Adjustment Method and Device for Course Information Based on Natural Language Processing

Abstract

A method and device for adjusting course information based on natural language processing, belonging to the technical field of course information adjustment. According to the numerical dispersion amplitude of any pair of types of past course data packet queues, the relevant importance of any pair of types of past course data packet queues at each storage time point is obtained; the numerical coherence of all course data packet queues of any pair of types is obtained; the numerical smooth amplitude of all course data packet queues of each type itself is obtained; the overall effect amplitude is used to obtain the final grouping of the most recent course data packet queue and all course data packet queues of all types; encoding is performed on all the most recent course data packets according to the final grouping. The present invention involves the coherence between different types of past course data packets and the influence of the most recent course data packets on the past course data packets, obtains a grouping that accurately reflects the approximate numerical arrangement, and thus obtains an accurate course data packet storage effect.

Description

Technical Field

[0001] The present invention belongs to the technical field of adjustment of course information, and particularly relates to a method and device for adjusting course information based on natural language processing. Background Art

[0002] The teaching completion degree is the progress of the teaching plan. Currently, it is often necessary to use the existing technical solution with the patent publication number "CN118279105A" to process course information based on natural language until the teaching completion degree corresponding to the course is formed. Among them, the method for processing course information based on natural language is to form a course data packet and store it in a database, and obtain an identification model through models such as a natural language model (the models such as the natural language model include a training vector machine, a natural language model, and an image model), and perform identification on the course data packet in the database to obtain a corresponding completion state vector. The course data packet contains manually sorted courses and corresponding teaching images, teaching audio, and classroom personnel coordinates.

[0003] During the process of forming the course data packet and storing it in the database, to improve the efficiency of its storage, encoding and reduction storage can be performed on several types of formed course data packets through an encoding method such as the DEFLATE method, which can well improve the storage efficiency and reduce the storage load of the database.

[0004] To improve the efficiency of course data packet encoding, several types of formed course data packets are often grouped and cut through a grouping method, and encoding and reduction storage are performed on the groups. Currently, when adding newly formed recent course data packets (that is, newly formed course data packets) to the groups formed by the existing previous course data packets (that is, previously stored course data packets), the influence of the recent course data packets on the previous course data packets is not taken into account, which often damages the existing groups and destroys the course data packets, resulting in incomplete encoded data, and thus the accurate storage effect of the course data packets cannot be obtained. Summary of the Invention

[0005] To solve the defects in the prior art, the present invention proposes an adjustment method and device for course information based on natural language processing, obtaining the past course data packet queues at each storage time point in the past period and the most recent course data packet queue to be stored; according to the numerical dispersion amplitude of any pair of types of past course data packet queues, obtaining the correlation importance of any pair of types of past course data packet queues at each storage time point, reflecting the stability of the data of any pair of types of past course data packet queues at each storage time point; adjusting the numerical fluctuation condition difference quantity of any pair of past course data packet queues at each storage time point through the correlation importance, obtaining the numerical correlation degree of all course data packet queues of any pair of types; obtaining the numerical smooth amplitude of all course data packet queues of each type; obtaining the overall influence amplitude of the most recent course data packet queue of all types on all course data packet queues of all types, and using the overall influence amplitude to obtain the final grouping of the most recent course data packet queue and all course data packet queues of all types; encoding all the most recent course data packets according to the final grouping. The present invention involves the correlation degree between different types of past course data packets and the influence of the most recent course data packet on the past course data packet, obtaining a grouping that accurately reflects the approximate numerical arrangement, thereby obtaining an accurate course data packet storage effect.

[0006] The present invention adopts the following technical solutions.

[0007] An adjustment method for course information based on natural language processing includes:

[0008] Based on natural language processing of course information until the teaching completion degree corresponding to the course is formed, wherein the method of natural language processing of course information includes forming course data packets and storing them in a database;

[0009] During the process of forming course data packets and storing them in a database, it further includes:

[0010] Step1, obtaining the most recent course data packet queues of all types to be stored and the past course data packet queues of each type at each storage time point within a preset past period; the past course data packet queues of each type at all storage time points form all course data packet queues of each type;

[0011] Step2, according to the numerical dispersion amplitude of any pair of types of past course data packet queues, until obtaining the numerical correlation degree of any pair of types of course data packet queues;

[0012] Step3, according to the numerical correlation degree between all course data packet queues of each type and all course data packet queues of each other type, until obtaining the overall influence amplitude of the most recent course data packet queue of all types on all course data packet queues of all types;

[0013] Step 4. Obtain the last grouping of the nearest course data packet queue and the overall course data packet queues of all types according to the overall action amplitude; perform encoding on all the nearest course data packets according to the last grouping and store them.

[0014] Further, in Step 1, transform the text information in the obtained course data packets into attribute vectors via the Word2Vec model, and process all the obtained attribute vectors by the ridge regression method.

[0015] In Step 2, the method for obtaining the numerical correlation degree of any pair of types of course data packet queues according to the numerical dispersion amplitude of the previous course data packet queues of any pair of types until the numerical correlation degree of any pair of types of course data packet queues is obtained includes: obtaining the correlation importance of any pair of types of previous course data packet queues at each storage time point according to the numerical dispersion amplitude of the previous course data packet queues of any pair of types; obtaining the numerical correlation degree of any pair of types of course data packet queues according to the numerical fluctuation condition difference quantity and the correlation importance of any pair of types of previous course data packet queues at each storage time point.

[0016] In Step 3, the method for obtaining the overall action amplitude of the nearest course data packet queues of all types on the overall course data packet queues of all types according to the numerical correlation degree between the overall course data packet queues of each type and the overall course data packet queues of each other type until the overall action amplitude is obtained includes: obtaining the numerical smooth amplitude of the overall course data packet queues of each type according to the numerical correlation degree between the overall course data packet queues of each type and the overall course data packet queues of each other type, and the numerical dispersion amplitude of the overall course data packet queues of each type; obtaining the overall correlation amplitude between the nearest course data packet queues of all types and the overall course data packet queues of all types according to the numerical reduction attribute between the nearest course data packet queues and the overall course data packet queues of the corresponding type, and the numerical smooth amplitude of the overall course data packet queues of each type; obtaining the overall action amplitude of the nearest course data packet queues of all types on the overall course data packet queues of all types according to the numerical definition difference quantity and the overall correlation amplitude between the nearest course data packet queues of all types and the overall course data packet queues of all types.

[0017] Further, in Step 2, the method for obtaining the numerical dispersion amplitude specifically includes:

[0018] In the previous course data packet queues of each type, sum up the difference quantity between the value of each numerical point and the average of all numerical points in the previous course data packet queues of each type to obtain the numerical dispersion amplitude of the previous course data packet queues of each type at each storage time point in the previous course data packet queues of each type.

