An intelligent interaction method and system based on big data

Abstract

The present application relates to the field of human-computer interaction technology, and in particular to a method and system for intelligent interaction based on big data. The method comprises: collecting training items and operation sequences of multiple interaction rounds of users, comparing the operation sequence of each training item with the standard sequence, and obtaining the proficiency of each operation item in the standard sequence of each training item as a real-time training status; screening the interaction samples according to the real-time training status to obtain multiple target samples; calculating the probability value of each training item in the target samples; and using the training item corresponding to the maximum probability value as the training item for the next interaction round. Through the technical solution of the present application, it is possible to determine the training items for each interaction round in the human-computer interaction process and submit the training effect of the human-computer interaction.

Description

Technical Field

[0001] The present application relates to the field of human-computer interaction technology, and in particular to an intelligent interaction method and system based on big data. Background Art

[0002] With the continuous development of smart terminal technology, the interaction process between people and smart terminals has gradually received attention. At present, through human-computer interaction with smart terminals, teaching and training tasks in various scenarios such as electrician maintenance, machine tool control and building automation control can be realized.

[0003] Patent application publication number CN115248955A discloses a system for developing aircraft training cases based on principle-based visual interactive simulation. The system includes a dynamic simulation development module for aircraft system principles and characteristics, a virtual cockpit visual and interactive control module, and a simulation training case development module. This system enables flexible and cost-effective development of dynamic simulation training cases for various aircraft models, including the operating principles, interactive control logic, and actual operating states of aircraft systems. Based on data exchange between a dynamic operating principle simulation model of the aircraft system and an interactive virtual cockpit model, these training cases intuitively demonstrate the complex operating principles, control logic, and operating processes within the aircraft system. Furthermore, these training cases simulate abnormal operating states by injecting faults, enabling trainees to master the operational procedures under various fault conditions.

[0004] The above method simulates abnormal operating conditions by injecting faults, allowing trainees to strengthen their understanding of the operating principles of aircraft systems during the human-computer interaction process. However, there are many different types of faults, and the number of interaction rounds in a human-computer interaction process is limited. How to determine the training items for each interaction round during the human-computer interaction process so that trainees can quickly master various training items and ensure training effectiveness is an urgent problem to be solved. Summary of the Invention

[0005] In order to solve the technical problem of inaccurate retrieval results of similar cases, this application provides an intelligent interaction method and system based on big data, which can determine the training items for each interaction round in the human-computer interaction process and submit the training effect of human-computer interaction.

[0006] In a first aspect, the present application provides an intelligent interaction method based on big data, the interaction method comprising: collecting training items and operation sequences of multiple interaction rounds of users, the operation sequence comprising multiple operation items and operation durations, comparing the operation sequence of each training item with a standard sequence, and obtaining the proficiency of each operation item in the standard sequence of each training item as a real-time training state, wherein the proficiency is negatively correlated with the operation duration; screening interaction samples according to the real-time training state to obtain multiple target samples, the interaction samples comprising training items of each interaction round during the historical interaction process and the historical training state after each interaction round ends; calculating the probability value of each training item in the target sample, the probability value being positively correlated with the proficiency increment of each operation item after the training item is executed in the target sample under the real-time training state, and negatively correlated with the number of interaction rounds after the training item is executed in the target sample; and using the training item corresponding to the maximum probability value as the training item for the next interaction round.

[0007] During the interaction process between the user and the smart device, multiple interaction rounds will be carried out. Each interaction round includes a training item and the user's operation sequence on the smart device. The operation sequence of each training item is compared with the standard sequence, and the proficiency of each operation item in the standard sequence of each training item is obtained as the real-time training status. The real-time training status can accurately reflect the user's actual mastery of each training item; further, the probability value of each training item is calculated in the target sample. The greater the proficiency increment of each operation item after executing the training item in the target sample under the real-time training state, the better the improvement effect of the user's mastery of the training item; the smaller the number of interaction rounds after executing the training item in the target sample, the faster the user masters the training items. At this time, it means that the training effect of executing the training item in the real-time training state is better and the probability value of the training item is larger; the training item corresponding to the maximum probability value is used as the training item for the next interaction round, which can determine the training item for each interaction round in the human-computer interaction process and submit the training effect of human-computer interaction.

[0008] Preferably, the operation duration of any operation item is the time interval between the timestamp of the operation item and the timestamp of the previous adjacent operation item.

