Teaching question-and-answer methods, equipment, software products, and storage media

Abstract

A teaching question-and-answer method, device, program product, and storage medium relate to the technical field of artificial intelligence education. The method includes: acquiring multi-source interactive behavior data of a user; calculating the user's interaction baseline data; calculating the behavior anomaly degree based on the multi-source interactive behavior data and the interaction baseline data; determining the current cognitive state based on the behavior anomaly degree; when the current cognitive state meets preset triggering conditions, generating an inquiry window containing inquiry content, the inquiry content being determined based on the current cognitive state and the context of the current interaction; acquiring the user's first answer based on the inquiry content; adjusting the inquiry content based on the first answer; acquiring the user's second answer based on the adjusted inquiry content; calculating the user's inquiry acceptance rate based on the first and second answers; and adjusting the triggering conditions based on the inquiry acceptance rate. Implementing the technical solution provided in this application can improve the effectiveness of personalized teaching.

Description

Technical Field

[0001] This application relates to the technical field of artificial intelligence education, specifically to a teaching question-and-answer method, device, program product, and storage medium. Background Technology

[0002] With the rapid development of information technology, online education and intelligent teaching systems have become an important part of the education field. These systems provide learners with personalized learning experiences through computer technology, with question-and-answer systems serving as a core component, providing timely guidance and assistance to users during the learning process.

[0003] Currently, mainstream educational question-and-answer systems typically employ rule-based methods to determine a user's learning status. For example, when the system detects that a user spends more than a preset threshold on a particular learning topic, or answers a question incorrectly in a quiz, it automatically triggers a help mechanism to provide the user with preset hints or solutions.

[0004] However, existing help mechanisms have limitations in their triggering strategies. These systems typically employ static triggering strategies, which are difficult to adapt to the learning habits and immediate states of different users. This can lead to a mismatch between the system's help behavior and the user's actual needs, thus affecting the final effectiveness of personalized teaching. Summary of the Invention

[0005] This application provides a teaching question-and-answer method, device, program product, and storage medium that can improve the effectiveness of personalized teaching.

[0006] The first aspect of this application provides a teaching question-and-answer method, specifically including:

[0007] Acquire multi-source interactive behavior data of users in the teaching Q&A system;

[0008] Calculate the user's interaction baseline data based on the user's historical interaction behavior data in the teaching Q&A system;

[0009] Calculate the user's behavior anomaly degree after a preset time period based on the multi-source interactive behavior data and the interactive baseline data;

[0010] The user's current cognitive state is determined based on the abnormality of the behavior. When the current cognitive state meets the preset triggering conditions, an inquiry window containing inquiry content is generated. The inquiry content is determined based on the current cognitive state and the context of the current interaction.

[0011] Obtain the user's first response based on the query content;

[0012] Based on the first answer, adjust the query content and obtain the user's second answer based on the adjusted query content;

[0013] The user's query acceptance rate is calculated based on the first and second answers, and the triggering conditions are adjusted based on the query acceptance rate.

[0014] By adopting the above technical solution, the system establishes a reference standard for user behavior by acquiring multi-source interactive behavior data of users and combining it with historical interactive behavior data to calculate interactive baseline data. Based on the comparison between real-time interactive behavior and baseline data, the system calculates the degree of behavioral anomaly, achieving accurate identification of the user's cognitive state. The system dynamically generates inquiry content based on the user's current cognitive state and adjusts the inquiry content by analyzing the user's answers, thereby ensuring the matching degree between the inquiry content and the user's cognitive needs. The system introduces inquiry acceptance rate as an evaluation indicator, and through dynamic adjustment of triggering conditions, makes the timing of the inquiry mechanism more in line with the user's learning patterns. This adaptive triggering mechanism based on multi-dimensional interactive data improves the system's accuracy in perceiving the user's learning state and enhances the personalized teaching effect of the teaching question-and-answer system.

[0015] Optionally, calculating the user's behavior anomaly degree after a preset time period based on the multi-source interaction behavior data and the interaction benchmark data includes:

[0016] The multi-source interactive behavior data and the interactive benchmark data are substituted into a preset behavior anomaly formula to calculate the user's behavior anomaly after a preset time period; wherein, the multi-source interactive behavior data includes multiple interactive behavior features, and the interactive benchmark data includes benchmark values ​​of multiple interactive behavior features;

[0017] The formula for the degree of behavioral abnormality is: ;

[0018] in, The abnormality score of user behavior at time t after a preset time period is given, where N is the total number of interaction behavior features. Let be the weight coefficient of the i-th interaction behavior feature. Let be the value of the i-th interaction behavior feature at time t. Let be the baseline value for the i-th interactive behavior feature. The preset stability constant, The time decay coefficient, Let be the time interval between the i-th interactive behavior feature and time t. The rate of change of the i-th interactive behavior feature at time t.

[0019] By adopting the above technical solution, the behavioral anomaly degree formula comprehensively considers the weighted influence of multiple interactive behavioral features, the degree of deviation from the benchmark value, the time decay effect, and the changing trend of behavioral features, thus achieving accurate quantification of the degree of user behavior anomalies. Introducing a stability constant avoids the case where the denominator is zero, ensuring the stability of the calculation; utilizing the exponential relationship between the time decay coefficient and the time interval, the influence of earlier behavioral data on anomaly degree gradually weakens; and combining this with the rate of change of behavioral features enables timely capture of dynamic changes in user behavior. This multi-dimensional behavioral anomaly degree calculation method improves the system's sensitivity and accuracy in recognizing changes in user learning states, providing a reliable data foundation for subsequent cognitive state judgment and inquiry triggering.

[0020] Optionally, the step of determining the user's current cognitive state based on the abnormality of the behavior, and generating an inquiry window containing inquiry content when the current cognitive state meets a preset triggering condition, includes:

[0021] The behavioral anomaly degree is matched with a preset cognitive state table to obtain the cognitive state corresponding to the behavioral anomaly degree. The cognitive state table includes multiple behavioral anomalies and their corresponding current cognitive states.

[0022] The user's learning rhythm parameters are calculated based on the historical interaction behavior data, and the trigger duration threshold for each cognitive state is calculated based on the learning rhythm parameters.

[0023] The duration of the user's continuous state of cognition is counted. When the user is in a preset state that requires intervention and the duration exceeds the corresponding trigger duration threshold, an inquiry window containing inquiry content is generated.