[0019] Further, in Step2, the operation equation for the numerical dispersion amplitude is:

[0020]

[0021] In the equation, represents the numerical dispersion amplitude of the previous course data packet queue of the th category; represents the number of numerical values of the previous course data packet queue of the th category at the th storage time point; represents the value of the th numerical point in the previous course data packet queue of the th category at the th storage time point; represents the mean of the numerical values in the previous course data packet queue of the th category.

[0022] Further, in Step2, the method for obtaining the difference quantity of the numerical fluctuation condition specifically includes:

[0023] Perform inverse allocation and standardization processing on the difference quantity of the numerical dispersion amplitude of any pair of previous course data packets of different categories, and obtain the correlation importance of any pair of previous course data packet queues of different categories at each storage time point.

[0024] Further, in Step2, the correlation importance operation equation is:

[0025]

[0026] In the equation, represents the correlation importance between the previous course data packet queue of the th category and the previous course data packet queue of the th category at the th storage time point; represents the numerical dispersion amplitude of the previous course data packet queue of the th category; represents the numerical dispersion amplitude of the previous course data packet queue of the th category; represents the standardization processing of using the Z-score method.

[0027] Further, in Step2, the method for obtaining the difference quantity of the numerical fluctuation situation specifically includes:

[0028] Construct the regression line of the historical course data packets using the least squares method for each queue of historical course data packets of each type within each storage time point; use the LCSS method to fit the regression lines of any pair of historical course data packets of each type within each storage time point to obtain the overall numerical fitting clusters of any pair of historical course data packet queues at each storage time point; sum up the derivative difference amounts between a pair of numerical points in the overall numerical fitting clusters at each storage time point to obtain the numerical fluctuation condition difference amount of any pair of historical course data packet queues at each storage time point.

[0029] Further, in Step2, the method for obtaining the numerical correlation degree specifically includes:

[0030] Under any pair of types, take the product of the correlation importance and the numerical fluctuation condition difference amount between the historical course data packet queues at each storage time point as the first multiplication quantity, and perform inverse allocation and standardization processing on the total sum of the first multiplication quantities at all storage time points to obtain the numerical correlation degree between all course data packet queues of any pair of types.

[0031] Further, in Step2, the numerical correlation degree operation equation is:

[0032]

[0033] In the equation, represents the numerical correlation degree between all course data packet queues of the th type and all course data packet queues of the th type; represents the number of storage time points in the preset historical period; represents the correlation importance between the historical course data packet queue of the th type and the historical course data packet queue of the th type at the th storage time point; represents the number of numerical fitting clusters between the historical course data packet queue of the th type and the historical course data packet queue of the th type at the th storage time point; represents the derivative value of the numerical point of the historical course data packet queue of the th type on its regression line within the th numerical fitting cluster at the th storage time point; represents the derivative value of the numerical point of the historical course data packet queue of the th type on its regression line within the th numerical fitting cluster at the The derivative value of the numerical points within a numerical adaptation cluster on its regression line; is the Euler number.

[0034] Furthermore, in Step3, the method for obtaining the numerical smoothness amplitude specifically includes:

[0035] Arbitrarily select a type as the target type; use the LCSS method to obtain all the adapted course data packets composed of all the adaptation clusters on the entire course data packet queue between each other type and the target type, and then execute the following equation:

[0036]

[0037] In the equation, represents the target type; represents the numerical smoothness amplitude of the entire course data packet queue of the target type; represents the number of values in the entire course data packet queue of the target type; represents the number of other types; represents the numerical coherence degree between the entire course data packet queue of the target type and the th other type; represents the th smooth coherence amplitude of the entire course data packet queue of the th other type facing the entire course data packet queue of the target type; represents the value of the th numerical point in the entire course data packet queue of the target type; represents the average of all the numerical points in the entire course data packet queue of the target type; represents the th number of adapted course data packets in the adaptation cluster where the th numerical point in the entire course data packet queue of the target type is located and the entire course data packet queue of the other type; represents the th value of the th adapted course data packet in the adaptation cluster where the th numerical point in the entire course data packet queue of the target type is located and the entire course data packet queue of the other type; represents the th average of all the numerical points in the entire course data packet queue of the other type; represents performing standardization processing on using the Z-score method; is the Euler number.

[0038] Further, in Step 3, the method for obtaining the overall coherence amplitude specifically includes:

[0039] In each of the nearest course data packet queues, sum up the minimum of the difference amounts between each numerical point and the corresponding type of course data packet queue to obtain the numerical reduction amplitude between each nearest course data packet queue and each type of course data packet queue; sum up the amounts obtained by multiplying the numerical smoothing amplitude and the numerical reduction amplitude of all types of course data packet queues in the preset previous period to obtain the overall coherence amplitude between all types of nearest course data packet queues and all types of all course data packet queues.

[0040] Further, in Step 3, the operation equation for the overall coherence amplitude is:

[0041]

[0042] In the equation, represents the overall coherence amplitude between all types of nearest course data packet queues and all types of all course data packet queues; represents the number of types of all course data packet queues; represents the numerical smoothing amplitude of all course data packet queues of the th type; represents the number of values of the th nearest course data packet queue; represents the th data of the

[0043] th nearest course data packet queue and the minimum of the difference amount between the corresponding type of all course data packet queues.

[0044] Take the quotient obtained by dividing the number of values in all types of nearest course data packet queues by the total number of values in all types of all course data packet queues as the numerical definition difference amount between the nearest course data packet queue and all types of course data packet queues.

[0045] Further, in Step 3, the method for obtaining the overall effect amplitude specifically includes:

[0046] Normalize the amount obtained by multiplying the numerical definition difference amount and the overall coherence amplitude using the Z - score method to obtain the overall effect amplitude of the nearest course data packet queue on all types of all course data packet queues.

[0047] Further, in Step 3, the operation equation for the overall effect amplitude is:

[0048]

[0049] In the equation, represents the overall amplitude of the action of the nearest course data packet queue of all types facing the overall course data packet queue of all types; represents the number of values of the nearest course data packet queue of all types; represents the total number of values of the overall course data packet queue of all types; represents the overall coherence amplitude between the nearest course data packet queue of all types and the overall course data packet queue of all types; represents the use of the Z-score method for performing standardization processing.

