to obtain rule dependency chains; performing loop detection on the rule dependency chains, and if loops exist, performing loop decomposition and recombination to obtain acyclic dependency chains; topologically sorting the acyclic dependency chains according to the execution order to obtain a rule execution sequence; and inserting preset dynamic checkpoints into the rule execution sequence based on biological change feature sequences to obtain a rule dependency graph.

[0013] In a preferred embodiment, the step of evaluating the credibility of a personal password based on a set of cryptographic structure rules and outputting the password credibility evaluation result specifically involves: inputting password input data and biometric data into the set of cryptographic structure rules, executing rule condition judgments one by one; recording the input data fragment and biometric fragment when each rule is triggered to form a set of triggering events; performing spatiotemporal clustering analysis on the set of triggering events to obtain event-dense regions; extracting the rule triggering frequency of the event-dense regions, and calculating the password credibility score based on the rule triggering frequency.

[0014] The technical effects and advantages of the personal password credibility assessment method integrating biometrics and rule engine of this invention are as follows:

[0015] This invention introduces collaborative modeling of cryptographic structure and biometric temporal behavior during password input. It sequentially performs structured parsing of the key sequence to identify character categories and locate cryptographic structure switching points. Based on this, using structure switching as anchor points, it delineates key time segments directly corresponding to the cryptographic structure from continuous biometric data. Furthermore, it extracts biometric change sequences reflecting input behavior changes from these segments. These biometric change sequences are then matched and judged against preset cryptographic structure rules, thereby forming a comprehensive credibility assessment of the password input process. Through these steps, this invention establishes a clear correspondence between internal cryptographic logic factors such as character type changes and structure switching and the natural fluctuations in biometrics generated during input. This allows for accurate identification and reasonable interpretation of local behavioral changes caused by cryptographic structure complexity, preventing them from interfering with the overall judgment result. Therefore, it effectively improves the precision, stability, and accuracy of the credibility assessment of the password input process. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a personal password credibility assessment method that integrates biometrics and a rule engine, according to the present invention.

[0017] Figure 2 This is a schematic diagram of the structure of a personal password credibility assessment system that integrates biometrics and a rule engine according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1, Figure 1This invention presents a method for evaluating the credibility of personal passwords that integrates biometrics and a rule engine, comprising the following steps:

[0020] S1, acquire the user's biometric data and password input data, and parse the password input data to obtain the password structure switching point;

[0021] In this embodiment, the user's biometric data and password input data are acquired, and the password input data is parsed to obtain the password structure switching point, specifically:

[0022] Collect the continuous biosignal stream of the user during the password input process to form biometric data;

[0023] The key sequence and timing information during the user's password input process are recorded synchronously to form password input data;

[0024] Perform syntax parsing on the key sequence and construct a directed graph of the state transitions of the password characters;

[0025] The directed graph of state transitions based on password characters identifies the location where each password character changes and marks it as the first candidate switching point;

[0026] By combining timing information, the time interval between adjacent keying operations is calculated and the keying operation position where the time interval is greater than a preset first threshold is identified to obtain the second candidate switching point;

[0027] The first candidate switching point and the second candidate switching point are merged, and the merged point set is topologically sorted according to the input order to obtain the cryptographic structure switching point.

[0028] It should be noted that during password input, the user's biometric data mainly includes continuous signals related to finger movements and input behavior. For example, the pressure of the fingertip on the keys, the sliding force on the touchpad, the speed at which the finger moves on the keyboard, the duration of the touch, and minute vibrations or tremors of the finger. This data is recorded as a continuous signal stream, with each data point containing a timestamp and a corresponding numerical value, reflecting the user's physiological characteristics during input.

[0029] Password input data consists of the actual sequence of password characters entered by the user, along with the order in which each character was entered. For example, if a user enters the password "Ab12#C", the password input data records the key code for each character (A, b, 1, 2, #, C) and the specific time each key was pressed. In this way, the password input data not only knows which keys the user pressed, but also the order in which those keys were pressed and released.

[0030] For example, suppose the time sequence of a user entering a password is as follows: User 1 presses: 0.00 seconds; User 1 releases: 0.12 seconds; User 2 presses: 0.15 seconds; User 2 releases: 0.27 seconds; User 3 presses: 0.35 seconds; User 3 releases: 0.47 seconds. This time sequence information, combined with biometric data, constitutes a complete record of input and biometric behavior.

