Note generation method, system and readable storage medium

Abstract

The present invention discloses a note generation method, system and readable storage medium, the method comprising: obtaining note keywords and pictures to be edited from information uploaded by users, and obtaining content keywords corresponding to the note keywords; identifying editing keywords corresponding to pictures to be edited, and encrypting and fusing them with pictures to be edited to form display information; obtaining historical browsing information of users, and identifying personality keywords therefrom, and generating note content based on each content keyword, each editing keyword and each personality keyword; generating a note title based on the note content and display information, and generating the note title, note content and display information into a complete note. The present invention realizes accurate reflection of user needs from multiple dimensions by generating note content based on each content keyword, editing keyword and personality keyword, and realizes fast and accurate generation of notes that meet user needs and are secure by encrypting and fusing editing keywords and pictures to be edited.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a note generation method, system and readable storage medium. Background Art

[0002] Xiaohongshu is a popular software and a lifestyle platform for young people. It supports posting notes and sharing life experiences, making it easier for people to discover the real, beautiful and diverse world and find the life they want. With the development of artificial intelligence technology, posting notes has become more convenient. People can generate notes by inputting keywords of the notes they want to post, without having to spend more time thinking and writing notes.

[0003] However, the notes formed in this way rely on the accurate and large-scale input of keywords by the note publisher. The more accurate and the greater the input, the more the notes formed will fit the needs of the note publisher. This also requires the publisher to spend time to construct keywords, otherwise the generated notes cannot meet the needs of the publisher. At the same time, Xiaohongshu's notes do not support full text publishing and need to carry pictures or videos. As a result, it is easy to cause the content of the published notes to be inconsistent with the pictures or videos uploaded by the publisher, which also cannot meet the publishing needs of the note publisher. In addition, the pictures or videos uploaded by the publisher are easily stolen by other users. Therefore, how to quickly and accurately generate notes that meet user needs and are secure is a technical problem that needs to be solved at present. Summary of the invention

[0004] The main purpose of the present invention is to provide a note generation method, system and readable storage medium, aiming to solve the technical problem of how to quickly and accurately generate notes that meet user needs and are secure.

[0005] To achieve the above object, the present invention provides a note generation method, the note generation method comprising:

[0006] Obtaining note keywords and pictures to be edited from information uploaded by the user based on a preset template, and obtaining content keywords corresponding to the note keywords based on a preset vocabulary;

[0007] Identifying an editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information;

[0008] Acquire the user's browsing history information, identify personality keywords from the browsing history information, and generate note content based on each of the content keywords, each of the editing keywords, and each of the personality keywords;

[0009] A note title is generated based on the note content and the display information, and the note title, the note content and the display information are generated as a complete note.

[0010] Optionally, the step of identifying an editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information includes:

[0011] Identify the person in the picture to be edited, obtain the editing keyword, and identify the area to be added and the characteristic elements corresponding to the picture to be edited;

[0012] Determine the pixel parameters of the editing keyword according to the characteristic elements, and add the editing keyword having the pixel parameters to the to-be-added area of ​​the to-be-edited picture to form the to-be-encrypted picture;

[0013] The picture to be encrypted is encrypted to form display information.

[0014] Optionally, the step of encrypting the picture to be encrypted to form display information includes:

[0015] Forming the picture to be encrypted into multiple sub-pictures, and obtaining characteristic parameters of each of the sub-pictures;

[0016] Based on the pixel mean of each sub-image and each characteristic parameter, the signature parameter of the image to be encrypted is calculated. The formula for calculating the signature parameter is:

[0017]

[0018] Randomly obtain pixel points from the image to be encrypted, and determine the coordinate values ​​of the pixel points;

[0019] An encryption password is calculated based on the signature parameter and the coordinate value, and the image to be encrypted is encrypted based on the encryption password to form display information. The formula for calculating the encryption password is:

[0020]

[0021] Among them, S is the signature parameter, ∑ is the accumulation operator, n is the number of word images, f(i) is the pixel mean function, is the (i-1)th characteristic parameter, k1 and k2 are passwords, π is the symbol of pi, sin is the sine function, cos is the dotted line function, and (x, y) are the coordinate values.