[0024] By adopting the above technical solution, a mapping relationship between behavioral anomaly and cognitive state is established, enabling objective quantitative judgment of users' cognitive states. Learning rhythm parameters are calculated by combining users' historical interaction data, and then the trigger duration thresholds for each cognitive state are dynamically determined based on these parameters, allowing the triggering mechanism to adapt to the learning characteristics of different users. The system also monitors the duration of users in specific cognitive states, only intervening when the duration exceeds the corresponding trigger duration threshold, avoiding overly frequent interruptions. This ensures the system responds promptly to users' learning difficulties while respecting their autonomous learning process, thereby improving the intelligence level and user experience of the teaching question-and-answer system.

[0025] Optionally, calculating the trigger duration threshold for each cognitive state based on the learning rhythm parameter includes:

[0026] Obtain the basic trigger duration corresponding to each cognitive state;

[0027] The learning pace parameter of the user is compared with the average learning pace parameter of all users in the teaching question and answer system to calculate the learning pace difference value.

[0028] When the user's learning rhythm parameter is higher than the average learning rhythm parameter, the corresponding basic trigger duration is shortened according to the learning rhythm difference value. When the user's learning rhythm parameter is lower than the average learning rhythm parameter, the corresponding basic trigger duration is extended according to the learning rhythm difference value, thus obtaining the trigger duration threshold for each cognitive state.

[0029] By adopting the above technical solution, dynamic adjustment of the trigger duration threshold is achieved based on the basic trigger duration and combined with the user's personalized learning characteristics. By comparing the user's learning pace parameters with the average level of all users in the system, the degree of individual difference is calculated, providing differentiated trigger duration settings for users with different learning paces. For users with a faster learning pace, the system shortens the trigger duration threshold accordingly to provide more timely support; for users with a slower learning pace, the trigger duration threshold is appropriately extended to allow more time for independent learning. This adaptive triggering mechanism based on learning pace enables the system to better match the learning characteristics of different users, improving the personalization of teaching support.

[0030] Optionally, calculating the user's inquiry acceptance rate based on the first and second answers includes:

[0031] The inquiry acceptance rate is calculated using a preset inquiry acceptance rate formula;

[0032] The formula for the inquiry acceptance rate is: ;

[0033] in, To explore acceptance rates, and These are the acceptance weights for the corresponding options in the first and second answers, respectively. and These represent the maximum and minimum acceptance weights among all options in the first and second answers, respectively. To select the coefficient of variation, and These are the response times for the first answer and the second answer, respectively. This is the reference response time for the user in their current cognitive state. For time weighting coefficients, The cumulative number of times the user selected the "Need Help" option in the first and second answers. The total number of times the user received queries in the historical interaction data. Historical acceptance coefficient is the numerical stability constant.

[0034] By adopting the above technical solution, a comprehensive formula for calculating inquiry acceptance rate was designed, taking into account multiple influencing factors: the immediate acceptance level of users is reflected by the acceptance weights of the first and second responses and their changes; a comparison between response time and reference time, combined with a time weight coefficient, reflects the user's responsiveness to the inquiry; and the long-term acceptance tendency of users is reflected by the frequency of users selecting "need help" in historical data, combined with a historical acceptance coefficient. The formula also includes a selection variation coefficient and a numerical stability constant to ensure the rationality and stability of the calculation results. This multi-dimensional inquiry acceptance rate calculation method can more accurately assess the user's acceptance level of the inquiry, providing effective data support for optimizing the system's inquiry strategy and improving the adaptability and effectiveness of the teaching question-and-answer system.

[0035] Optionally, adjusting the triggering condition based on the query acceptance rate includes:

[0036] When there is a first cognitive state where the inquiry acceptance rate is lower than a preset lower threshold, the trigger duration threshold of the first cognitive state is extended according to the behavioral abnormality of the first cognitive state.

[0037] When there is a second cognitive state where the inquiry acceptance rate is not lower than a preset upper limit threshold and the behavioral abnormality is lower than a preset abnormality benchmark value, the trigger duration threshold of the second cognitive state is shortened.

[0038] When adjusting the trigger duration threshold of the first cognitive state or the second cognitive state, other cognitive states with similar learning parameters to the first cognitive state or the second cognitive state are identified based on the multi-source interaction behavior data, and the trigger duration threshold of each of the other cognitive states is adjusted synchronously.

[0039] By employing the aforementioned technical solutions, for cognitive states with low inquiry acceptance rates, the system reduces inquiry frequency by extending the trigger duration threshold, avoiding excessive intervention that could negatively impact the user's learning experience. Conversely, for cognitive states with high inquiry acceptance rates and low behavioral abnormality, the trigger duration threshold is appropriately shortened to provide more timely support and fully leverage user engagement. Furthermore, the system identifies cognitive states with similar learning parameters to achieve correlated adjustment of trigger conditions, allowing the adjustment effect to be reasonably generalized across similar scenarios. This multi-layered trigger condition adjustment strategy considers both the characteristics of individual cognitive states and the correlations between different cognitive states, enabling the system to more intelligently balance the timeliness of inquiry intervention with user acceptance, thereby improving the accuracy and effectiveness of teaching support. Simultaneously, the synchronous adjustment of similar states also improves the efficiency of the system's adaptive adjustment.

[0040] Optionally, after adjusting the triggering condition based on the query acceptance rate, the method further includes:

[0041] The changes in the response rate of the inquiry under each cognitive state within a preset period after statistical adjustment are analyzed, and the increase in the response rate of the inquiry before and after adjustment is calculated.

[0042] When the improvement exceeds a preset effect threshold, the baseline value of the user interaction behavior feature in the interaction baseline data is updated;

[0043] When the improvement is lower than the effect threshold, the learning rhythm parameter is recalculated based on the historical interaction behavior data, and the trigger duration threshold of each cognitive state is adjusted based on the recalculated learning rhythm parameter.

[0044] By employing the above technical solution, the system evaluates the effectiveness of trigger condition adjustments by statistically analyzing changes in the inquiry acceptance rate within a preset period. When the adjusted inquiry acceptance rate significantly improves, it indicates that the current triggering strategy is relatively effective, and the system updates the baseline values ​​of user interaction behavior characteristics in the interaction benchmark data accordingly to adapt to the dynamic changes in user behavior characteristics. When the improvement effect is not ideal, the system recalculates the learning rhythm parameters and adjusts the trigger duration threshold accordingly to ensure that the inquiry triggering mechanism can better match the user's actual learning patterns. This dynamic optimization mechanism based on effect evaluation enables the system to continuously adjust and improve the inquiry strategy, constantly enhancing the relevance and effectiveness of inquiries. Simultaneously, through dynamic data updates, the system maintains its sensitivity to user learning characteristics, thereby achieving continuous optimization of teaching support.