[0031] Secondly, the password character sequence is parsed, and each character is classified according to its type, such as uppercase letters, lowercase letters, numbers, and special symbols. Then, the adjacent transition relationships of each type of character are statistically analyzed according to the input order, and the frequency of transition from one type of character to another is recorded. For example, in the password "Ab12#C", the transition from uppercase A to lowercase b is an "uppercase -> lowercase" transition, and the transition from b to 1 is a "lowercase -> number" transition. Next, these transition relationships are constructed into a directed graph of state transitions. Each node in the graph represents a character type (such as uppercase, lowercase, number, special symbol), and the edges represent transitions from one character type to the next, along with the number of transitions or weights. By analyzing the path of the input sequence in the graph, the locations where character types change can be found. For example, if the user inputs two consecutive numbers "12", there is no type change between these two numbers, and the graph path remains within the number node; however, the transition from b to 1 crosses "lowercase -> number", so this location is marked as the first candidate switching point. In addition to variations in character types, the time interval between user input of a password is also crucial information about the password structure. An unusually long interval between two keystrokes usually indicates that the user is thinking or pausing input, which is a potential point of change in the password's logical structure.

[0032] Finally, the first and second candidate switching points are identified based on changes in character type and time interval, respectively. These may or may not overlap. The union operation combines the two sets of points into a single set, removing duplicates to obtain a complete set of potential structural change locations. For example, if the first candidate switching points are A->b, b->1, 2->#, and #->C, and the second candidate switching points are b->1 and 2->#, the union result is A->b, b->1, 2->#, and #->C. The merged set is then sorted according to the input order, i.e., the order in which the user entered the password. This is the core idea of ​​topological sorting: ensuring that all switching points are arranged sequentially in time, without any conflicts. For example, the sorted set is still: A->b, b->1, 2->#, and #->C. The final password structure switching points are these key nodes marked during the input process that may reflect changes in the password's logical structure. They are used in subsequent analysis to locate areas of biometric variation, helping to assess whether the password input conforms to user habits, thereby assisting in determining password credibility.

[0033] In this embodiment, the key sequence is parsed, and a directed graph of the state transitions of the password characters is constructed, specifically as follows:

[0034] Perform syntax parsing on the key sequence to obtain character category labels;

[0035] Transition edges for character categories are constructed based on temporal information, and the transition frequencies between adjacent character categories are counted to form a transition frequency matrix;

[0036] A directed graph of state transitions is constructed based on the transition frequency matrix. The nodes of the directed graph of state transitions are character categories, and the edge weights are the transition frequencies.

[0037] It's important to note that during password input, the key sequence first undergoes syntactic parsing. Each specific keystroke is transformed into a more abstract character category label to facilitate password structure analysis. For example, if a user enters the password "Aa1#B3", each character is parsed into a corresponding category label: uppercase letters A and B are labeled "Uppercase", lowercase letter a is labeled "Lowercase", numbers 1 and 3 are labeled "Digit", and the special symbol # is labeled "Symbol". After parsing, the entire password sequence is transformed into a category sequence of "Uppercase->Lowercase->Digit->Symbol->Uppercase->Digit". These category labels reflect the type attributes of the characters, rather than the specific characters themselves, allowing subsequent analysis to focus on the input logic rather than the specific values.

[0038] Next, based on the temporal order of the input, we need to construct transition edges between character categories and count the frequency of transitions between adjacent categories. For example, in the category sequence above, the transition from Uppercase to Lowercase occurs once, from Lowercase to Digit once, from Digit to Symbol once, from Symbol to Uppercase once, and from Uppercase to Digit once. These statistics form a transition frequency matrix, where each row represents the current category, each column represents the next category, and the value of each cell in the matrix represents the number of times the transition occurs from the current category to the next category. This allows us to see which categories transition more frequently and which transition less frequently, thus quantifying the structural changes in the password input.

[0039] Finally, using this transition frequency matrix, a directed graph of state transitions can be constructed, with each character category as a node in the graph. Directed edges represent the transitions between categories, and the weight of the edge corresponds to the frequency of the transition. For example, the edge weight from Uppercase node to Lowercase node is 1, the edge weight from Uppercase to Digit is 1, the edge weight from Digit to Symbol is 1, and the edge weight from Symbol to Uppercase is 1. This directed graph intuitively depicts the transformation relationships between different character categories during password input.

[0040] S2, based on the cryptographic structure switching point, biometric data is isolated and identified to obtain the first isolation zone;

[0041] In this embodiment, biometric data is isolated and identified based on cryptographic structure switching points to obtain a first isolation zone, specifically:

[0042] Centered on each cryptographic structure switching point, a buffer of a preset time length is extended to the front and back sides to obtain several biological segments;

[0043] The biological signal sequence within each biological segment is segmented at multiple scales to obtain several biological signal sub-segments;

[0044] Finger movement response parameters are extracted from each biological signal sub-segment, and feature fusion is performed on the finger movement response parameters to obtain the biological feature vector of each biological signal sub-segment;

[0045] Unsupervised clustering of biological feature vectors within all biological signal sub-segments yields several biological response clusters;

[0046] The first isolation zone is obtained by filtering out regions that overlap with the cryptographic structure switching point in time and whose number of samples in the biological response cluster exceeds a preset second threshold.