[0022] Optionally, the steps of identifying the person in the picture to be edited, obtaining the editing keyword, and identifying the area to be added and the characteristic elements corresponding to the picture to be edited include:

[0023] Identify a prominent area containing a person in the picture to be edited, and the attributes of the person contained therein, and obtain a keyword corresponding to the attribute as the editing keyword;

[0024] The area outside the prominent area that does not contain the person is regarded as a non-salient area, and the area to be added is determined from the non-salient area according to the note keyword;

[0025] Acquire a first pixel mean value of the salient area and a second pixel mean value of the non-salient area, and generate an absolute value of a difference between the first pixel mean value and the second pixel mean value;

[0026] The characteristic elements of the image to be edited are determined according to the magnitude relationship between the absolute value of the difference and a preset first threshold.

[0027] Optionally, the step of identifying personality keywords from the historical browsing information includes:

[0028] Filtering out historical attribute keywords from each of the historical browsing information, and forming each of the historical attribute keywords into a first feature vector;

[0029] The second feature vectors corresponding to the preset personality keywords are obtained, and the first feature vector and each of the second feature vectors are calculated respectively to generate a vector similarity value. The calculation formula is:

[0030] Wi=

[0031] (α*p*ln (α*p+(1-α)*qi) ) / ((1-α)*qi*ln (α*qi+(1-α)*p) );

[0032] Each of the vector similarity values ​​is compared with a preset second threshold value, a target vector similarity value greater than the preset second threshold value is screened out, and a preset personality keyword corresponding to the target vector similarity value is determined as the personality keyword.

[0033] Among them, Wi is the vector similarity value, α is the historical key parameter corresponding to the historical attribute keyword, p is the first eigenvector, ln is the natural logarithm, and qi is the second eigenvector corresponding to the i-th preset personality keyword.

[0034] Optionally, the step of generating note content based on each of the content keywords, each of the editing keywords and each of the personality keywords includes:

[0035] Generate content texts corresponding to each of the content keywords and edit texts corresponding to each of the edit keywords based on a preset generation model;

[0036] The content text and the edit text are combined into a text to be adjusted, and the text to be adjusted is adjusted based on the personality keywords to obtain the note content.

[0037] Optionally, before the step of generating the content text corresponding to each of the content keywords and the editing text corresponding to each of the editing keywords based on the preset generation model, the step includes:

[0038] Obtaining popularity information of published notes within a preset period, and filtering out target notes from the published notes according to the popularity information;

[0039] Dividing each of the target notes into training samples and test samples, and transmitting each training sample to a preset generation model for optimization training;

[0040] After the number of training times reaches the preset number of times, the preset generation model is tested based on the test sample, and the loss function value of the preset generation model is calculated, and the preset generation model is optimized based on the loss function value. The calculation formula is:

[0041]

[0042] Among them, L is the loss function value, exp is the exponential function, h is the model parameter of the preset generation model, m is the number of test samples, γ is the label feature corresponding to the test sample, and γi is the test feature generated based on the i-th test.

[0043] Optionally, the step of acquiring content keywords corresponding to the note keywords based on a preset vocabulary includes:

[0044] Identify the subject words in the note keywords, and identify the sub-word library corresponding to the subject words in the preset word library;

[0045] The note keywords are converted into a first feature matrix, and the second feature matrix corresponding to each feature phrase in the sub-lexicon is obtained, and the overlap value between the first feature matrix and each second feature matrix is ​​calculated. The calculation formula is:

[0046]

[0047] Determine a target coincidence value with the largest value among the coincidence values, and determine a characteristic phrase corresponding to the target coincidence value as the content keyword;

[0048] Among them, G represents the overlap value, max is the maximum value, B is the first feature matrix of note keyword conversion, Di is the i-th second feature matrix corresponding to the i-th feature phrase, and cos is the cosine function.

[0049] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a note generation system, the note generation system comprising: a memory, a processor, a communication bus, and a control program stored in the memory:

[0050] The communication bus is used to realize the connection and communication between the processor and the memory;

[0051] The processor is used to execute the control program to implement the steps of the note generating method as described above.