[0050] Further, in Step4, according to the overall amplitude of the action, a method for grouping the nearest course data packet queue is as follows:

[0051] Group all the previous course data packets within the preset previous time period through the fuzzy C-means grouping method to obtain the source grouping; preset a critical quantity one, and cut all the nearest course data packet queues with an overall amplitude lower than the critical quantity one into the source grouping through the DBSCAN method to obtain the final grouping; directly group all the nearest course data packet queues with an overall amplitude higher than the critical quantity one to obtain all the latest groupings of the nearest course data packet queues; regard all the latest groupings and the source grouping as the final grouping.

[0052] Further, in Step4, the RLE method has a good coding reduction effect on redundant values, and encodes and stores all the values in the final grouping through the RLE method.

[0053] An adjustment device for course information based on natural language processing, including:

[0054] A formation module, which is used to obtain the nearest course data packet queue of all types to be stored and the previous course data packet queues of each type at each storage time point within the preset previous time period; the previous course data packet queues of each type at all storage time points form the overall course data packet queue of each type;

[0055] A coherence module, which is used to calculate the numerical dispersion amplitude of the previous course data packet queues of any pair of types until the numerical coherence degree of the course data packet queues of any pair of types is obtained;

[0056] An action module, which is used to obtain the overall action amplitude of all types of recent course data packet queues on all types of all course data packet queues according to the numerical correlation between all course data packet queues of each type and all course data packet queues of each other type until the nearest course data packet queue of all types is obtained;

[0057] A storage module, which is used to obtain the final grouping of the nearest course data packet queue and all types of all course data packet queues according to the overall action amplitude; encode and store all the nearest course data packets according to the final grouping.

[0058] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0059] Obtain the past course data packet queues at each storage time point in the past period and the nearest course data packet queue to be stored; because the correlation between course data packet queues of different types will have a great impact on the grouping and cutting of all types of course data packet queues, and the past course data packet queues at different storage time points often change with the change of the storage time point, so according to the numerical dispersion amplitude of any pair of types of past course data packet queues, obtain the correlation importance of any pair of types of past course data packet queues at each storage time point, which reflects the stability of the data of any pair of types of past course data packet queues at each storage time point; adjust the numerical fluctuation condition difference quantity of any pair of past course data packet queues at each storage time point through the correlation importance to obtain the numerical correlation between all course data packet queues of any pair of types; because the numerical dispersion amplitude of all course data packet queues of one type often has a great impact on the data stability amplitude of all course data packet queues of other types, so obtain the numerical smooth amplitude of all course data packet queues of each type; in addition, because the nearest course data packet queue will have a certain amount of influence on all types of all course data packet queues, which involves the existing grouping of all past course data packets, so obtain the overall action amplitude of all types of the nearest course data packet queue on all types of all course data packet queues, use the overall action amplitude to obtain the final grouping of the nearest course data packet queue and all types of all course data packet queues; encode all the nearest course data packets according to the final grouping. The present invention involves the correlation between past course data packet queues of different types and the influence of the nearest course data packet queue on past course data packets, obtains a grouping that accurately reflects the approximate numerical arrangement, and thus obtains an accurate course data packet storage effect. Description of the Drawings

[0060] Figure 1 is a flowchart of the adjustment method for course information based on natural language processing described in the present invention;

[0061] Figure 2It is a partial structural diagram of the adjustment device for course information based on natural language processing described in the present invention. Detailed implementation manners

[0062] To improve the efficiency of course data packet encoding, several types of course data packets formed by a grouping method are often subjected to packet cutting, and encoding reduction storage is performed on the packets. Currently, when adding newly formed recent course data packets (i.e., newly formed course data packets) into the packets composed of existing past course data packets (i.e., course data packets stored in the past), the impact of the recent course data packets on the past course data packets is not considered, which often damages the existing packets and destroys the course data packets, making the encoded data incomplete, and thus an accurate course data packet storage effect cannot be obtained. The present invention relates to the relevance between different types of past course data packets and the impact of recent course data packets on past course data packets, and obtains packets that accurately represent the approximate numerical arrangement, so as to obtain an accurate course data packet storage effect.

[0063] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. The embodiments described in this application are only some embodiments of the present invention, not all embodiments. According to the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the protection scope of the present invention.

[0064] As Figure 1 shown, a method for adjusting course information based on natural language processing described in the present invention includes:

[0065] Based on natural language processing of course information until the teaching completion degree corresponding to the course is formed. The method of processing course information based on natural language is to form course data packets and store them in a database, obtain an identification model through models such as natural language models (models such as natural language models include training vector machines, natural language models, and image models), and perform identification on the course data packets in the database to obtain corresponding completion state vectors. The course data packets include manually sorted courses and corresponding teaching images, teaching audio, and classroom personnel coordinates; obtaining an identification model through models such as natural language models (models such as natural language models include training vector machines, natural language models, and image models), and performing identification on the course data packets in the database to obtain corresponding completion state vectors until the method of forming the teaching completion degree corresponding to the course can be achieved by the corresponding method in the patent publication number "CN118279105A". The types of course data packets are several types such as manually sorted courses and corresponding teaching images, teaching audio, and classroom personnel coordinates.

[0066] During the formation of the course data packet and its storage in the database, it also includes:

[0067] Step1: Obtain the queue of the most recent course data packets of all types to be stored and the queues of the historical course data packets of each type at each storage time point within a preset historical period; the queues of the historical course data packets of each type at all storage time points form the queues of all course data packets of each type; the queue of the most recent course data packets is the queue formed by arranging the most recent course data packets in the order of their formation. The most recent course data packets are the newly formed course data packets, and the historical period is the previous period. The queue of the historical course data packets is the queue formed by arranging the historical course data packets in the order of their storage. The historical course data packets are the course data packets that have been stored within the preset historical period.

[0068] In the method for adjusting course information based on natural language processing of the present application, when storing the formed course data packet, first obtain the formed course data packet. Since the course data includes several types such as manually sorted courses and corresponding teaching images, teaching audio, and classroom personnel coordinates, all types of course data packets are obtained. In specific situations, as long as a certain number of historical course data packets are formed, a course data packet storage will be performed once. Therefore, a certain number of historical course data packets formed each time form the queues of historical course data packets at each storage time point. Since the newly formed course data packets will be stored subsequently, in the present application, the queues of all types of historical course data packets at each storage time point and the queue of the most recent course data packets to be stored are obtained, and the queues of all types of historical course data packets at all storage time points in the preset historical period form the queues of all course data packets of each type.

[0069] In a preferred but non-limiting embodiment of the present invention, in Step1, since the quantity of the type numbers and numerical values of the formed historical course data packets and the most recent course data packets is very large, directly performing subsequent analysis will often result in an imbalance in the historical course data packets when performing subsequent grouping of different types of historical course data packets, thus resulting in errors. Therefore, the text information in the obtained course data packets is transformed into attribute vectors through the Word2Vec model, and the obtained all attribute vectors are processed by the ridge regression method to centrally transform the information amounts of several types of course data packets into a length-normalized structure for subsequent processing.