[0045] In a second aspect, this application provides a teaching question-and-answer device, which includes: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the teaching question-and-answer device to perform the method described in the first aspect and any possible implementation thereof.

[0046] Thirdly, this application provides a computer program product containing instructions that, when run on an instructional question-and-answer device, cause the instructional question-and-answer device to perform the method described in the first aspect and any possible implementation thereof.

[0047] Fourthly, this application provides a computer-readable storage medium including instructions that, when executed on an instructional question-and-answer device, cause the instructional question-and-answer device to perform the method described in the first aspect and any possible implementation thereof. Attached Figure Description

[0048] Figure 1 This is a system architecture diagram of a teaching question-and-answer system provided in an embodiment of this application;

[0049] Figure 2 This is a flowchart illustrating a teaching question-and-answer method provided in an embodiment of this application;

[0050] Figure 3 This is a timing interaction diagram of a teaching question and answer provided in an embodiment of this application.

[0051] Figure 4 This is a schematic diagram of an exemplary hardware structure of a teaching question-and-answer device provided in an embodiment of this application. Detailed Implementation

[0052] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0053] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0054] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0055] Figure 1 An architecture for a teaching question-and-answer system is shown. For example... Figure 1 As shown, the system architecture may include a client 011, a network 012, and an electronic device 013. The network 012 provides a data transmission link between the client 011 and the electronic device 013. The network 012 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0056] Client 011 can send interactive behavior data to electronic device 013 via network 012. Client 011 is mainly responsible for collecting multi-source interactive behavior data of users in the teaching question and answer system, and assisting in the display of the inquiry window and the collection of user answers.

[0057] Client 011 can be hardware, such as a mobile terminal, tablet computer, personal computer, or other device with a user interface, used to realize human-computer interaction between the user and the teaching question-and-answer system.

[0058] Electronic device 013 is responsible for receiving and comprehensively analyzing interactive behavior data, including core functions such as interaction benchmark calculation, behavior anomaly analysis, cognitive state judgment, inquiry content generation, inquiry acceptance rate calculation, and trigger condition adjustment. Electronic device 013 can adaptively adjust inquiry strategies based on user responses and preset thresholds, calculate the optimal trigger timing, and, combined with preset evaluation rules, ultimately achieve dynamic monitoring of the user's learning status. These analysis and processing results can be used to improve the personalized teaching effectiveness of the teaching question-and-answer system.

[0059] It should be noted that electronic devices can be either hardware or software. When an electronic device is hardware, it can be implemented as a distributed cluster of multiple electronic devices or as a single electronic device. When an electronic device is software, it can be implemented as multiple software programs or software modules (e.g., multiple software programs or software modules used to provide distributed processing) or as a single software program or software module. No specific limitations are set here.

[0060] It should be understood that Figure 1 The number of clients 011, networks 012, and electronic devices 013 shown is merely illustrative. Depending on implementation needs, there can be any number of clients 011, networks 012, and electronic devices 013. Specifically, in a local teaching question-and-answer system, the above system architecture may exclude network 012 and only include clients 011 or electronic devices 013.

[0061] This application provides a teaching question-and-answer method, for reference. Figure 2 , Figure 2 This is a flowchart illustrating a teaching question-and-answer method provided in an embodiment of this application, including steps S101 to S106, as follows:

[0062] S101: Obtain multi-source interactive behavior data of users in the teaching Q&A system.

[0063] In this embodiment, multi-source interactive behavior data refers to multi-dimensional behavioral data records generated by users in the teaching Q&A system, including but not limited to user click operation data, dwell time data, input content data, page switching data, and resource access data. Specifically, click operation data represents the location, frequency, and time information of user clicks, swipes, selections, and other operations performed on the system interface; dwell time data represents the time users spend on various learning content pages; input content data represents the text content entered by users in Q&A, practice, and other sections; page switching data represents the frequency and order of user jumps between different learning modules; and resource access data represents user records of downloading, saving, and sharing learning resources.

[0064] Specifically, by setting up a data acquisition module on the client side, various user interactions within the teaching Q&A system are captured and recorded in real time. For clicks, the system records the coordinates, trigger time, and duration of each click; for page dwells, it tracks the user's entry time, exit time, and cumulative duration on each content page; for content input, it saves the text content entered in the input box, input speed, and number of modifications; for page switching, it records the timestamps of user access to each functional module and the switching path; and for resource access, it tracks the frequency and methods of user use of different types of learning resources. These multi-dimensional interaction data are then timestamped and formatted to form multi-source interaction data.

[0065] S102: Calculate the user's interaction baseline data based on the user's historical interaction behavior data in the teaching Q&A system.

[0066] In this embodiment, historical interaction behavior data refers to the records of user interaction behavior accumulated during past learning processes, including historical data information such as learning tasks, interactive operations, and question-and-answer records completed by the user in the teaching question-and-answer system. Interaction baseline data represents the standard behavioral characteristic parameters of a user under normal learning conditions, used to measure whether the user's current behavior has deviated abnormally.

[0067] Specifically, the first step is to calculate the mean of each interaction behavior feature in the historical interaction behavior data. Taking clicks as an example, the average number of clicks per user in each learning task is calculated to obtain a baseline value for click count; the average response time per user in each Q&A session is calculated to obtain a baseline value for response time; and the average dwell time per user in each learning module is calculated to obtain a baseline value for dwell time. For each interaction behavior feature, the mean value is calculated based on its performance data from the last 30 learning tasks and set as the baseline value for that feature. The baseline values ​​of all interaction behavior features are then integrated to form the interaction baseline data.

[0068] S103: Calculate the degree of user behavior abnormality after a preset time period based on multi-source interactive behavior data and interactive baseline data.

[0069] In this embodiment of the application, the abnormality of behavior refers to the degree of deviation between the user's current interactive behavior and its historical normal behavior, and is used to quantitatively assess whether the user's behavior is abnormal.