[0047] It's important to note that during password input, each password structure switching point signifies a potential change in input logic. To capture related biometric features, a preset time length, such as 0.3 seconds, needs to be extended forward and backward from this switching point to create a time buffer covering the changes in actions before and after. Each switching point corresponds to a biometric segment. For example, when a user inputs "Aa1#B3", the switching point from b to 1 occurs at 0.27 seconds. The biometric segment at this switching point will cover the finger pressure, swipe speed, and touch area data from 0.3 seconds before to 0.3 seconds after 0.27 seconds, forming a complete fragment reflecting the characteristics of actions before and after the switch. Similar biometric segments are generated at each switching point throughout the password input process, providing a foundation for subsequent multi-scale analysis.

[0048] Secondly, each biometric segment contains continuous finger movement signals. To capture both subtle changes in movement and analyze overall trends, the biometric segment needs to be segmented at multiple scales. For example, a 0.6-second biometric segment can be divided into several sub-segments of three different lengths: 0.1 seconds, 0.2 seconds, and 0.3 seconds. This ensures that the movement characteristics within each sub-segment are relatively stable while preserving dynamic changes at different time scales. In this way, if the finger pressure changes rapidly within a certain period while the movement is relatively stable within another period, multi-scale segmentation can capture both types of features simultaneously. Each sub-segment becomes the basic unit of analysis for subsequent feature extraction, thus providing a more refined reflection of the user's movement characteristics before and after the password structure switching point.

[0049] Furthermore, within each biosignal sub-segment, parameters describing the user's finger operation characteristics need to be extracted. These include pressure intensity (the magnitude and trend of force applied to the fingertip when pressing); key duration (the time interval between pressing and releasing); touch area (the change in the size of the finger's contact surface); sliding speed or displacement (reflecting the finger's movement on the key or touchpad); and minute jitter or vibration characteristics (reflecting the stability and subtle differences in finger movements during input). For example, within a 0.2-second sub-segment, the user's key pressure increases from 0.5 Newtons to 0.8 Newtons over a duration of 0.12 seconds, with a sliding displacement of 0.3 centimeters, accompanied by slight jitter. These parameters collectively depict the action characteristics of this sub-segment, providing fundamental information for subsequent biosignal fusion.

[0050] Furthermore, the extracted finger movement response parameters need to be integrated into a unified representation, called a biometric vector. This allows each sub-segment to be described by a vector representing its comprehensive movement characteristics. For example, by sequentially integrating pressure changes, duration, touch area, swipe speed, and jitter parameters into a single vector, the feature vector for the 0.2-second sub-segment might be [0.8 Newtons, 0.12 seconds, 1.2 square centimeters, 0.3 centimeters / second, 0.02 Newtons vibration amplitude]. This vector can simultaneously reflect finger force, speed, contact area, and stability, enabling direct comparison, clustering, or subsequent analysis of the movement characteristics of different sub-segments.

[0051] Next, after collecting the feature vectors of all segments throughout the entire password input process, unsupervised clustering methods are used to group vectors with similar features into one category, forming biological response clusters. For example, in a certain password input, the feature vectors of 20 segments, after clustering, are divided into three clusters: cluster A reflects the action pattern of high-pressure, rapid input; cluster B reflects the action pattern of medium-pressure, stable input; and cluster C reflects the action pattern of low-pressure, slow operation. These clusters correspond to the typical action characteristics of users at different input stages. Through clustering, the massive segment data can be simplified into several representative action patterns, providing a basis for locating key biological change areas.

[0052] Finally, after obtaining the biometric response clusters, it is necessary to select the key regions that are truly related to the cryptographic structure switching points. This is done by filtering out clusters that overlap with the switching points in time and whose sample count exceeds a preset threshold. For example, within a biometric segment corresponding to a switching point, cluster A contains 18 sub-segment samples, and the threshold is set to 15. The time period containing this cluster is then preserved, forming the first isolation zone. The first isolation zone can be understood as the time interval during password input that is closely related to the cryptographic structure switching and contains significant changes in finger movements. It is a key area for subsequently extracting features from biometric changes and evaluating password credibility, ensuring that the analysis focuses on action segments highly correlated with changes in user input logic, while ignoring irrelevant or noisy actions.

[0053] S3, extract biological change feature sequences from the first isolation zone, and reorganize the judgment rules in the preset rule engine based on the biological change feature sequences to obtain a cryptographic structure rule group;

[0054] In this embodiment, biological change feature sequences are extracted within the first isolation zone, specifically as follows:

[0055] Differential processing was performed on the biometric data in the first isolation zone along the time axis to obtain the pressure change gradient and rhythm interval derivative;

[0056] Identify positive and negative jump points exceeding a preset amplitude from the pressure change gradient, and record the time and duration of the jump points.