[0052] Furthermore, to achieve the above-mentioned purpose, the present invention also provides a readable storage medium, on which a control program is stored, and when the control program is executed by a processor, the steps of the note generation method as described above are implemented.

[0053] The note generation method, system and readable storage medium of the present invention are pre-set with a preset template, and the user can input the keywords for the notes to be generated and upload the pictures or videos to be published in the preset module. Once the keywords, pictures or videos and other information uploaded through the preset template are received, the note keywords and the pictures to be edited used to generate the notes are obtained from the information, and the content keywords corresponding to the note keywords are obtained based on the preset vocabulary, and the editing keywords corresponding to the pictures to be edited are identified at the same time, and then the editing keywords and the pictures to be edited are encrypted and merged to form display information; in addition, the user's historical browsing information is also obtained, and the personality keywords are identified from the acquired historical browsing information, and then the note content is generated according to each content keyword, each editing keyword and each personality keyword, and the note title is generated according to the note content and the display information, and the note title, note content and display information are jointly generated into a complete note. In this way, the user is allowed to input information including text and pictures or videos through the preset template, so that the obtained note keywords and the pictures to be edited are correlated, which reflects the user's need to generate notes. At the same time, content keywords reflect content related to note keywords, editing keywords reflect content related to pictures or videos, and personality keywords are used to reflect the user's favorite style, so that the note content generated based on each content keyword, each editing keyword, and each personality keyword can accurately reflect user needs from multiple dimensions, avoiding users spending a lot of time to accurately construct keywords. In addition, by encrypting and integrating editing keywords and the pictures to be edited, other users need to enter the correct password when downloading, avoiding theft by other users, thereby achieving fast and accurate generation of notes that meet user needs and are secure. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of the first embodiment of the note generation method of the present invention;

[0055] Figure 2 It is a flowchart of the second embodiment of the note generation method of the present invention;

[0056] Figure 3 It is a flowchart of the third embodiment of the note generation method of the present invention;

[0057] Figure 4 This is a schematic diagram of the structure of the hardware operating environment involved in an embodiment of the note generation system of the present invention.

[0058] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0059] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0060] The present invention provides a note generation method, please refer to Figure 1 , Figure 1 It is a flowchart of the first embodiment of the note generation method of the present invention.

[0061] The embodiment of the present invention provides an embodiment of a note generation method. It should be noted that although the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than that shown here. Specifically, the note generation method in this embodiment includes:

[0062] Step S10, obtaining note keywords and pictures to be edited from the information uploaded by the user based on a preset template, and obtaining content keywords corresponding to the note keywords based on a preset vocabulary.

[0063] The note generation method of this embodiment is applied to a server, and the server supports communication connection with multiple smart terminals. The smart terminal can be a mobile terminal such as a mobile phone and a tablet computer. The mobile terminal is installed with a Xiaohongshu client. When a Xiaohongshu user has the need to generate notes, a generation instruction is triggered on the display interface of the mobile terminal. After receiving the generation instruction, the server sends the pre-set preset template to the display interface for display, so that the user can set the keywords of the notes to be generated. The setting method can be to select information related to the notes to be generated from the many options contained in the preset template, such as selecting travel, travel destination, etc., or to enter information related to the notes to be generated in the input box of the preset template, such as entering food, food names, etc. In addition, the preset template also supports uploading pictures or videos, so that users can select the pictures or videos to be uploaded, and upload the uploaded pictures or videos together with the set keywords as information corresponding to the generated notes.