[0070] Specifically, multi-source interactive behavior data and interaction benchmark data are substituted into a preset behavior anomaly formula to calculate the user's behavior anomaly degree after a preset time period. The multi-source interactive behavior data includes multiple interactive behavior features, and the interaction benchmark data includes benchmark values ​​for multiple interactive behavior features. The behavior anomaly formula is as follows: ;in, The abnormality score of user behavior is defined as the time point after a preset time period, where N is the total number of interaction behavior features, determined by statistically analyzing the types of behavior features recorded in the interaction behavior database. Let be the weight coefficient of the i-th interactive behavior feature, calculated by analyzing the importance of this behavior feature in historical query-triggered events. Let be the value of the i-th interaction behavior feature at time t, extracted from multi-source interaction behavior data. The baseline value for the i-th interactive behavior feature is extracted from the interactive baseline data. The preset stability constant, The time decay coefficient is determined by analyzing the impact of the timeliness of behavioral data. Let be the time interval between the i-th interactive behavior feature and the time point. The rate of change of the i-th interactive behavior feature at time t is obtained by calculating the slope of the numerical change of the interactive behavior feature within a preset time window.

[0071] The behavioral anomaly formula adopts a multi-feature weighted fusion structure, mainly consisting of four core components: the basic deviation calculation part. Used to quantify the relative deviation of the current behavioral feature value from the benchmark value, normalization is achieved by dividing by the benchmark value; time decay part Weights used to reduce the impact of behavioral features with longer time intervals on the current anomaly level; rate of change enhancement component. This is used to capture the dynamic changing trends of behavioral features, increasing the anomaly score when behavior changes; the weight fusion part achieves comprehensive evaluation by assigning different importance weights to different interactive behavioral features. The advantage of this multi-dimensional fusion structure is that it considers both static deviations of behavior and dynamic changing features, while ensuring the timeliness of anomaly score calculation through time decay.

[0072] For example, suppose a user has three interaction behavior characteristics, with weights as follows: =0.4、 =0.3、 =0.3, the values ​​of the interaction behavior features at the current time are respectively =50、 =8、 =120, the baseline values ​​for interactive behavior features are respectively =40、 =10、 =100, time intervals are all 1 hour, and change rates are 5, 2, and 10 respectively. Parameter settings. =0.1、 =0.05, then the calculated abnormality degree A(t) is 0.4×(|50-40| / (40+0.1))×e^(-0.05×1)×(1+5)+0.3×(|8-10| / (10+0.1))×e^(-0.05×1)×(1+2)+0.3×(|120-100| / (100+0.1))×e^(-0.05×1)×(1+10)≈0.89.

[0073] S104: Determine the user's current cognitive state based on the degree of behavioral abnormality. When the current cognitive state meets the preset triggering conditions, generate an inquiry window containing inquiry content. The inquiry content is determined based on the current cognitive state and the context of the current interaction.

[0074] In this embodiment, the current cognitive state refers to the user's current learning cognitive level state determined based on the degree of user behavior abnormality, including different types such as focused state, distracted state, confused state, and fatigued state. The inquiry window refers to an interactive interface that pops up proactively by the system when an abnormal user cognitive state is detected, containing targeted inquiry content, used to obtain the user's true learning feelings and needs.

[0075] Specifically, the calculated behavioral anomaly score is first matched with a pre-defined cognitive state table to obtain the current cognitive state corresponding to the behavioral anomaly score. The cognitive state table pre-establishes a mapping relationship between multiple behavioral anomaly score ranges and corresponding cognitive states; for example, anomaly scores of 0-0.3 correspond to a focused state, 0.3-0.6 to a slightly distracted state, 0.6-0.8 to a confused state, and 0.8-1.0 to a severely fatigued state. Then, the user's learning rhythm parameters are calculated based on historical interaction data, and the trigger duration threshold for each cognitive state is dynamically calculated according to these parameters. Next, the duration for which the user is continuously in the current cognitive state is recorded. When the user is in a pre-defined cognitive state requiring intervention (such as a confused state or a fatigued state) and the duration exceeds the corresponding trigger duration threshold, the system automatically generates an inquiry window containing inquiry content. The inquiry content is personalized based on the current cognitive state and the context of the current interaction. For example, when the user is in a confused state, the inquiry content might include targeted questions such as "Are you finding the current learning content difficult? Do you need additional explanations?"

[0076] Based on the above embodiments, as an optional embodiment, S104: determining the user's current cognitive state based on the degree of behavioral abnormality, and generating an inquiry window containing inquiry content when the current cognitive state meets the preset triggering conditions, may specifically include the following steps:

[0077] S301: Match the behavioral abnormality degree with the preset cognitive state table to obtain the cognitive state corresponding to the behavioral abnormality degree. The cognitive state table includes multiple behavioral abnormalities and their corresponding current cognitive states.

[0078] In this embodiment of the application, the cognitive state table refers to a pre-established mapping table that corresponds to the numerical range of behavioral abnormality and the type of user cognitive state. It is used to convert the quantified behavioral abnormality into an interpretable cognitive state label, which facilitates subsequent personalized intervention decisions.

[0079] Specifically, the calculated behavioral anomaly score is matched against a preset cognitive state table to determine the range to which the behavioral anomaly score belongs, thus obtaining the corresponding current cognitive state. The cognitive state table is established based on statistical analysis of a large amount of user behavior data and includes multiple behavioral anomaly score threshold ranges and corresponding cognitive state types. For example, an anomaly score in the range [0, 0.2] corresponds to "highly focused state", an anomaly score in the range (0.2, 0.4] corresponds to "normal learning state", an anomaly score in the range (0.4, 0.6] corresponds to "mildly distracted state", an anomaly score in the range (0.6, 0.8] corresponds to "confused state", and an anomaly score in the range (0.8, 1.0] corresponds to "fatigue state".

[0080] S302: Calculate the user's learning rhythm parameters based on historical interaction behavior data, and calculate the trigger duration threshold for each cognitive state based on the learning rhythm parameters.

[0081] In this embodiment, the learning rhythm parameter refers to a quantitative indicator reflecting the user's learning speed and rhythm characteristics, derived from statistical data of the user's historical interaction behavior. This includes parameters such as average task completion time, interaction frequency, and learning duration distribution. The trigger duration threshold refers to a time threshold set for different cognitive states. When the duration of a user's cognitive state exceeds this threshold, an inquiry window is triggered.

[0082] Specifically, the system first acquires historical user interaction data within a preset time window, including learning task completion records, page access logs, and interaction operation sequences. Key behavioral indicators are extracted, such as average completion time for different learning tasks, number of clicks per unit time, and average dwell time on each page. Based on these extracted behavioral indicators, a weighted average is used to obtain a learning pace parameter that comprehensively reflects the user's learning pace. Then, the basic trigger duration corresponding to each cognitive state is obtained as an initial reference value. The user's learning pace parameter is compared with the average learning pace parameter of all users in the teaching Q&A system to calculate the learning pace difference value. When the user's learning pace parameter is higher than the average learning pace parameter, the corresponding basic trigger duration is shortened proportionally to the learning pace difference value; when the user's learning pace parameter is lower than the average learning pace parameter, the corresponding basic trigger duration is extended proportionally to the learning pace difference value, ultimately obtaining trigger duration thresholds for each cognitive state that adapt to the user's personalized learning pace.