[0057] Extract monotonic segments from the derivative of rhythm intervals and calculate the time span and mean slope of the monotonic segments;

[0058] By fusing the occurrence time and duration of the transition point, the time span of the monotonic segment, and the mean slope, a biological change feature sequence is obtained.

[0059] It's important to note that in the first isolation zone, each biometric segment contains continuous finger movement feature data. To analyze the dynamic changes during password input, this continuous data needs to be differentially processed along the time axis, which means calculating the change between each time point and the previous time point. For example, if the finger pressure is 0.5 Newtons at 0.05 seconds and increases to 0.65 Newtons at 0.07 seconds, the differential result is 0.15 Newtons. This continuous pressure change forms a pressure change gradient, reflecting the trend of changes in the speed and force of the finger pressing or releasing action. The rhythm interval derivative is the differential of the change in the time interval between each key press. For example, if the time interval between consecutive key presses changes from 0.12 seconds to 0.08 seconds, the derivative is -0.04 seconds, reflecting the trend of the input rhythm accelerating or slowing down over time. Through the pressure change gradient and the rhythm interval derivative, the fluctuations in the force of finger movements and the changes in input rhythm can be quantified, thereby capturing the subtle movement features of the user near the password structure switching points.

[0060] Secondly, the pressure change gradient obtained through differential processing can reflect the rapid increase or decrease in pressure. To identify significant changes in action, it is necessary to identify transition points exceeding a preset amplitude. For example, if the preset threshold is 0.1 Newtons, then if the pressure suddenly increases from 0.5 Newtons to 0.65 Newtons at a certain time (gradient of 0.15 Newtons), this point is marked as a positive transition point; conversely, if the pressure drops sharply from 0.65 Newtons to 0.5 Newtons, it is marked as a negative transition point. Each transition point records its occurrence time, such as 0.07 seconds, and its duration, such as from 0.07 seconds to 0.09 seconds lasting 0.02 seconds. The entire transition process is also recorded, which can accurately describe the rapid fluctuations in the user's actions near the password structure switching point.

[0061] Furthermore, the derivative of the rhythm interval describes the trend of continuous key press interval changes. To analyze the overall pattern of user input rhythm, it is necessary to extract monotonic segments from the derivative sequence, that is, time periods where the direction of continuous change remains consistent. For example, if the user's key press interval gradually decreases from 0.12 seconds to 0.08 seconds, and then gradually increases to 0.15 seconds, then the first half (0.12->0.08 seconds) is a decreasing monotonic segment, and the second half (0.08->0.15 seconds) is an increasing monotonic segment. For each monotonic segment, its time span needs to be recorded, such as a decreasing segment lasting 0.1 seconds and an increasing segment lasting 0.12 seconds. Simultaneously, the average slope is calculated, which is the average speed of rhythm change within that monotonic segment. For example, the average slope of the decreasing segment is -0.04 seconds per step, and the average slope of the increasing segment is 0.06 seconds per step. These indicators can quantify the stability and trend of user input rhythm, providing important information for subsequent feature fusion.

[0062] Finally, the features of pressure gradient transition points and rhythmic interval monotonic segments need to be integrated into a unified sequence to form a biological change feature sequence. For example, the occurrence time (0.07 seconds) and duration (0.02 seconds) of each transition point are integrated with the corresponding rhythmic monotonic segment duration (0.1 seconds) and mean slope (-0.04 seconds per step) to obtain a comprehensive descriptive vector containing changes in action intensity and rhythm. By similarly integrating all transition points and monotonic segments in the entire first isolation zone, a complete biological change feature sequence is formed. This sequence not only reflects the user's pressure fluctuations at cryptographic structure switching points but also quantifies the acceleration or deceleration of the input rhythm, providing a precise dynamic feature basis for subsequent reorganization rules and cryptographic credibility assessment.

[0063] In this embodiment, the judgment rules in the preset rule engine are reorganized based on the biological change feature sequence to obtain a cryptographic structure rule group, specifically:

[0064] Obtain the judgment rule library from the preset rule engine, the judgment rule library containing rule condition statements and execution actions;

[0065] By mapping the biological change feature sequence to the variable space in the rule condition statement, several rule subsets are obtained;

[0066] Perform dependency analysis on several subsets of rules and construct a rule dependency graph;

[0067] Based on the biological change feature sequence, the nodes in the rule dependency graph are reordered and merged to obtain a recombined rule set;

[0068] The reorganization rule set is grouped according to the preset execution priority and logical relevance to obtain the cryptographic structure rule group.