[0070] The size of the historical period can be set to one day.

[0071] Thus, the queues of all types of historical course data packets at each storage time point and the queue of the most recent course data packets to be stored are obtained.

[0072] Step 2. Obtain the relevant importance of the historical course data packet queues of any two types at each storage time point according to the value dispersion amplitude of the historical course data packet queues of any two types; obtain the numerical correlation degree of the historical course data packet queues of any two types according to the difference quantity and the relevant importance of the historical course data packet queues of any two types at each storage time point.

[0073] There is often some correlation between historical course data packets of different types. For example, the larger the number of teaching images, the generally larger the number of teaching audio. And the correlation between historical course data packets of different types will have an impact on the grouping and cutting of the overall course data packet queues of all types, and there will often be an imbalance in the grouping and cutting. Therefore, in this application, it is necessary to obtain the numerical correlation degree of the historical course data packet queues of any two types at each storage time point.

[0074] The historical course data packet queues at different storage time points often show some changes with the change of the storage time point. Therefore, it is necessary to perform separate analysis on the historical course data packet queues at each storage time point. For the historical course data packet queues at different storage time points, the value dispersion amplitude of the historical course data packet queues of each type reflects its stability. And the higher the stability of the historical course data packet queue, the closer the correlation degree between calculating the historical course data packet queue of this type and the historical course data packet queues of other types. Therefore, in this application, according to the value dispersion amplitude of the historical course data packet queues of any two types, obtain the relevant importance of the historical course data packet queues of any two types at each storage time point.

[0075] In a preferred but non-limiting embodiment of the present invention, in Step 2, the method for obtaining the value dispersion amplitude specifically includes:

[0076] In the historical course data packet queues of each type, sum up the difference quantity between the value of each numerical point and the average of all numerical points in the historical course data packet queues of each type to obtain the value dispersion amplitude of the historical course data packet queues of each type at each storage time point. In the historical course data packet queues of each type, the higher the difference quantity between the value of each numerical point and the average of all numerical point values, the more dispersed each historical course data packet is relative to the whole of the historical course data packet queue. A numerical point is a corresponding course data packet (here it is a historical course data packet), and the value of the numerical point is the specific value of the binary value of the numerical point.

[0077] In a preferred but non-limiting embodiment of the present invention, in Step 2, the operation equation of the value dispersion amplitude is:

[0078]

[0079] Within the equation, represents the numerical dispersion amplitude of the previous course data packet queue of the th category; represents the number of values of the previous course data packet queue of the th storage time point for the th category; represents the value of the th numerical point within the previous course data packet queue of the th category at the th storage time point; represents the mean of the values within the previous course data packet queue of the th category (i.e., the mean of the values of all numerical points in this queue).

[0080] In the operation equation of the numerical dispersion amplitude, the higher the difference between each numerical point and the mean of the values of all numerical points in the previous course data packet queue of the th category, it means that each numerical point is more dispersed relative to the entire previous course data packet queue of the th category. Summing up the differences between each numerical point and the mean of the values of all numerical points, the higher the total sum, it means that the numerical dispersion amplitude of the previous course data packet queue of the th category is higher.

[0081] In a preferred but non - limiting embodiment of the present invention, in Step2, the method for obtaining the coherence importance specifically includes:

[0082] Performing inverse allocation and normalization on the difference in the numerical dispersion amplitude between any pair of previous course data packet categories to obtain the coherence importance between the previous course data packet queues of any pair of categories at each storage time point.

[0083] In a preferred but non - limiting embodiment of the present invention, in Step2, the coherence importance operation equation is:

[0084]

[0085] Within the equation, represents the coherence importance between the previous course data packet queue of the th category and the previous course data packet queue of the th category at the th storage time point; represents the numerical dispersion amplitude of the previous course data packet queue of the th category; represents the numerical dispersion amplitude of the previous course data packet queue of the th category; Represent the use of the Z-score method for Perform standardization processing.

[0086] In the relevant importance calculation equation, the lower the difference in the magnitude of the numerical dispersion between a pair of past course data packet queues means that the th type of past course data packet queue and the th type of past course data packet queue have a more similar numerical arrangement at the th storage time point, and the th storage time point, the th type of past course data packet queue and the th type of past course data packet queue have a higher degree of correlation, and the relevant importance is greater.

[0087] In a preferred but non-limiting embodiment of the present invention, in Step2, the method for obtaining the difference in the numerical fluctuation situation specifically includes:

[0088] Construct a regression line of the past course data packet for each type of past course data packet queue in each storage time point by using the least squares method (the abscissa is the storage time point, and the ordinate is the value of the course data packet); use the LCSS method to perform adaptation on the regression lines of any pair of types of past course data packets in each storage time point (adaptation means that each point on one regression line in the pair of regression lines obtains an adapted point on the other regression line), and obtain the overall numerical adaptation cluster of any pair of types of past course data packet queues in each storage time point (a numerical adaptation cluster is two points on a pair of adapted regression lines); sum up the derivative difference amount between a pair of numerical points in the overall numerical adaptation cluster of each storage time point (that is, the absolute value of the amount obtained by subtracting the derivative values of a pair of numerical points on their respective regression lines), and obtain the difference in the numerical fluctuation situation between any pair of types of past course data packet queues. The higher the difference in the numerical fluctuation situation, the lower the degree of correlation between any pair of types of past course data packet queues.

[0089] In a preferred but non-limiting embodiment of the present invention, in Step2, the method for obtaining the numerical correlation specifically includes:

[0090] Under any pair of types, obtain the product of the relevant importance and the difference in the numerical fluctuation situation between the past course data packet queues in each storage time point as the first product quantity, and perform inverse distribution and standardization processing on the total amount of the first product quantity of all storage time points to obtain the numerical correlation between all course data packet queues of any pair of types.

[0091] In a preferred but non-limiting embodiment of the present invention, in Step2, the numerical correlation calculation equation is:

[0092]

[0093] In the equation, represents the numerical coherence degree between the entire course data packet queue of the th category and the entire course data packet queue of the th category; represents the number of storage time points in a preset past period; represents the th storage time point, and the coherence importance degree between the past course data packet queue of the th category and the past course data packet queue of the th category; represents the th storage time point, and the number of numerical adaptation clusters between the past course data packet queue of the th category and the past course data packet queue of the th category; represents the derivative value of the numerical points of the past course data packet queue of the th category at the th storage time point on its regression line within the th numerical adaptation cluster; represents the derivative value of the numerical points of the past course data packet queue of the th category at the th storage time point on its regression line within the th numerical adaptation cluster; represents the difference quantity of the numerical fluctuations between the past course data packet queue of the th category and the past course data packet queue of the th category at the th storage time point; is the Euler number.