[0083] Based on the above embodiments, as an optional embodiment, S302: the step of calculating the trigger duration threshold of each cognitive state according to the learning rhythm parameter may specifically include the following steps:

[0084] S401: Obtain the basic trigger duration corresponding to each cognitive state, compare the user's learning rhythm parameter with the average learning rhythm parameter of all users in the teaching question and answer system, and calculate the learning rhythm difference value.

[0085] In this embodiment of the application, the basic trigger duration refers to the standard time threshold preset for each cognitive state, which is used to indicate the reference length of time that the user should stay in the cognitive state at the standard learning speed. For example, the basic trigger duration of the confused state can be set to 30 seconds, which means that intervention should be considered when the user stays in the confused state for more than 30 seconds.

[0086] Specifically, the basic trigger duration corresponding to each cognitive state is first obtained from the preset cognitive state configuration table. Then, the learning rhythm parameter of the current user is compared with the average learning rhythm parameter of all users in the system. The difference between the two is calculated and normalized to obtain the learning rhythm difference value.

[0087] S402: When the user's learning rhythm parameter is higher than the average learning rhythm parameter, shorten the corresponding basic trigger duration according to the learning rhythm difference value. When the user's learning rhythm parameter is lower than the average learning rhythm parameter, extend the corresponding basic trigger duration according to the learning rhythm difference value to obtain the trigger duration threshold for each cognitive state.

[0088] Specifically, the process first determines the relationship between the user's learning pace parameter and the average learning pace parameter. When the user's learning pace parameter is higher than the average, the base trigger duration for each cognitive state is shortened proportionally to the difference in learning pace, with the shortening amount being directly proportional to the difference in learning pace. When the user's learning pace parameter is lower than the average, the base trigger duration for each cognitive state is extended proportionally to the difference in learning pace, with the extension amount being directly proportional to the absolute value of the difference in learning pace. Through this adjustment process, the trigger duration thresholds for each cognitive state are ultimately obtained to adapt to the user's current personalized learning pace.

[0089] For example, suppose the system presets the basic trigger duration for three cognitive states: 300 seconds for confusion, 180 seconds for doubt, and 120 seconds for hesitation. A user named Xiaoming has a learning pace parameter of 1.5, while the system calculates an average learning pace parameter of 1.2. The difference in learning pace is 1.5 - 1.2 = 0.3. Since Xiaoming's learning pace parameter is higher than the average, it indicates that his learning speed is faster, and the system needs to shorten the trigger duration to initiate inquiries more promptly. Assuming a shortening ratio of 0.2, the trigger duration threshold for confusion would be adjusted to 300 × (1 - 0.3 × 0.2) = 300 × 0.94 = 282 seconds, for doubt to 180 × 0.94 = 169.2 seconds, and for hesitation to 120 × 0.94 = 112.8 seconds. Conversely, if another user, Xiao Li, has a learning rhythm parameter of 0.9, which is lower than the average of 1.2, the absolute value of the difference in learning rhythm is |0.9-1.2|=0.3. Assuming the extension ratio is 0.15, the trigger duration threshold for the confused state is adjusted to 300×(1+0.3×0.15)=300×1.045=313.5 seconds. The trigger duration thresholds for other cognitive states can be adjusted in the same way.

[0090] S303: Count the duration of a user’s current cognitive state. When a user is in a preset cognitive state that requires intervention and the duration exceeds the corresponding trigger duration threshold, generate an inquiry window containing inquiry content.

[0091] Specifically, the system first continuously tracks the duration of a user's current cognitive state, where duration represents the continuous time from entering a cognitive state to the current moment. Then, it determines whether the user's current cognitive state falls under the category of cognitive states requiring intervention, and simultaneously checks whether the duration of this state has exceeded the corresponding trigger duration threshold. When both conditions are met—the user being in a cognitive state requiring intervention and the duration exceeding the corresponding trigger duration threshold—the system generates an inquiry window, ultimately forming a complete inquiry window containing the inquiry content and presenting it to the user.

[0092] S105: Obtain the user's first answer based on the query content, adjust the query content based on the first answer, and obtain the user's second answer based on the adjusted query content.

[0093] In this embodiment of the application, the adjusted inquiry content refers to a new inquiry text that has been intelligently optimized and improved based on the user's initial response feedback. It is used to represent the personalized inquiry content that the system dynamically adjusts according to the user's response. For example, when the user answers "I don't quite understand this concept" to the initial inquiry "What difficulties have you encountered?", the system may adjust the inquiry content to "Which specific concept would you like me to explain in detail?".

[0094] Specifically, the system first obtains the user's initial response to the query. This initial response refers to the user's direct reaction and feedback to the system's initial query window content. It includes the options selected by the user in the query window, the text content entered, and the user's response behavior data (such as response time, click location, and duration of hesitation). For example, when the system asks "How is your current learning status?", the user might select "feeling confused" or enter "I don't quite understand this knowledge point" in the text box. These selections and inputs, along with behavioral characteristics such as response time, constitute the initial response. The system then performs semantic analysis and intent recognition on the initial response to extract the user's true needs, level of confusion, and specific problem points. Based on these analysis results, the system intelligently adjusts the original query content, including refining the question's wording and adjusting the angle of inquiry. Finally, the adjusted query content is presented back to the user to obtain the user's second response based on the adjusted query content. The second response refers to the user's further response and confirmation information to the query content intelligently adjusted by the system based on the first response. It also includes the user's selection results, text input, and corresponding behavioral data, but its content is more specific and precise because the query content has been specifically optimized based on the first response. For example, if the user said "I don't quite understand this concept" in the first response, the system adjusts the query content to "Which specific concept point would you like me to explain in detail?" The user's second response may be to select a specific concept option or enter a more detailed description of the question. Through the comparative analysis of the two rounds of queries and responses, the system can more accurately assess the user's acceptance and cooperation with the queries.

[0095] S106: Calculate the user's query acceptance rate based on the first and second answers, and adjust the triggering conditions based on the query acceptance rate.