[0069] It's important to note that in password credibility assessment, the default rule engine's judgment rule base is a predefined set of rules used to logically judge user input actions and biometric features. Each rule consists of two parts: a conditional statement and an execution action. The conditional statement describes the prerequisites for triggering the rule. For example, "pressure gradient exceeding 0.1 Newtons and rhythm interval derivative less than -0.05 seconds" indicates that the user pressed keys rapidly within a short period and the input rhythm accelerated. The variables in these conditional statements correspond to specific indicators in the biometric feature sequence. The execution action is the judgment or processing method to be taken when the condition is met, such as "increasing the credibility score of this structure point," "triggering further verification," or "recording abnormal behavior." For example, in one rule, if a pressure jump point is detected near a password switching point for a duration exceeding 0.02 seconds and the average slope of the monotonic segment is positive, the execution action might be to mark that segment as a high-credibility pattern; another rule might be that the pressure jump amplitude is too large and the rhythm slows down, in which case the execution action is to decrease the credibility score. By acquiring the entire rule base, the assessment method can perform logical analysis of the password structure based on different biometric patterns.

[0070] Secondly, the mapping of the biological change feature sequences extracted from the first isolation zone to the judgment rule base is accomplished by mapping the specific values ​​in the feature sequences to the variable space in the rule condition statements. For example, if the rule conditions include "pressure jump amplitude > X" and "rhythm monotonic segment slope > Y", then the amplitude of each jump point and the slope of the monotonic segment in the biological change feature sequence are used as the actual values ​​of X and Y to determine whether the conditions are met. After mapping, rules that satisfy the same type of conditions can be aggregated together to form several rule subsets. For example, in a password input, if the amplitudes of three jump points are all greater than the threshold and the slopes of two monotonic segments are positive, then 10 rules in the corresponding rule base are triggered. Among them, 5 rule conditions completely match the current feature sequence, and these 5 rules form a subset. Another set of rules that meet different features form a second subset, thus decomposing the entire rule base into targeted subsets based on the actual biological change feature sequences.

[0071] Furthermore, logical dependencies exist between rule subsets; for example, the execution of some rules may depend on the judgment results of other rules. To ensure proper execution, a rule dependency graph needs to be constructed, with each rule as a node and the input-output dependencies between rules represented by directed edges. Then, by combining the current biological change feature sequence, the nodes in the graph can be reordered and merged. For example, if two rules detect stress jumps and rhythm changes respectively, but the stress jump amplitude is significantly greater than the rhythm change threshold under the current feature sequence, the two rules can be merged into a single comprehensive rule. The node order is arranged from largest to smallest influence, prioritizing the evaluation of the rule most relevant to the current input. This results in a reorganized rule set that retains the logic of the original rule base while being optimized based on actual biological characteristics, improving evaluation efficiency and accuracy.

[0072] Finally, after obtaining the recombined rule set, the rules need to be further grouped according to execution priority and logical relevance. This ensures that each group of rules maintains logical coherence while facilitating group-by-group execution. Execution priority can be based on the importance of the rules or their impact on password credibility; for example, rules with the largest pressure jumps are executed first. Logical relevance reflects the dependencies between rules; for example, rules with continuous jumps and monotonic segments are grouped together for execution. Ultimately, each group of rules forms a logical unit. These combinations constitute a cryptographic structure rule group, which can be executed group-by-group in subsequent credibility assessments of user-input passwords. This considers both the intensity of actions and rhythmic variations to comprehensively judge whether the input conforms to user habits. For example, in a single input, the first group of rules might assess pressure jump patterns, the second group assesses rhythmic stability, and the third group assesses the combined pattern of jumps and rhythm, forming a complete rule logic framework for final scoring.

[0073] In this embodiment, dependency analysis is performed on several subsets of rules, and a rule dependency graph is constructed, specifically as follows:

[0074] Extract the condition variables and action variables of each rule in the rule subset and construct a variable association matrix;

[0075] The input-output dependencies between rules are identified based on the variable correlation matrix, thus obtaining the rule dependency chain;

[0076] Perform loop detection on the rule-dependent chain. If a loop exists, decompose and reorganize the loop to obtain a loop-free dependency chain.

[0077] The acyclic dependency chains are topologically sorted according to their execution order to obtain the rule execution sequence;

[0078] Based on biological change feature sequences, preset dynamic checkpoints are inserted into the rule execution sequence to obtain a rule dependency graph.

[0079] It's important to note that within the rule subset, each rule consists of a conditional statement and an action. The metrics involved in the conditional statement are the conditional variables. For example, the pressure jump amplitude in "pressure jump amplitude exceeds 0.1 Newtons" is a conditional variable, and the slope in "the slope of the monotonic rhythm segment is positive" is another conditional variable. The action, which affects the results of other rules or scores, is called the action variable, such as "increase confidence score" or "trigger further verification." By extracting the conditional and action variables of each rule in the rule subset, a variable association matrix can be constructed. The rows of the matrix represent the conditional variables of the rules, and the columns represent the action variables. If the conditional variable of one rule has a dependency on the action variable of another rule, then the association is marked in the matrix. For example, if the conditional variable of rule R1 is the pressure jump amplitude and the action variable is marking high confidence, while the conditional variable of rule R2 is the high confidence mark, then there is a dependency between R1 and R2 in the matrix. This matrix clearly describes the variable connections and potential dependencies between rules, providing a foundation for constructing a rule dependency graph later.