[0094] In the numerical coherence degree operation equation, the lower the derivative difference quantity between a pair of numerical points in each numerical adaptation cluster between the past course data packet queue of the th category and the past course data packet queue of the th category, the closer the numerical fluctuations between each numerical adaptation cluster are. By summing up the numerical fluctuation difference quantities of each numerical adaptation cluster between the past course data packet queue of the th storage time point, the th category, and the past course data packet queue of the th category, the total numerical fluctuation difference quantity of each numerical adaptation cluster is obtained for the past course data packet queue of the th storage time point, the th category, and the past course data packet queue of the The difference in the numerical fluctuation status between the previous course data packet queues of each type, and the higher the difference in the numerical fluctuation status at this time; if the coherence importance between the previous course data packet queue of the th storage time point and the previous course data packet queue of the th type and the previous course data packet queue of the th type is higher, and the higher the coherence degree between a pair of previous course data packet queues of different types, that is, the th storage time point and the previous course data packet queue of the th type and the previous course data packet queue of the th type, the higher the amount of influence of the difference in the numerical fluctuation status between the previous course data packet queues of a pair of types on the coherence degree between the pair of course data packet queues. The sum of the amounts obtained by multiplying the coherence importance and the difference in the numerical fluctuation status of any pair of previous course data packet queues within the storage time point is calculated, and the higher the numerical coherence degree of any pair of previous course data packet types obtained.

[0095] Step3. According to the numerical coherence degree between the entire course data packet queues of each type and the entire course data packet queues of each other type, and the numerical dispersion amplitude of the entire course data packet queues of each type, obtain the numerical smoothness amplitude of the entire course data packet queues of each type; according to the numerical decrement attribute between the most recent course data packet queue and the entire course data packet queues of the corresponding type, and the numerical smoothness amplitude of the entire course data packet queues of each type, obtain the overall coherence amplitude between the most recent course data packet queues of all types and the entire course data packet queues of all types; according to the numerical definition difference amount and the overall coherence amplitude between the most recent course data packet queues of all types and the entire course data packet queues of all types, obtain the overall influence amplitude of the most recent course data packet queues of all types on the entire course data packet queues of all types;

[0096] Because there is often some coherence between the previous course data packets of different types, the numerical dispersion amplitude of the previous course data packets of one type often has a corresponding effect on the steady amplitude of the previous course data packets of other types. To prevent the previous course data packets of each type from being affected by the previous course data packets of other types, in this application, according to the numerical coherence degree between the course data packet queues of each type and the entire course data packet queues of each other type, and the numerical dispersion amplitude of the course data packet queues of each type, the numerical smoothness amplitude of the course data packet queues of each type is obtained.

[0097] In a preferred but non-limiting embodiment of the present invention, in Step3, the method for obtaining the numerical smoothness amplitude specifically includes:

[0098] Arbitrarily select a type as the target type; use the LCSS method to obtain all the adaptation clusters on the entire course data packet queue between each other type and the target type (the adaptation cluster is that for each course data packet on one course data packet queue in a pair of course data packet queues, find the adapted course data packet on the other course data packet queue using the LCSS method), and then execute the following equation:

[0099]

[0100] In the equation, represents the target type; represents the value smooth amplitude of the entire course data packet queue of the target type; represents the number of values in the entire course data packet queue of the target type; represents the number of other types; represents the numerical coherence degree between the entire course data packet queue of the target type and the entire course data packet queue of the th other type; represents the th value point's smooth coherence amplitude of the entire course data packet queue of the th other type facing the entire course data packet queue of the target type; represents the value of the th value point in the entire course data packet queue of the target type; represents the mean of all value points in the entire course data packet queue of the target type; represents the th number of adapted course data packets in the adaptation cluster where the value point in the entire course data packet queue of the th value point in the entire course data packet queue of the target type is located in the entire course data packet queue of the th other type; represents the value of the th adapted course data packet in the adaptation cluster where the value point in the entire course data packet queue of the th value point in the entire course data packet queue of the target type is located in the entire course data packet queue of the th other type; represents the mean of all value points in the entire course data packet queue of the represents performing standardization processing on using the Z-score method; is the Euler number.

[0101] Regarding the smooth coherence amplitude of the th value point of the entire course data packet queue of the th other type facing the entire course data packet queue of the target type, if the The lower the difference in the mean value between the value of an adapted course data packet and the value of the entire course data packet queue of another type, the lower it means that the th adapted course data packet is flatter overall with respect to the entire course data packet queue of another type. By taking the mean of the quotient obtained by dividing the dispersion amplitude of all adapted course data packets by one, the smooth coherence amplitude of the entire course data packet queue of the target type at the th numerical point of the adapted course data packet can be obtained.

[0102] In the numerical smooth amplitude operation equation, for each numerical point of the target type course data packet queue, calculate the total amount of the product of the smooth coherence amplitude of the adapted course data packet of another type and the coherence degree between a pair of types for each numerical value of the target type course data packet queue, and regard it as the affected amplitude of each numerical point of the target type. Here, the higher the coherence degree and the higher the smooth coherence amplitude of the adapted course data packet of another type, the more the numerical fluctuation situation of each numerical point of the target type will be affected by the previous course data packets of another type, and the higher the affected amplitude of each numerical point of the target type; sum up the affected amplitudes of each numerical point of the target type. The higher the total amount, the lower the numerical smooth amplitude of the entire course data packet queue of the target type. The numerical smooth amplitude reflects the amount of influence exerted by another type on the target type.

[0103] When storing previous course data packets, numerical values with similar arrangements are often grouped and segmented, and then each group is encoded and stored. When storing the most recent course data packet queue to be stored, it is necessary to store the most recent course data packet queue under the condition of the previously stored previous course data packets, that is, each newly formed most recent course data packet queue will not damage the segmentation situation of the previous course data packet groups. Because the defined difference amount between the newly formed most recent course data packet queue and all previous course data packets will have a significant impact on the segmentation of the previous course data packet groups, it is necessary to consider the influence of the most recent course data packet queue to be stored on all previous course data packets. Therefore, in this application, the overall influence amplitude of the most recent course data packet queue on the entire course data packet queue of all types is obtained.