[0096] In this embodiment of the application, the inquiry acceptance rate refers to a quantitative indicator calculated by comprehensively analyzing the quality and acceptance of users' responses to system inquiries. It is used to represent the user's cooperation and satisfaction level with the system's proactive intervention. For example, when a user responds positively to the inquiry content and chooses to accept help, the inquiry acceptance rate may reach 0.85, while when a user responds negatively or refuses help, the inquiry acceptance rate may drop to 0.3.

[0097] Specifically, the response rate (RR) is first calculated using a preset formula: ;

[0098] in, To explore acceptance rates, and These are the acceptance weights for the options corresponding to the first and second answers, calculated by analyzing the correlation between each option and the user's subsequent learning effect in the historical interaction data. and These represent the maximum and minimum acceptance weights among all options in both the first and second answers, respectively. To select the coefficient of variation, the impact of users' choices to change during continuous exploration on learning engagement was determined by statistical analysis. and The response times for the first and second answers are extracted and calculated from the user interaction timestamps. The reference response time for the user in their current cognitive state is obtained from the median response time in the historical interaction data. For time weighting coefficients, To calculate the cumulative number of times a user selected the "Need Help" option in the first and second answers. The total number of queries a user receives in historical interaction behavior data is calculated from that historical interaction behavior data. Historical acceptance coefficient is the numerical stability constant.

[0099] The inquiry acceptance rate formula adopts a multi-factor weighted structure. The first part reflects the user's basic cooperation level through the average acceptance weight. The second part reflects the consistency of the user's attitude through the selection of the variation coefficient. The third part considers the impact of response time on acceptance through the exponential decay function. The fourth part reflects the user's long-term behavioral pattern through the historical acceptance ratio. This construction method can comprehensively and accurately quantify the user's inquiry acceptance level.

[0100] For example, suppose =0.7, =0.8, =1.0, =0.1, =0.5, =3 seconds, =2.5 seconds, =4 seconds, =1.2, =2, =10, =0.3, =0.01, then =0.75×1.056×0.632×1.06≈0.53.

[0101] Then, the triggering conditions are adjusted according to the inquiry acceptance rate. When there is a first cognitive state where the inquiry acceptance rate is lower than the preset lower threshold, the triggering duration threshold of the first cognitive state is extended according to the behavioral abnormality of the first cognitive state. When there is a second cognitive state where the inquiry acceptance rate is not lower than the preset upper threshold and the behavioral abnormality is lower than the preset abnormality benchmark value, the triggering duration threshold of the second cognitive state is shortened. When adjusting the triggering duration threshold of the first or second cognitive state, other cognitive states with similar learning parameters to the first or second cognitive state are identified based on multi-source interactive behavior data, and the triggering duration threshold of each other cognitive state is adjusted synchronously.

[0102] Based on the above embodiments, as an optional embodiment, S106: the step of adjusting the triggering condition according to the inquiry acceptance rate may specifically include the following steps:

[0103] S601: When there is a first cognitive state where the inquiry acceptance rate is lower than the preset lower threshold, the trigger duration threshold of the first cognitive state is extended according to the behavioral abnormality of the first cognitive state.

[0104] In the embodiments of this application, the first cognitive state refers to a specific user cognitive state type where the inquiry acceptance rate is lower than a preset lower threshold. It is used to indicate a learning cognitive state in which the user shows a low degree of cooperation or resistance to the system's inquiry intervention. For example, when the user is in a confused state but his inquiry acceptance rate is only 0.2 (lower than the lower threshold of 0.4), the confused state is identified as the first cognitive state.

[0105] Specifically, the system first detects whether there is a first cognitive state where the inquiry acceptance rate is lower than a preset lower threshold. When a user's inquiry acceptance rate in a certain cognitive state does not reach the lower threshold, that cognitive state is marked as the first cognitive state. Once the first cognitive state is identified, the system further analyzes the user's behavioral abnormality in that state. By comparing the user's behavioral data in this cognitive state with their personal historical behavioral patterns, the system calculates behavioral deviation indicators across multiple dimensions, including abnormal dwell time, changes in operation frequency, and interaction response delays. After identifying the first cognitive state, the system increases the trigger duration threshold value corresponding to that cognitive state. The difference between the lower threshold and the actual inquiry acceptance rate is used as the first extension factor, which is proportional to the extension magnitude. Behavioral abnormality is used as the second extension factor, which is also proportional to the extension magnitude. The original trigger duration threshold is multiplied by a composite extension coefficient composed of the first and second extension factors to obtain a new trigger duration threshold with a larger value.

[0106] S602: When there is a second cognitive state where the inquiry acceptance rate is not lower than the preset upper limit threshold and the behavioral abnormality is lower than the preset abnormality benchmark value, shorten the trigger duration threshold of the second cognitive state.

[0107] In this embodiment of the application, the second cognitive state refers to a specific user cognitive state type where the inquiry acceptance rate is not lower than a preset upper limit threshold and the behavioral abnormality is lower than a preset abnormality benchmark value. It is used to represent a learning cognitive state in which the user shows a high degree of cooperation with the system's inquiry intervention and the behavior pattern is normal and stable. For example, when the user is in a focused state and his inquiry acceptance rate reaches 0.85 (not lower than the upper limit threshold of 0.7) while the behavioral abnormality is only 0.15 (lower than the benchmark value of 0.3), the focused state is identified as the second cognitive state.

[0108] Specifically, the system first detects whether there are cognitive states where the inquiry acceptance rate is not lower than a preset upper threshold. Then, it further filters out cognitive states where the behavioral anomalousness is lower than a preset anomalousness benchmark value, and marks cognitive states that meet both conditions as the second cognitive state. When the second cognitive state is identified, the system reduces the trigger duration threshold value corresponding to that cognitive state. This is done by using the inquiry acceptance rate as the first shortening factor, which is proportional to the shortening magnitude, and the reciprocal of the behavioral anomalousness as the second shortening factor, which is also proportional to the shortening magnitude. The original trigger duration threshold is then divided by a composite shortening coefficient composed of the first and second shortening factors to obtain a new trigger duration threshold with a smaller value.

[0109] S603: When adjusting the trigger duration threshold of the first cognitive state or the second cognitive state, other cognitive states with similar learning parameters to the first cognitive state or the second cognitive state are identified based on multi-source interactive behavior data, and the trigger duration threshold of each other cognitive state is adjusted synchronously.

[0110] In the embodiments of this application, similarity learning parameters refer to the similarity indicators exhibited by different cognitive states in terms of multi-dimensional learning behavior characteristics such as learning efficiency, attention duration, interaction response mode, and task completion speed, which are used to measure the degree of similarity of user learning behavior patterns under different cognitive states.