[0080] Secondly, using the variable correlation matrix obtained in the previous step, we can analyze which rule outputs serve as inputs for other rules, thus forming rule dependencies. For example, if the action variable of rule R1 is "marking high confidence," and the condition variable of rule R2 needs to reference the "high confidence mark," then we can determine that R2 depends on the execution result of R1. By connecting rules according to their dependencies in this way, a rule dependency chain is formed, such as R1->R2->R3. R3 may depend on the action output of R2, while R2 depends on the action output of R1. This dependency chain clarifies the sequential logic of rule execution, ensuring that when evaluating user biometrics, the impact of key actions is calculated first, followed by subsequent related judgments.

[0081] Furthermore, during the construction of rule dependency chains, loops may occur, meaning that rules form circular dependencies. For example, R1 depends on the action output of R2, while R2 also depends on the action output of R1. This situation leads to ambiguous execution logic. To solve this problem, loop detection is needed for the dependency chains. Once a loop is detected, the rules in the loop are decomposed and reorganized. For example, the rules in the loop can be split into independent logical units or conditional priorities can be introduced to adjust the order of the originally circularly dependent rules to linear dependencies. Ultimately, an acyclic dependency chain is obtained, ensuring that the execution of each rule can explicitly depend on the results of previous rules without creating an infinite loop.

[0082] Furthermore, for acyclic dependency chains that have been eliminated, topological sorting can be performed based on dependencies, placing each rule in the earliest position that satisfies its input dependencies, thus forming a rule execution sequence. For example, if the dependency chain is R1->R2->R3, the execution sequence after topological sorting is to execute R1 first, then R2, and finally R3. Each rule can be computed after its dependent antecedent rules have been executed. Topological sorting not only ensures the logical consistency of rule execution but also avoids evaluation errors caused by disordered dependency order, enabling the rule evaluation process to proceed according to the actual sequence of changes in biometric characteristics.

[0083] Finally, after obtaining the rule execution sequence, dynamic checkpoints can be inserted at key time points or nodes with significant feature changes, in conjunction with the biometric change feature sequence. These checkpoints indicate moments requiring focused attention or reassessment during rule execution. For example, during password input, if a user experiences a significant shift in pressure and a faster pace at a certain switching point, a dynamic checkpoint can be inserted before the relevant rules in the execution sequence. This allows the rules to re-evaluate the features or trigger specific actions at that point, improving the sensitivity of the assessment. Ultimately, the entire rule execution sequence, along with the dynamic checkpoints, forms a complete rule dependency graph. This graph not only demonstrates the logical dependencies between rules but also identifies key checkpoints when critical changes occur in biometric features. This enables dynamic responses to changes in user actions during the assessment process, improving the accuracy and robustness of password credibility analysis.

[0084] S4 evaluates the credibility of individual passwords based on the set of cryptographic structure rules and outputs the password credibility evaluation results.

[0085] In this embodiment, the credibility of an individual's password is evaluated based on a set of cryptographic structure rules, and the password credibility evaluation result is output, specifically as follows:

[0086] Input the password data and biometric data into the password structure rule group, and execute the rule condition judgment one by one;

[0087] Record the input data fragments and biometric fragments when each rule is triggered to form a set of trigger events;

[0088] Spatiotemporal clustering analysis is performed on the set of triggering events to obtain event-dense regions;

[0089] Extract the rule trigger frequency of event-intensive areas and calculate the password credibility score based on the rule trigger frequency.

[0090] It's important to note that when executing the password structure rule set, the user's password input data (e.g., key sequence, key timing, and structure transition location) is first aligned with the biometric data extracted within the corresponding time period (e.g., pressure change features, rhythm change features, and finger movement feature vectors) along the time axis and then fed into the rule set. Then, following a pre-determined execution order within the rule set, each rule is evaluated sequentially. For example, if a rule's condition is "a significant positive pressure jump and a shortened rhythm occur when switching from letters to numbers," then the corresponding pressure change features and rhythm features are read from the current time segment and matched against the rule's condition. If the condition is met, the rule is triggered; otherwise, the next rule is executed. This step-by-step evaluation ensures that each rule is independently evaluated based on actual password input behavior and corresponding biometric features, thus comprehensively covering various structural change scenarios during password input.