[0009] Preferably, comparing the operation sequence of each training item with the standard sequence includes: in the operation sequence and the standard sequence of any training item, judging in turn whether the operation items at each position are the same; if they are the same, setting the operation value of the operation item at the corresponding position in the standard sequence to 1, otherwise setting it to 0; calculating the difference between the relative operation time of 1 and any operation item in the standard sequence, and taking the product of the difference and the operation value as the proficiency of the operation item, and the relative operation time is the ratio of the operation time of the operation item in the operation sequence to the total operation time of the operation sequence.

[0010] The longer the relative operation time, the longer the user thinks when performing the operation item, and the lower the user's proficiency in the operation item. Therefore, proficiency is negatively correlated with operation time. The operation value can represent the correctness of executing each operation item. An operation value of 1 indicates correct execution, and an operation value of 0 indicates incorrect execution. The proficiency of incorrect execution is set to 0. The correctness of the operation item and the operation time are comprehensively considered to achieve accurate quantification of proficiency.

[0011] Preferably, before comparing the operation sequence of each training item with the standard sequence, the interaction method also includes: in response to the operation sequence being greater than the standard sequence, deleting a preset number of operation items at the end of the operation sequence; in response to the operation sequence being less than the standard sequence, adding a preset number of special symbols at the end of the operation sequence; the preset number is the difference in length between the operation sequence and the standard sequence.

[0012] When the user omits or redundantly performs an operation item in training project 1, the length of the operation sequence and the standard sequence will be different. By deleting or adding operations, the length of the operation sequence and the standard sequence are ensured to be consistent, so that each operation item in the operation sequence and the standard sequence corresponds one to one.

[0013] Preferably, obtaining the plurality of target samples comprises: calculating the similarity between the real-time training state and each historical training state in the interactive sample, and in response to the maximum similarity being greater than a similarity threshold, taking the interactive sample as the target sample.

[0014] When the maximum similarity value is greater than the similarity threshold, it means that the historical training state corresponding to the maximum similarity value has the same degree of mastery of each training item as the real-time training state, ensuring that the target sample contains historical states similar to the real-time training state to facilitate the subsequent calculation of probability values.

[0015] Preferably, the similarity is calculated using a similarity calculation method based on Euclidean distance or cosine similarity.

[0016] Preferably, calculating the probability value of each training item in the target sample includes: taking the interaction round corresponding to the historical training state with the greatest similarity to the real-time training state in the target sample as the target round; comparing the historical training state of the target round and the next adjacent interaction round to obtain the proficiency increment of each operation item in the standard sequence of each training item; taking the average value of all proficiency increments and the ratio of the number of interaction rounds after the target round in the target sample as the score value of the training item in the next adjacent interaction round in the target sample; calculating the mean score value of each training item, and taking the normalized result of the mean score value as the probability value of each training item.

[0017] The average of all proficiency increments reflects the training effect after executing the next adjacent interaction round of training items in real-time training. Therefore, the larger the average of all proficiency increments, the higher the score of the next adjacent interaction round of training items. Since the interaction sample represents a complete interaction process in which the user masters each training item, the fewer interaction rounds after the target round, the faster the user masters each training item, and the higher the score of the next adjacent interaction round of training items. Therefore, the probability value can accurately represent the training effect of executing the corresponding training item in real-time training.

[0018] Preferably, the training program Rating value for:

[0019] ; is the average of all proficiency increments, is the number of interaction rounds after the target round in the target sample.

[0020] Preferably, the training program The probability value of for:

[0021] , For training projects The mean rating value of is the sum of the mean scores of all training items.

[0022] In the second aspect of the present application, there is also provided an intelligent interaction system based on big data, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent interaction method based on big data as described in the first aspect of the present application is implemented.

[0023] The technical solution of this application has the following beneficial technical effects:

[0024] During the interaction process between the user and the smart device, multiple interaction rounds will be carried out. Each interaction round includes a training item and the user's operation sequence on the smart device. The operation sequence of each training item is compared with the standard sequence, and the proficiency of each operation item in the standard sequence of each training item is obtained as the real-time training status. The real-time training status can accurately reflect the user's actual mastery of each training item; further, the probability value of each training item is calculated in the target sample. The greater the proficiency increment of each operation item after executing the training item in the target sample under the real-time training state, the better the improvement effect of the user's mastery of the training item; the smaller the number of interaction rounds after executing the training item in the target sample, the faster the user masters the training items. At this time, it means that the training effect of executing the training item in the real-time training state is better and the probability value of the training item is larger; the training item corresponding to the maximum probability value is used as the training item for the next interaction round, which can determine the training item for each interaction round in the human-computer interaction process and submit the training effect of human-computer interaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a flowchart of an intelligent interaction method based on big data according to an embodiment of the present application.