[0111] Specifically, after the system adjusts the trigger duration threshold for the first or second cognitive state, it constructs a learning parameter vector for each cognitive state by analyzing multi-source interactive behavior data. This vector includes multi-dimensional values ​​such as learning efficiency, attention concentration, interaction frequency, and task switching frequency. The similarity of the learning parameters between the target cognitive state and other cognitive states is calculated. The similarity metric is the cosine similarity or the reciprocal of the Euclidean distance between the learning parameter vectors. When the similarity exceeds a preset similarity threshold, the corresponding cognitive state is marked as another cognitive state with similar learning parameters. The adjustment range of the trigger duration threshold for the target cognitive state is multiplied by the corresponding similarity coefficient to obtain the adjustment range for each other cognitive state, thus achieving synchronous adjustment of the trigger duration thresholds for multiple other cognitive states with similar learning parameters.

[0112] Based on the above embodiments, as an optional embodiment, S106: after the step of adjusting the triggering condition according to the inquiry acceptance rate, a step of adjusting the interaction baseline data is further included, which may specifically include the following steps:

[0113] S701: Statistically analyze the changes in the inquiry acceptance rate under each cognitive state within the preset period after adjustment, and calculate the increase in inquiry acceptance rate before and after adjustment.

[0114] Specifically, after the system adjusts the trigger duration threshold, it statistically analyzes the query acceptance rate for each cognitive state within a preset period after the adjustment. The adjusted query acceptance rate is calculated by dividing the number of queries accepted by each cognitive state within the preset period by the total number of queries, thus constructing a statistical vector containing the adjusted query acceptance rate for each cognitive state. The difference between the adjusted query acceptance rate and the baseline query acceptance rate before adjustment is calculated to obtain the absolute improvement for each cognitive state. Simultaneously, the ratio of the adjusted query acceptance rate to the pre-adjustment query acceptance rate is subtracted by one to obtain the relative improvement for each cognitive state. Furthermore, the improvement for each cognitive state is weighted and averaged according to its query frequency within the preset period to obtain the overall improvement in the system's query acceptance rate.

[0115] S702: When the improvement exceeds the preset effect threshold, update the baseline value of the user interaction behavior feature in the interaction baseline data.

[0116] In this embodiment, the effect threshold refers to a critical value used to determine whether the trigger duration threshold adjustment strategy has produced a significant improvement effect. When the increase in the inquiry acceptance rate reaches this threshold, it indicates that the user's interaction behavior pattern has undergone a positive adaptive change, and the system's behavior judgment benchmark needs to be updated accordingly.

[0117] Specifically, when the system detects that the increase in the inquiry acceptance rate exceeds a preset effect threshold, it extracts the latest interactive behavior data of users in various cognitive states within a preset period, including behavioral characteristic indicators such as response time distribution, operation frequency statistics, task completion efficiency, and interaction duration. The latest behavioral data is then weighted and fused with the original baseline value. A new baseline value is obtained by multiplying the original baseline value by historical weight coefficients and adding the latest behavioral characteristic mean multiplied by updated weight coefficients. This new baseline value serves as the benchmark value for user interactive behavior characteristics in the interactive baseline data.

[0118] S703: When the improvement is lower than the effect threshold, the learning rhythm parameters are recalculated based on historical interaction behavior data, and the trigger duration threshold of each cognitive state is adjusted based on the recalculated learning rhythm parameters.

[0119] In this embodiment of the application, when the increase in the inquiry acceptance rate is lower than the effect threshold, it indicates that the current trigger duration threshold adjustment strategy based on the learning rhythm parameter has failed to produce the expected improvement effect, and it is necessary to re-examine and optimize the understanding and parameterization of the user's learning rhythm pattern.

[0120] Specifically, when the system detects that the improvement is below the effectiveness threshold, the analysis time window for historical interaction data is extended to a longer period. More complex data processing methods, such as time series analysis and frequency domain analysis, are introduced to re-identify deeper learning rhythm characteristics, such as the periodic patterns of user learning activities, attention decay patterns, and task switching frequency distribution. Cluster analysis is used to group historical behavior data by pattern, identifying differences in learning rhythm across different time periods and learning content. Statistical characteristic values ​​of learning rhythm parameters, such as peak learning efficiency time, attention duration distribution, and optimal inquiry interval, are recalculated. Based on the recalculated learning rhythm parameters, the optimal triggering time for each cognitive state is matched with the updated learning rhythm characteristics. By using key time features such as peak learning efficiency time, attention stability interval, and task completion node as weighting factors, the triggering duration threshold for each cognitive state is reset.

[0121] Please see Figure 3 , Figure 3 This is a timing interaction diagram of a teaching question and answer provided in an embodiment of this application.

[0122] Figure 3This demonstrates the complete interaction process among three participants: the user, the client, and the electronic device. The entire process begins with the user's actions within the educational Q&A system. As the user engages in various learning activities through the client, multi-source interactive behavior data is generated. The client is responsible for collecting this data and sending it to the electronic device for analysis. Upon receiving the interactive behavior data, the electronic device first calculates baseline interaction data based on the user's historical interaction data. Then, it combines this with the current multi-source interactive behavior data to calculate the user's behavioral anomaly level. Both of these steps are internal calculations within the electronic device. Next, the electronic device determines the user's current cognitive state based on the calculated behavioral anomaly level and checks if this state meets preset trigger conditions. If the trigger conditions are met, the electronic device generates corresponding inquiry content and sends an inquiry window containing this content to the client. The client then displays this inquiry window to the user. Upon seeing the inquiry window, the user provides an initial response. The client sends this initial response back to the electronic device, which intelligently adjusts the inquiry content based on this response. The adjusted inquiry window is then sent back to the client, which displays the adjusted window to the user. The user then provides a second response to the adjusted inquiry content, which the client also sends back to the electronic device. Finally, the electronic device comprehensively analyzes the user's first and second answers, calculates the user's inquiry acceptance rate, and dynamically adjusts the system's triggering conditions based on this inquiry acceptance rate, thereby completing the closed-loop process of the entire teaching question-and-answer method and realizing the system's adaptive optimization of the user's learning state.

[0123] The following describes an exemplary teaching question-and-answer device provided in the embodiments of this application. Figure 4 This is a schematic diagram of an exemplary hardware structure of the teaching question-and-answer device provided in the embodiments of this application.