[0091] Secondly, when a rule is triggered during execution, the corresponding input data segment and biometric feature segment are immediately recorded. This includes the key sequence interval, start and end times, pressure gradient, rhythm trend, and finger movement feature vectors within that time period. This information collectively constitutes a complete trigger event. As the rule group continues to execute, different rules may be triggered multiple times at different points in time. For example, similar pressure jumps and rhythm changes may occur at multiple cryptographic structure switching points. Each trigger generates an independent trigger event. All trigger events are aggregated in chronological order to form a trigger event set. This set retains both the rule trigger type information and the corresponding temporal sequence and biometric context, providing the initial basis for subsequent overall analysis.

[0092] Furthermore, after obtaining the set of triggering events, these events can be jointly analyzed along the time and input location dimensions. Triggering events occurring close in time and with similar input structures are grouped together. For example, if multiple rules are triggered consecutively within a short period before and after the same password structure switch point, a high-density region will form on the timeline. Through this spatiotemporal clustering analysis, event-dense areas can be identified. These areas typically correspond to locations where users exhibit significant changes in their biological responses when entering passwords. For instance, when switching from letter segments to special character segments, multiple rules related to pressure jumps and rhythm changes are triggered within a short period; this time window is identified as an event-dense area. Event-dense areas reflect the concentration of rule triggering and are an important basis for assessing whether password input behavior conforms to stable user habits.

[0093] Finally, after identifying event-intensive areas, the frequency of triggering various rules within these areas is statistically analyzed. For example, in a certain event-intensive area, 10 rules are triggered, with rules related to pressure changes triggered 6 times and rules related to rhythm changes triggered 4 times. These trigger frequencies reflect the stability and consistency of the user's input behavior within these areas. Subsequently, based on the importance of different rules in the cipher structure analysis, the trigger frequencies are comprehensively evaluated to form the final cipher credibility score. For instance, when high-weight rules are consistently triggered in multiple event-intensive areas, the score will increase accordingly, indicating that the cipher input behavior is highly consistent with the user's historical input habits. Conversely, if the number of triggered rules is small and scattered, or if only low-weight rules are triggered, the score will decrease. For example, in a complete cipher input process, if obvious event-intensive areas are formed at the three key structural switching points, and the main rules are triggered multiple times in each area, the final cipher credibility score will be in a higher range, indicating that the input behavior has high credibility.

[0094] Figure 2 This invention presents a personal password credibility assessment system that integrates biometrics and a rule engine, comprising a data analysis module, an isolation and identification module, a rule recombination module, and a password assessment module.

[0095] The data analysis module is used to acquire the user's biometric data and password input data, and to parse the password input data to obtain the password structure switching points;

[0096] The isolation and identification module is used to isolate and identify biometric data based on the cryptographic structure switching point to obtain the first isolation zone;

[0097] The rule recombination module is used to extract biological change feature sequences from the first isolation zone and recombine the judgment rules in the preset rule engine based on the biological change feature sequences to obtain a cryptographic structure rule group.

[0098] The password evaluation module is used to evaluate the credibility of an individual's password based on a set of password structure rules and output the password credibility evaluation result.

[0099] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0100] Those skilled in the art will recognize that the modules and algorithm steps of the various examples 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 implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0101] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0102] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0103] In conclusion, 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 within the protection scope of the present invention.

Claims

1. A method for evaluating the credibility of personal passwords by integrating biometrics and rule engines, characterized in that, Includes the following steps: The system acquires the user's biometric data and password input data, and performs password logic structure parsing on the password input data to obtain password structure switching points. The first isolation zone is obtained by isolating and identifying biometric data based on the cryptographic structure switching point. Biological change feature sequences are extracted from the first isolation zone, and the judgment rules in the preset rule engine are reorganized based on the biological change feature sequences to obtain a cryptographic structure rule group; The credibility of personal passwords is evaluated based on a set of cryptographic structure rules, and the password credibility evaluation result is output.

2. The personal password credibility assessment method integrating biometrics and rule engine according to claim 1, characterized in that, The process of acquiring the user's biometric data and password input data, and parsing the password input data to obtain the password structure switching point, specifically involves: Collect the continuous biosignal stream of the user during the password input process to form biometric data; The key sequence and timing information during the user's password input process are recorded synchronously to form password input data; Perform syntax parsing on the key sequence and construct a directed graph of the state transitions of the password characters; The directed graph of state transitions based on password characters identifies the location where each password character changes and marks it as the first candidate switching point; By combining timing information, the time interval between adjacent keying operations is calculated and the keying operation position where the time interval is greater than a preset first threshold is identified to obtain the second candidate switching point; The first candidate switching point and the second candidate switching point are merged, and the merged point set is topologically sorted according to the input order to obtain the cryptographic structure switching point.