[0026] Figure 2 This is a structural block diagram of an intelligent interactive system based on big data according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of this application.

[0028] According to the first aspect of the present application, the present application provides an intelligent interaction method based on big data, which is used to determine the training items for each interaction round during the interaction between the user and the smart device, so that the user can quickly master the content of multiple training items.

[0029] For example, the smart device can be an electrician skills training device, which is equipped with multiple training items, and one training item corresponds to a standard sequence. One training item can be regarded as a fault type, and the standard sequence is the standard operating procedure under the fault type. After executing the standard operating procedure, the fault type can be eliminated.

[0030] Figure 1 This is a flow chart of an intelligent interaction method based on big data according to an embodiment of the present application. Figure 1As shown, the intelligent interaction method based on big data includes steps S101 to S104, which are described in detail below.

[0031] S101, collecting training items and operation sequences of multiple rounds of user interaction, wherein the operation sequence includes multiple operation items and operation duration, comparing the operation sequence of each training item with the standard sequence, and obtaining the proficiency of each operation item in the standard sequence of each training item as the real-time training status, wherein the proficiency is negatively correlated with the operation duration.

[0032] In one embodiment, during the interaction between a user and a smart device, there are multiple interaction rounds. One interaction round includes a training item and a sequence of operations performed by the user on the smart device. The operation sequence includes multiple operation items and operation durations. The operation duration of any operation item is the time interval between the timestamp of the operation item and the timestamp of the previous adjacent operation item. In this way, the training items and operation sequences of each interaction round in the user interaction process are collected. Taking interaction round 1 as an example, the training item is item A. The collected operation sequence includes a total of 3 operation items, specifically: {press button 1, rotate knob 2, turn on switch 3}, and the corresponding operation durations are {t1, t2, t3}.

[0033] It can be understood that the operation items include clicking a button, adjusting a lever angle, starting a device, etc.; the training items of the first interactive round can be pre-set.

[0034] In one embodiment, for any training item in multiple interaction rounds, the operation sequence and standard sequence of the training item can be obtained. Specifically, comparing the operation sequence and standard sequence of each training item includes: in the operation sequence and standard sequence of any training item, determining whether the operation items at each position are the same in turn; if they are the same, setting the operation value of the operation item at the corresponding position in the standard sequence to 1, otherwise setting it to 0; calculating the difference between the relative operation time of 1 and any operation item in the standard sequence, and taking the product of the difference and the operation value as the proficiency of the operation item, wherein the relative operation time is the ratio of the operation time of the operation item in the operation sequence to the total operation time of the operation sequence.

[0035] It is understandable that when the user makes an error in an operation item in training project 1, the length of the operation sequence and the standard sequence are the same; when the user omits an operation item in training project 1, the length of the operation sequence will be shorter than the standard sequence. In this case, it is necessary to add a special symbol at the end of the operation sequence to make the operation sequence and the standard sequence the same length, wherein the special symbol is any symbol different from the operation item; when the user makes a redundant operation item in training project 1, the length of the operation sequence will be longer than the standard sequence. In this case, it is necessary to delete multiple operation items at the end of the operation sequence to make the operation sequence and the standard sequence the same length. Therefore, before comparing the operation sequence and the standard sequence of each training item, the interaction method further includes: in response to the operation sequence being longer than the standard sequence, deleting a preset number of operation items at the end of the operation sequence; in response to the operation sequence being shorter than the standard sequence, adding a preset number of special symbols at the end of the operation sequence; the preset number is the difference in length between the operation sequence and the standard sequence.

[0036] By "deleting a preset number of operation items at the end of the operation sequence" or "adding a preset number of special symbols at the end of the operation sequence", the length of the operation sequence and the standard sequence can be guaranteed to be the same, and the operation items in the operation sequence and the operation items in the standard sequence correspond one to one. In this way, the relative operation time of any operation item in the standard sequence can be calculated. The longer the relative operation time, the longer the user thinks when executing the operation item, and the lower the user's proficiency in the operation item. Therefore, proficiency is negatively correlated with operation time; the operation value can represent the correctness of executing each operation item. The operation value of 1 indicates correct execution, and the operation value of 0 indicates incorrect execution. The proficiency of incorrect execution is set to 0, and the correctness of the operation item and the operation time are comprehensively considered to achieve accurate quantification of proficiency.