[0104] During the grouping of the latest course data packet queue, the difference between the latest course data packet queue and the corresponding type of all course data packet queue is involved. The higher the difference, the higher the impact of the latest course data packet queue on the all course data packet queue. And all course data packet queues of each type are often affected by all course data packet queues of other types. Therefore, when the overall correlation amplitude between the latest course data packet queue and the corresponding type of course data packet queue is involved, the numerical smooth amplitude of all course data packet queues of each type must be analyzed. Therefore, in this application, the overall correlation amplitude between the latest course data packet queue of all types and all course data packet queues of all types is obtained based on the numerical reduction attribute between the latest course data packet queue and the corresponding type of all course data packet queue, as well as the numerical smooth amplitude of all course data packet queues of each type.

[0105] In a preferred but non-limiting embodiment of the present invention, in Step 3, the method for obtaining the overall coherent amplitude specifically includes:

[0106] In each recent course data packet queue, the minimum amount of difference between each numerical point and the corresponding type of course data packet queue is summed up to obtain the numerical reduction amplitude between each recent course data packet queue and each type of course data packet queue; the amount obtained by multiplying the numerical flat amplitude and the numerical reduction amplitude of all course data packet queues of each type in a pre-set past time period is summed up to obtain the overall correlation amplitude between all types of recent course data packet queues and all types of course data packet queues.

[0107] In a preferred but non-limiting embodiment of the present invention, in Step 3, the operation equation of the overall coherent amplitude is:

[0108]

[0109] In the equation, Represents the total correlation amplitude between the most recent course data packet queues of all categories and the total course data packet queues of all categories; Represents the number of types of all course data packet queues; Representative The numerical amplitude of the whole course data packet queue of each type; Representative The number of values ​​in the most recent course packet queue; Representative The most recent course packet queue The minimum amount of difference between individual data and the corresponding type of all course data packet queues.

[0110] In the overall coherence amplitude calculation equation, Via the lowest amount of the difference between each value of the most recent course data packet queue and all types of course data packet queues reflects the effect of the most recent course data packet on all previous course data packets. Here, the higher the lowest amount of the overall difference between the most recent course data packet queue and all types of course data packet queues, the more special the most recent course data packet queue is for all types of course data packet queues, and the more analysis needs to be performed on all types of course data packet queues. At this time, if the overall stability amplitude of all types of course data packet queues itself is higher, it means that the relevant amplitude of the most recent course data packet queue for all types of course data packet queues of the corresponding type is higher. Summing up the relevant amplitudes, at this time, the overall relevant amplitude between all types of most recent course data packet queues and all types of course data packet queues is higher.

[0111] Because the difference amount between the value definition of the most recent course data packet queue and the value definition of all types of course data packet queues also affects the subsequent grouping of the most recent course data packet queue. The higher the value definition of the most recent course data packet queue, the higher the effect of the most recent course data packet queue data on all types of course data packet queues. Therefore, in this application, according to the difference amount of value definition and the overall relevant amplitude between all types of most recent course data packet queues and all types of course data packet queues, the overall effect amplitude of the most recent course data packet queue on all types of course data packet queues is obtained.

[0112] In a preferred but non-limiting embodiment of the present invention, in Step3, the method for obtaining the difference amount of value definition specifically includes:

[0113] Taking the quotient obtained by dividing the number of values in all types of most recent course data packet queues by the total number of values in all types of course data packet queues as the difference amount of value definition between the most recent course data packet queue and all types of course data packet queues. The closer the ratio is to one, the closer the value definition between the most recent course data packet queue data and all types of course data packet queues is, and at this time, the higher the effect of the most recent course data packet queue data on all types of course data packet queues; when the quotient is higher than one, at this time, the value definition of the most recent course data packet queue is higher than that of all types of course data packet queues, and the effect of the most recent course data packet queue data on all types of course data packet queues is higher.

[0114] In a preferred but non-limiting embodiment of the present invention, in Step3, the method for obtaining the overall effect amplitude specifically includes:

[0115] The quantity obtained by multiplying the numerical definition difference quantity and the overall coherence amplitude is standardized using the Z-score method to obtain the overall effect amplitude of the nearest course data packet queue for the overall course data packet queues of all types.

[0116] In a preferred but non-limiting embodiment of the present invention, in Step 3, the overall effect amplitude operation equation is:

[0117]

[0118] In the equation, represents the overall effect amplitude of the nearest course data packet queue of all types facing the overall course data packet queues of all types; represents the number of values of the nearest course data packet queue of all types; represents the total number of values of the overall course data packet queues of all types; represents the overall coherence amplitude between the nearest course data packet queue of all types and the overall course data packet queues of all types; represents the use of the Z-score method for performing a standardization process.

[0119] In the overall effect amplitude operation equation, the higher the effect amplitude of the nearest course data packet queue of all types facing the overall course data packet queues of all types, and the higher the numerical definition difference quantity between the overall nearest course data packet queue and the overall course data packet queues of all types, it means that when grouping the overall nearest course data packet queue, the higher the overall effect amplitude of the nearest course data packet queue facing the overall course data packet queues of all types.

[0120] Thus, the overall effect amplitude of the nearest course data packet queue of all types for the overall course data packet queues of all types is obtained.

[0121] Step 4, obtain the final grouping of the nearest course data packet queue and the overall course data packet queues of all types according to the overall effect amplitude; encode and store all the nearest course data packets according to the final grouping. The method for identifying the course data packets in the database to obtain the corresponding completed state vectors includes taking out the encoded course data packets in the database, first performing decoding and restoration, and then identifying the decoded and restored course data packets to obtain the corresponding completed state vectors.

[0122] When grouping the most recent course data packet queue, it often presents splitting the most recent course data packet queue into the existing groups of all previous course data packets, and it also often re - groups the most recent course data packet queue. Here, the specific grouping situation depends on all the action amplitudes obtained in Step4. Therefore, in this application, the final grouping of the most recent course data packet queue and all types of course data packet queues is obtained according to all the action amplitudes.

[0123] In a preferred but non - restrictive embodiment of the present invention, in Step4, the method of grouping the most recent course data packet queue according to all the action amplitudes specifically includes:

[0124] Group all the previous course data packets within a preset previous time period through the fuzzy C - means grouping method to obtain source groups; preset a critical quantity one, and cut the most recent course data packet queue with all action amplitudes lower than the critical quantity one into the source groups through the DBSCAN method to obtain the final groups; directly group the most recent course data packet queue with all action amplitudes higher than the critical quantity one (that is, form a separate group for this most recent course data packet queue) to obtain all the latest groups of this most recent course data packet queue; regard all the latest groups and the source groups as the final groups. The value of the critical quantity one can be sixty percent.