[0124] In some embodiments, the teaching question-and-answer device is a computer device or includes a computer device. The computer device includes a processor, memory, and a network interface connected via a system bus. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores data. The network interface of the computer device is used to communicate with other external terminals or servers via a network connection. In some embodiments, the network interface can be a wired network interface; in some embodiments, the network interface can also be a wireless network interface. When the computer program is executed by the processor, it implements the methods in the embodiments of this application.

[0125] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0126] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0127] As used in the above embodiments, depending on the context, the term "when..." can be interpreted as meaning "if...", "after...", "in response to determining...", or "in response to detecting...". Similarly, depending on the context, the phrase "when determining..." or "if (the stated condition or event) is interpreted as meaning "if determining...", "in response to determining...", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0128] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. 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 integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0129] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A method of teaching question and answer, characterized by, The method includes: Acquire multi-source interactive behavior data of users in the teaching Q&A system; Calculate the user's interaction baseline data based on the user's historical interaction behavior data in the teaching Q&A system; Calculate the user's behavior anomaly degree after a preset time period based on the multi-source interactive behavior data and the interactive baseline data; The user's current cognitive state is determined based on the abnormality of the behavior. When the current cognitive state meets the preset triggering conditions, an inquiry window containing inquiry content is generated. The inquiry content is determined based on the current cognitive state and the context of the current interaction. Obtain the user's first answer based on the query content, adjust the query content based on the first answer, and obtain the user's second answer based on the adjusted query content; The user's query acceptance rate is calculated based on the first and second answers, and the triggering conditions are adjusted based on the query acceptance rate.

2. The teaching Q&A method of claim 1, wherein, The step of calculating the user's behavior anomaly degree after a preset time period based on the multi-source interaction behavior data and the interaction benchmark data includes: The multi-source interactive behavior data and the interactive benchmark data are substituted into a preset behavior anomaly formula to calculate the user's behavior anomaly after a preset time period; wherein, the multi-source interactive behavior data includes multiple interactive behavior features, and the interactive benchmark data includes benchmark values ​​of multiple interactive behavior features; The behavior abnormality degree formula is: ; wherein, is the behavior anomaly degree of the user at time t after a preset time period, N is the total number of interaction behavior features, is the weight coefficient of the i-th interaction behavior feature, is the value of the i-th interaction behavior feature at time t, is the reference value of the i-th interaction behavior feature, is a preset stability constant, is a time decay coefficient, is the time interval of the i-th interaction behavior feature from time t, is the change rate of the i-th interaction behavior feature at time t.

3. The teaching question-and-answer method according to claim 1, characterized in that, The step involves determining the user's current cognitive state based on the degree of behavioral anomaly. When the current cognitive state meets a preset triggering condition, an inquiry window containing inquiry content is generated, including: The behavioral anomaly degree is matched with a preset cognitive state table to obtain the cognitive state corresponding to the behavioral anomaly degree. The cognitive state table includes multiple behavioral anomalies and their corresponding current cognitive states. The user's learning rhythm parameters are calculated based on the historical interaction behavior data, and the trigger duration threshold for each cognitive state is calculated based on the learning rhythm parameters. The duration of the user's continuous state of cognition is counted. When the user is in a preset state that requires intervention and the duration exceeds the corresponding trigger duration threshold, an inquiry window containing inquiry content is generated.

4. The teaching question-and-answer method according to claim 3, characterized in that, The step of calculating the trigger duration threshold for each cognitive state based on the learning rhythm parameter includes: Obtain the basic trigger duration corresponding to each cognitive state, compare the user's learning rhythm parameter with the average learning rhythm parameter of all users in the teaching question and answer system, and calculate the learning rhythm difference value. When the user's learning rhythm parameter is higher than the average learning rhythm parameter, the corresponding basic trigger duration is shortened according to the learning rhythm difference value. When the user's learning rhythm parameter is lower than the average learning rhythm parameter, the corresponding basic trigger duration is extended according to the learning rhythm difference value, thus obtaining the trigger duration threshold for each cognitive state.

5. The teaching question-and-answer method according to claim 1, characterized in that, The step of calculating the user's inquiry acceptance rate based on the first answer and the second answer includes: The inquiry acceptance rate is calculated using a preset inquiry acceptance rate formula; The formula for the inquiry acceptance rate is: ; in, To explore acceptance rates, and These are the acceptance weights for the corresponding options in the first and second answers, respectively. and These represent the maximum and minimum acceptance weights among all options in the first and second answers, respectively. To select the coefficient of variation, and These are the response times for the first answer and the second answer, respectively. This is the reference response time for the user in their current cognitive state. For time weighting coefficients, The cumulative number of times the user selected the "Need Help" option in the first and second answers. The total number of times the user received queries in the historical interaction data. Historical acceptance coefficient is the numerical stability constant.

6. The teaching question-and-answer method according to claim 1, characterized in that, Adjusting the triggering condition based on the inquiry acceptance rate includes: When there is a first cognitive state where the inquiry acceptance rate is lower than a preset lower threshold, the trigger duration threshold of the first cognitive state is extended according to the behavioral abnormality of the first cognitive state. When there is a second cognitive state where the inquiry acceptance rate is not lower than a preset upper limit threshold and the behavioral abnormality is lower than a preset abnormality benchmark value, the trigger duration threshold of the second cognitive state is shortened. When adjusting the trigger duration threshold of the first cognitive state or the second cognitive state, other cognitive states with similar learning parameters to the first cognitive state or the second cognitive state are identified based on the multi-source interaction behavior data, and the trigger duration threshold of each of the other cognitive states is adjusted synchronously.

7. The teaching question-and-answer method according to claim 1, characterized in that, After adjusting the triggering condition based on the inquiry acceptance rate, the method further includes: The changes in the response rate of the inquiry under each cognitive state within a preset period after statistical adjustment are analyzed, and the increase in the response rate of the inquiry before and after adjustment is calculated. When the improvement exceeds a preset effect threshold, the baseline value of the user interaction behavior feature in the interaction baseline data is updated; When the improvement is lower than the effect threshold, the learning rhythm parameter is recalculated based on the historical interaction behavior data, and the trigger duration threshold of each cognitive state is adjusted based on the recalculated learning rhythm parameter.

8. A teaching question-and-answer device, characterized in that, The teaching question-and-answer device includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the teaching question-and-answer device to perform the method as described in any one of claims 1-7.

9. A computer program product containing instructions, characterized in that, When the computer program product is run on the teaching question-and-answer device, the teaching question-and-answer device performs the method as described in any one of claims 1-7.

10. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the teaching question-and-answer device, the teaching question-and-answer device performs the method as described in any one of claims 1-7.

Images