[0037] It can be understood that the operation items in the operation sequence correspond one-to-one to the operation items in the standard sequence. If an operation item in the standard sequence corresponds to a special symbol, the relative operation duration of the operation item is set to 0 because the operation duration corresponding to the special symbol cannot be found.

[0038] In this way, during the interaction between the user and the smart device, the training items and operation sequences of each interaction round are analyzed to obtain the proficiency of each operation item in the standard sequence of each training item, accurately quantify the user's actual mastery of each training item after the current interaction round, and obtain the user's real-time training status.

[0039] S102 , screening the interaction samples according to the real-time training status to obtain a plurality of target samples, wherein the interaction samples include training items of each interaction round in the historical interaction process and the historical training status after each interaction round ends.

[0040] In one embodiment, the interaction sample includes the training items for each interaction round during the historical interaction process and the historical training status after each interaction round. During the historical interaction process, the user and the smart device generate multiple interaction rounds. After each interaction round, a historical training status for that interaction round is obtained. This historical training status can represent the user's mastery of each training item after the end of the interaction round. The user continues to interact with multiple rounds until the user masters each training item. The training items for each interaction round and the historical training status after each interaction round are stored to obtain an interaction sample.

[0041] Specifically, obtaining the plurality of target samples includes: calculating the similarity between the real-time training state and each historical training state in the interaction sample, and in response to a maximum similarity being greater than a similarity threshold, taking the interaction sample as the target sample.

[0042] The similarity may be calculated using a similarity calculation method based on Euclidean distance or cosine similarity, and the similarity threshold is 0.8. When the maximum similarity is greater than the similarity threshold, it indicates that the historical training state corresponding to the maximum similarity has the same degree of mastery of each training item as the real-time training state.

[0043] In this way, the real-time training state is determined based on the training items and operation sequences of multiple interaction rounds of the user; the interaction sample includes the training items of each interaction round in a complete historical interaction process and the historical training state after the end of each interaction round; and multiple target samples are determined based on the similarity between the real-time training state and the historical training state.

[0044] S103, calculating the probability value of each training item in the target sample, wherein the probability value is positively correlated with the proficiency increment of each operation item after the training item is executed in the real-time training state in the target sample, and is negatively correlated with the number of interaction rounds after the training item is executed in the target sample.

[0045] In one embodiment, calculating the probability value of each training item in the target sample includes:

[0046] The interaction round corresponding to the historical training state with the greatest similarity to the real-time training state in the target sample is taken as the target round; the historical training state of the target round and the next adjacent interaction round are compared to obtain the proficiency increment of each operation item in the standard sequence of each training item; the average value of all proficiency increments and the ratio of the number of interaction rounds after the target round in the target sample are taken as the score value of the training item in the next adjacent interaction round in the target sample; the mean score value of each training item is calculated, and the normalized result of the mean score value is used as the probability value of each training item.

[0047] The historical training status includes the proficiency of each operation item in the standard sequence of each training item; the number of training items is recorded as Since the number of operation items in each training project is different, the training project The number of operation items contained in ; then training items Can get The proficiency increment of the operation item; The mean of the proficiency increments of the operation items is used as the training item The proficiency increments of all training items are further calculated to obtain the average value of all proficiency increments. The average value of all proficiency increments can reflect the training effect after executing the training items of the next adjacent interactive round in the real-time training state. Therefore, the larger the average value of all proficiency increments, the higher the score value of the training items of the next adjacent interactive round.

[0048] Since the interaction sample is a complete interaction process in which the user can master each training item, the fewer the number of interaction rounds after the target round, the faster the user masters each training item, and the higher the score value of the training item in the next adjacent interaction round.

[0049] Specifically, the training program Rating value for:

[0050] ; is the average of all proficiency increments, is the number of interaction rounds after the target round in the target sample.

[0051] It can be understood that the score value of each training item can be obtained based on multiple target samples, and there may be a situation where one training item corresponds to multiple score values. Therefore, the average score value of each training item is calculated and the average score value is normalized to obtain the probability value of each training item. The larger the probability value, the better the training effect of executing the corresponding training item in the real-time training state.

[0052] Specifically, the training program The probability value of for:

[0053] , For training projects The mean rating value of is the sum of the mean scores of all training items.

[0054] S104: The training item corresponding to the maximum probability value is used as the training item for the next interaction round.

[0055] In one embodiment, the larger the probability value, the better the training effect of executing the corresponding training item in the real-time training state. Therefore, the training item corresponding to the maximum probability value is used as the training item for the next interaction round to complete the user's next interaction round.