3. The personal password credibility assessment method integrating biometrics and rule engine according to claim 2, characterized in that, The specific steps for parsing the key sequence and constructing a directed graph of the state transitions of the password characters are as follows: Perform syntax parsing on the key sequence to obtain character category labels; Transition edges for character categories are constructed based on temporal information, and the transition frequencies between adjacent character categories are counted to form a transition frequency matrix; A directed graph of state transitions is constructed based on the transition frequency matrix. The nodes of the directed graph of state transitions are character categories, and the edge weights are the transition frequencies.

4. The personal password credibility assessment method integrating biometrics and rule engine according to claim 3, characterized in that, The method of isolating and identifying biometric data based on cryptographic structure switching points to obtain the first isolation zone is as follows: Centered on each cryptographic structure switching point, a buffer of a preset time length is extended to the front and back sides to obtain several biological segments; The biological signal sequence within each biological segment is segmented at multiple scales to obtain several biological signal sub-segments; Finger movement response parameters are extracted from each biological signal sub-segment, and feature fusion is performed on the finger movement response parameters to obtain the biological feature vector of each biological signal sub-segment; Unsupervised clustering of biological feature vectors within all biological signal sub-segments yields several biological response clusters; The first isolation zone is obtained by filtering out regions that overlap with the cryptographic structure switching point in time and whose number of samples in the biological response cluster exceeds a preset second threshold.

5. The personal password credibility assessment method integrating biometrics and rule engine according to claim 4, characterized in that, The extraction of biological change feature sequences from the first isolation zone specifically involves: Differential processing was performed on the biometric data in the first isolation zone along the time axis to obtain the pressure change gradient and rhythm interval derivative; Identify positive and negative jump points exceeding a preset amplitude from the pressure change gradient, and record the time and duration of the jump points. Extract monotonic segments from the derivative of rhythm intervals and calculate the time span and mean slope of the monotonic segments; By fusing the occurrence time and duration of the jump point, the time span of the monotonic segment, and the mean slope, a biological change feature sequence is obtained.

6. The personal password credibility assessment method integrating biometrics and rule engine according to claim 5, characterized in that, The process of reorganizing the judgment rules in the preset rule engine based on biological change feature sequences to obtain a cryptographic structure rule group is as follows: Obtain the judgment rule library from the preset rule engine, the judgment rule library containing rule condition statements and execution actions; By mapping the biological change feature sequence to the variable space in the rule condition statement, several rule subsets are obtained; Perform dependency analysis on several subsets of rules and construct a rule dependency graph; Based on the biological change feature sequence, the nodes in the rule dependency graph are reordered and merged to obtain a recombined rule set; The reorganization rule set is grouped according to the preset execution priority and logical relevance to obtain the cryptographic structure rule group.

7. The personal password credibility assessment method integrating biometrics and rule engine according to claim 6, characterized in that, The process of performing dependency analysis on several subsets of rules and constructing a rule dependency graph specifically involves: Extract the condition variables and action variables of each rule in the rule subset and construct a variable association matrix; The input-output dependencies between rules are identified based on the variable correlation matrix, thus obtaining the rule dependency chain; Perform loop detection on the rule-dependent chain. If a loop exists, decompose and reorganize the loop to obtain a loop-free dependency chain. The acyclic dependency chains are topologically sorted according to their execution order to obtain the rule execution sequence; Based on biological change feature sequences, preset dynamic checkpoints are inserted into the rule execution sequence to obtain a rule dependency graph.

8. The personal password credibility assessment method integrating biometrics and rule engine according to claim 7, characterized in that, The method for evaluating the credibility of personal passwords based on a set of cryptographic structure rules, and outputting the password credibility evaluation result, is as follows: Input the password data and biometric data into the password structure rule group, and execute the rule condition judgment one by one; Record the input data fragments and biometric fragments when each rule is triggered to form a set of trigger events; Spatiotemporal clustering analysis is performed on the set of triggering events to obtain event-dense regions; Extract the rule trigger frequency of event-intensive areas and calculate the password credibility score based on the rule trigger frequency.

9. A personal password credibility assessment system integrating biometrics and a rule engine, applied to the personal password credibility assessment method integrating biometrics and a rule engine as described in any one of claims 1-8, characterized in that, It includes a data analysis module, an isolation and identification module, a rule recombination module, and a cryptographic evaluation module: The data analysis module is used to acquire the user's biometric data and password input data, and to parse the password input data to obtain the password structure switching points; The isolation and identification module is used to isolate and identify biometric data based on the cryptographic structure switching point to obtain the first isolation zone; The rule recombination module is used to extract biological change feature sequences from the first isolation zone and recombine the judgment rules in the preset rule engine based on the biological change feature sequences to obtain a cryptographic structure rule group. The password evaluation module is used to evaluate the credibility of an individual's password based on a set of password structure rules and output the password credibility evaluation result.

Images