[0125] In a preferred but non - restrictive embodiment of the present invention, in Step4, the values cut into the same group will have the same type of value arrangement situation. Many approximate redundant values can be encoded and stored through the encoding reduction method. The RLE method has a good encoding reduction effect on redundant values, and the RLE method is used to encode and store all the values in the final groups.

[0126] As Figure 2 shown, an adjustment device for course information based on natural language processing according to the present invention includes:

[0127] A formation module, which is used to obtain all types of the most recent course data packet queues to be stored and all types of previous course data packet queues at each storage time point within a preset previous time period; all types of previous course data packet queues at all storage time points form all types of all course data packet queues;

[0128] A correlation module, which is used to obtain the numerical correlation degree of any pair of types of course data packet queues according to the numerical dispersion amplitude of any pair of types of previous course data packet queues;

[0129] An action module, which is used to obtain the overall action amplitude of all types of recent course data packet queues on all types of all course data packet queues according to the numerical correlation between all course data packet queues of each type and all course data packet queues of each other type until the nearest course data packet queue of all types is obtained;

[0130] A storage module, which is used to obtain the final grouping of the nearest course data packet queue and all types of all course data packet queues according to the overall action amplitude; encode and store all the nearest course data packets according to the final grouping.

[0131] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0132] Obtain the past course data packet queues at each storage time point in the past period and the nearest course data packet queue to be stored; because the correlation between course data packet queues of different types will have a significant impact on the grouping and cutting of all types of course data packet queues, and the past course data packet queues at different storage time points often change with the change of the storage time point, so according to the numerical dispersion amplitude of any pair of types of past course data packet queues, obtain the correlation importance of any pair of types of past course data packet queues at each storage time point, reflecting the stability of the data of any pair of types of past course data packet queues at each storage time point; adjust the numerical fluctuation condition difference quantity of any pair of past course data packet queues at each storage time point through the correlation importance to obtain the numerical correlation between all course data packet queues of any pair of types; because the numerical dispersion amplitude of all course data packet queues of one type often has a significant impact on the data stability amplitude of all course data packet queues of other types, so obtain the numerical smooth amplitude of all course data packet queues of each type; in addition, because the nearest course data packet queue will have a certain amount of impact on all types of all course data packet queues, thus involving the existing grouping of all past course data packets, so obtain the overall action amplitude of all types of the nearest course data packet queue on all types of all course data packet queues, use the overall action amplitude to obtain the final grouping of the nearest course data packet queue and all types of all course data packet queues; encode all the nearest course data packets according to the final grouping. The present invention involves the correlation between past course data packet queues of different types and the impact of the nearest course data packet on past course data packets, obtains a grouping that accurately reflects the approximate numerical arrangement, and thus obtains an accurate course data packet storage effect.

[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications or equivalent replacements can still be made to the specific implementation manners of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention shall be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for adjusting course information based on natural language processing, characterized in that: include: Processing the course information based on natural language until the teaching completion degree corresponding to the course is formed, wherein the method of processing the course information based on natural language includes forming a course data package and storing it in a database; During the process of forming the course data package and storing it in the database, it also includes: Step 1, obtain the latest course data packet queue of all types to be stored and the past course data packet queue of all types at each storage time point in the pre-set past period; the past course data packet queue of all types at all storage time points forms the whole course data packet queue of all types; Step 2, according to the numerical dispersion amplitude of the previous course data packet queues of any pair of types, until the numerical coherence of the course data packet queues of any pair of types is obtained; Step 3, according to the numerical correlation between the whole course data packet queue of each type and the whole course data packet queue of each other type, until the whole effect amplitude of the nearest course data packet queue of all types on the whole course data packet queue of all types is obtained; Step 4, obtain the final grouping of the latest course data packet queue and all course data packet queues of all types according to the total effect amplitude; encode and store all the latest course data packets according to the final grouping.

2. The method for adjusting course information based on natural language processing according to claim 1, characterized in that: In Step 1, the text information in the obtained course data package is transformed into an attribute vector through the Word2Vec model, and the entire attribute vector obtained is processed by the ridge regression method; In Step 2, the method of obtaining the numerical coherence of a random pair of types of course data packet queues according to the numerical dispersion amplitude of a random pair of types of past course data packet queues includes: obtaining the coherence importance of a random pair of types of past course data packet queues at each storage time point according to the numerical dispersion amplitude of a random pair of types of past course data packet queues; obtaining the numerical coherence of a random pair of types of course data packet queues according to the numerical fluctuation state distinction and coherence importance of a random pair of types of past course data packet queues at each storage time point; In Step 3, the method of obtaining the overall effect amplitude of the latest course data packet queue of all types on all course data packet queues of all types according to the numerical coherence between all course data packet queues of all types and all course data packet queues of all other types, includes: obtaining the numerical smooth amplitude of all course data packet queues of all types according to the numerical coherence between all course data packet queues of all types and all course data packet queues of all other types, as well as the numerical dispersion amplitude of all course data packet queues of all types; obtaining the overall coherence amplitude between the latest course data packet queue of all types and all course data packet queues of all types according to the numerical decrement attribute between the latest course data packet queue and all course data packet queues of the corresponding type, as well as the numerical smooth amplitude of all course data packet queues of all types; obtaining the overall effect amplitude of the latest course data packet queue of all types on all course data packet queues of all types according to the numerical definition difference and the overall coherence amplitude between the latest course data packet queue of all types and all course data packet queues of all types.

3. The method for adjusting course information based on natural language processing according to claim 2, characterized in that: In Step 2, the method for obtaining the numerical dispersion amplitude includes: In each type of past course data packet queue, the difference between the value of each numerical point and the average of all numerical points in each type of past course data packet queue is summed up to obtain the numerical dispersion amplitude of each type of past course data packet queue at each storage time point, in each type of past course data packet queue; In Step 2, the calculation equation of the numerical dispersion amplitude is: In the equation, Representative The numerical dispersion amplitude of the data packet queues of the past courses of each category; Representative The first storage point The number of values ​​of the previous course data packet queues of each type; Representative The first storage point The first in the queue of past course data packets of the category The value of the numerical point; Representative The average of the values ​​in the data packet queues of the past courses of each category; In Step 2, the method for obtaining the coherent importance includes: The difference in the numerical dispersion amplitude of any pair of past course data packets is inversely proportionally adjusted and normalized to obtain the relevant importance of any pair of past course data packet queues at each storage time point; In Step 2, the coherence importance calculation equation is: In the equation, Representative The first storage point The previous course data packet queues and the The coherence importance between the data packets of past courses of various categories; Representative The numerical dispersion amplitude of the data packet queues of the past courses of each category; Representative The numerical dispersion amplitude of the data packet queues of the past courses of each category; Represents the use of Z-score method to Perform standardized disposal.