[0056] In this way, after each interaction round is completed, the user's real-time training status can be calculated, and the real-time training status can reflect the user's mastery of each training item; based on the interaction sample, the training item with the best training effect in the real-time training state is determined, and the training item is used as the training item for the next interaction round, realizing human-computer interaction between the user and the smart device, and submitting the training effect of human-computer interaction.

[0057] According to the second aspect of the present application, the present application also provides an intelligent interaction system based on big data. Figure 2 This is a block diagram of a big data-based intelligent interaction system according to an embodiment of the present application. Figure 2 As shown, the system 50 includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, the intelligent interaction method based on big data according to the first aspect of the present application is implemented. The system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are well known in the art and are therefore not described in detail here.

[0058] It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present application, and these all fall within the scope of protection of the present application.

Claims

1. An intelligent interaction method based on big data, characterized in that: The interaction method includes: Collect training items and operation sequences from multiple rounds of user interaction, where the operation sequence includes multiple operation items and operation durations. Compare the operation sequence of each training item with the standard sequence to obtain the proficiency of each operation item in the standard sequence of each training item as the real-time training status. The proficiency is negatively correlated with the operation duration. The interaction samples are screened according to the real-time training status to obtain multiple target samples, wherein the interaction samples include the training items of each interaction round in the historical interaction process and the historical training status after each interaction round ends; Calculating a probability value for each training item in the target sample, wherein the probability value is positively correlated with the proficiency increment of each operation item after the target sample executes the training item in the real-time training state, and is negatively correlated with the number of interaction rounds after the target sample executes the training item; The training item corresponding to the maximum probability value is used as the training item for the next interaction round.

2. The intelligent interaction method based on big data according to claim 1, characterized in that: The operation duration of any operation item is the time interval between the timestamp of the operation item and the timestamp of the previous adjacent operation item.

3. The intelligent interaction method based on big data according to claim 1, characterized in that: Comparison of the operation sequence and standard sequence of each training item includes: In the operation sequence and standard sequence of any training item, determine in turn whether the operation items at each position are the same. If they are the same, set the operation value of the operation item at the corresponding position in the standard sequence to 1, otherwise it is set to 0; calculate the difference in the relative operation time between 1 and any operation item in the standard sequence, and use the product of the difference and the operation value as the proficiency of the operation item. The relative operation time is the ratio of the operation time of the operation item in the operation sequence to the total operation time of the operation sequence.

4. The intelligent interaction method based on big data according to claim 3, characterized in that: Before comparing the operation sequence of each training item with the standard sequence, the interactive method further includes: In response to the operation sequence being greater than the standard sequence, deleting a preset number of operation items at the end of the operation sequence; In response to the operation sequence being shorter than the standard sequence, a preset number of special symbols is added to the end of the operation sequence; the preset number is the difference between the lengths of the operation sequence and the standard sequence.

5. The intelligent interaction method based on big data according to claim 1, characterized in that: Get multiple target samples including: The similarity between the real-time training state and each historical training state in the interaction sample is calculated, and in response to the maximum similarity being greater than a similarity threshold, the interaction sample is used as a target sample.

6. The intelligent interaction method based on big data according to claim 5, characterized in that: The similarity is calculated using a similarity calculation method based on Euclidean distance or cosine similarity.

7. The intelligent interaction method based on big data according to claim 1, characterized in that: Calculating the probability value of each training item in the target sample includes: The interaction round corresponding to the historical training state with the greatest similarity to the real-time training state in the target sample is taken as the target round; Compare the historical training status of the target round and the next adjacent interaction round to obtain the proficiency increment of each operation item in the standard sequence of each training item; The ratio of the average value of all proficiency increments to the number of interaction rounds after the target round in the target sample is used as the score value of the training item in the next adjacent interaction round in the target sample; Calculate the mean score of each training item, and use the normalized result of the mean score as the probability value of each training item.

8. The intelligent interaction method based on big data according to claim 7, characterized in that: Training Program Rating value for: ; is the average of all proficiency increments, is the number of interaction rounds after the target round in the target sample.

9. The intelligent interaction method based on big data according to claim 7, characterized in that: Training Program The probability value of for: , For training projects The mean rating value of is the sum of the mean scores of all training items.

10. An intelligent interactive system based on big data, characterized in that: The invention comprises a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent interaction method based on big data according to any one of claims 1 to 9 is implemented.

Images