4. The method for adjusting course information based on natural language processing according to claim 3, characterized in that: In Step 2, the method for obtaining the value fluctuation status distinction quantity specifically includes: The regression line of the past course data packet of each type at each storage time point is constructed by the least square method; the regression line of any pair of types of past course data packets at each storage time point is fitted by the LCSS method to obtain the whole value fitting cluster of any pair of types of past course data packet queues at each storage time point; the derivative difference between a pair of value points in the whole value fitting cluster of each storage time point is summed up to obtain the value fluctuation state difference of any pair of types of past course data packet queues at each storage time point.

5. The method for adjusting course information based on natural language processing according to claim 4, characterized in that: In Step 2, the method for obtaining the numerical coherence degree specifically includes: Under any pair of categories, the value obtained by multiplying the correlation importance and the difference in numerical fluctuation between the past course data packet queues at each storage time point is taken as the multiplication value one, and the total amount of the multiplication value one of all storage time points is inversely proportionally allocated and normalized to obtain the numerical correlation between all course data packet queues of any pair of categories; In Step 2, the numerical coherence calculation equation is: In the equation, Representative All course data packets of the category are queued and The numerical correlation between all course data package queues of a category; Represents the number of storage points in the pre-set past period; Representative The first storage point The previous course data packet queues and the The coherence importance between the data packets of past courses of various categories; Representative The first storage point The previous course data packet queues and the The number of numerical adaptation clusters between the queues of past course data packets of each type; Representative The first storage point The previous course data packet queue of the category is in The derivative value of the numerical point in the numerical adaptation cluster on its regression line; Representative The first storage point The previous course data packet queue of the category is in The derivative value of the numerical point in the numerical adaptation cluster on its regression line; is the Euler number.

6. The method for adjusting course information based on natural language processing according to claim 5, characterized in that: In Step 3, the method for obtaining the numerical smooth amplitude includes: Randomly select a class as the target class; use the LCSS method to obtain the entire adaptation course data packet consisting of all adaptation clusters on the entire course data packet queue between each other class and the target class, and then execute the following equation: In the equation, Representative purpose type; The numerical amplitude of the whole course data packet queue representing the purpose category; The number of values ​​in the queue of all course data packets representing the target category; Represents the number of other kinds; Representative purpose type and The numerical correlation between all course data sets of another type; Representative Another type of all course data packet queue facing the target type of all course data packet queue The smooth coherence amplitude of the numerical points; The first in the queue of all course data packets representing the target category The value of the numerical point; The average of all numerical points in all course data packet queues representing the target category; Representative In the whole course package queue of another type, and in the whole course package queue of the target type The number of adapted course data packets in the adapted cluster where the numerical point is located; Representative In the whole course package queue of another type, and in the whole course package queue of the target type The first value of the adaptation cluster where the value point is located The value of the adapted course data package; Representative The mean of all numerical points of all course data sets of another type; Represents the use of Z-score method to Implement standardized disposal; is the Euler number.

7. The method for adjusting course information based on natural language processing according to claim 6, characterized in that: In Step 3, the method for obtaining the overall coherent amplitude includes: In each recent course data packet queue, the minimum amount of the difference between each numerical point and the corresponding type of course data packet queue is summed up to obtain the numerical reduction amplitude between each recent course data packet queue and each type of course data packet queue; the amount obtained by multiplying the numerical flat amplitude and the numerical reduction amplitude of all course data packet queues of all types in the pre-set past period is summed up to obtain the overall correlation amplitude between all types of recent course data packet queues and all types of course data packet queues; In Step 3, the operational equation for the overall coherent amplitude is: In the equation, Represents the total correlation amplitude between the most recent course data packet queues of all categories and the total course data packet queues of all categories; Represents the number of types of all course data packet queues; Representative The numerical amplitude of the whole course data packet queue of each type; Representative The number of values ​​in the most recent course packet queue; Representative The most recent course packet queue The minimum amount of difference between individual data and the corresponding type of all course data packet queues.

8. The method for adjusting course information based on natural language processing according to claim 7, characterized in that: In Step 3, the method for obtaining the numerical definition difference quantity includes: The quotient obtained by dividing the number of values ​​in the latest course data package queue of all types by the total number of values ​​in the entire course data package queue of all types is used as the value definition difference between the latest course data package queue and the entire course data package queue; In Step 3, the method for obtaining the overall action amplitude includes: The quantity obtained by multiplying the numerical definition difference and the overall correlation amplitude is normalized using the Z-score method to obtain the overall effect amplitude of the nearest course data packet queue on all course data packet queues of all types; In Step 3, the equation for calculating the overall action amplitude is: In the equation, Represents the total effect magnitude of the latest course data packet queues of all categories on all course data packet queues of all categories; The number of values ​​representing the most recent course packet queues of all categories; The total number of values ​​representing all course data packet queues of all categories; Represents the total correlation amplitude between the most recent course data packet queues of all categories and the total course data packet queues of all categories; Represents the use of Z-score method to Perform standardized disposal.

9. The method for adjusting course information based on natural language processing according to claim 8, characterized in that: In Step 4, the method of grouping the recent course data packet queue according to the overall action amplitude includes: The fuzzy C-means grouping method is used to group all past course data packets within a preset past period to obtain source groups; a critical value of one is set in advance, and the DBSCAN method is used to cut all recent course data packet queues with action amplitudes lower than the critical value of one into source groups to obtain final groups; the recent course data packet queues with all action amplitudes higher than the critical value of one are directly grouped to obtain all the latest groups of the recent course data packet queues; all the latest groups and source groups are regarded as final groups; In Step 4, the RLE method can achieve good coding reduction effect for redundant values, and the coding and storage are performed by applying the RLE method to all values ​​in the final group.

10. A device for adjusting course information based on natural language processing, characterized in that: include: A forming module is used to obtain the latest course data packet queue of all types to be stored and the past course data packet queue of each type at each storage time point in the pre-set past period; the past course data packet queue of each type at all storage time points forms the whole course data packet queue of each type; A coherence module, which is used to obtain the numerical coherence of a pair of course data packet queues according to the numerical dispersion amplitude of a pair of past course data packet queues; An effect module, which is used to obtain the total effect amplitude of the nearest course data packet queue of all types on the total course data packet queue of all types according to the numerical correlation between the total course data packet queue of each type and the total course data packet queue of each other type; The storage module is used to obtain the latest course data packet queue and the final grouping of all course data packet queues of all types according to the total action amplitude; and encode and store all the latest course data packets according to the final grouping.

Images