[0064] Furthermore, after receiving such information uploaded by the user through a preset template, the server obtains the keywords set by the user as note keywords, and obtains the pictures or videos therein for processing. The processing method includes beautifying the pictures, extracting pictures from the videos, and using the beautified pictures or extracted pictures as pictures to be edited. In addition, the server is pre-set with a preset vocabulary library, which contains a large number of words related to the content of the notes. For each note keyword obtained, it is compared with the words contained in the preset vocabulary library, and the words corresponding to the note keywords are determined as content keywords, so as to facilitate the subsequent generation of corresponding content texts based on the content keywords. Specifically, the step of obtaining content keywords corresponding to the note keywords based on the preset vocabulary library includes:

[0065] Step S11, identifying a subject word in the note keyword, and identifying a sub-thesaurus corresponding to the subject word in the preset thesaurus;

[0066] Step S12, converting the note keywords into a first feature matrix, obtaining a second feature matrix corresponding to each feature phrase in the sub-lexicon, and calculating a coincidence value between the first feature matrix and each of the second feature matrices;

[0067] Step S13, determining a target coincidence value having the largest value among the coincidence values, and determining a characteristic phrase corresponding to the target coincidence value as the content keyword.

[0068] Furthermore, the note keywords can reflect the subject of the notes to be generated. They can be words that directly include the subject in the note keywords, or words related to the subject can be identified through the note keywords. For example, if the note keywords include the name of a hotel, the notes to be generated may be related to the hotel accommodation, and the hotel name and accommodation are used as subject words; for another example, if the note keywords include not only the name of a hotel but also the name of a scenic spot, the city where the scenic spot and the hotel are located can be determined through this type of note keywords, and the city name and travel are used as subject words.

[0069] Furthermore, a preset word library reflecting the content of the notes is pre-set, and the preset word library is provided with different sub-word libraries according to different subject words, one sub-word library corresponds to one subject word, and contains a plurality of characteristic phrases of the note content related to the subject word. For the identified subject words, the sub-word libraries corresponding to each subject word are determined from the preset word library. Then, the note keywords are converted into a first characteristic matrix, and a plurality of second characteristic matrices corresponding to the characteristic phrases in the sub-word library are obtained. Among them, the sub-word library contains a plurality of characteristic phrases related to the subject words, and each characteristic phrase contains a plurality of words. The note keywords can be formed into a one-dimensional matrix according to the number of note keywords, that is, the number of columns of the first characteristic matrix is ​​one column, and the number of rows is the same as the number of note keywords. The corresponding second characteristic matrix can be formed according to the number of rows of the first characteristic matrix, and the words contained in the characteristic phrases used to generate the second characteristic matrix can be the same, but the arrangement positions of the words are different. For example, the sub-lexicon includes feature phrases A and B. The words contained in the feature phrases A and B are the same, but the arrangement positions of the words are different, resulting in a different second feature matrix, which in turn causes the same words to appear in different positions in the note content, forming different note contents.

[0070] Furthermore, the coincidence values ​​of the first characteristic matrix and each second characteristic matrix are calculated respectively, and the calculation formula is shown in the following formula (1).

[0071]

[0072] Among them, G represents the overlap value, max is the maximum value, B is the first feature matrix of note keyword conversion, Di is the i-th second feature matrix corresponding to the i-th feature phrase, and cos is the cosine function.

[0073] The calculated overlap values ​​are compared to determine the target overlap value with the largest value. The target overlap value reflects the characteristic phrase that is closest to the note keyword. Therefore, the characteristic phrase that generates the target overlap value is searched as the content keyword for generating the note content, so as to generate note content that meets the user's content needs based on the content keyword.

[0074] Step S20, identifying the editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information.

[0075] Furthermore, the image to be edited obtained from the preset template is identified to determine the portraits and / or object images contained therein and the features of the portraits and / or object images, and the features are identified as editing keywords, and then the identified editing keywords are added to the image to be edited, encrypted and fused with the image to be edited to form secure display information in the form of graphics and text.

[0076] Step S30, obtaining the user's historical browsing information, identifying personality keywords from the historical browsing information, and generating note content based on each of the content keywords, each of the editing keywords and each of the personality keywords.

[0077] Furthermore, in order to reflect the user's preferred style, the historical browsing information formed by the user browsing the notes posted by other users on Xiaohongshu is also obtained, and then the obtained historical browsing information is identified to obtain the personality keywords reflecting the user's preferences. For example, if it is determined that the styles of multiple notes browsed by the user are all cute, or all have the same elements, the personality keyword is identified as cute, or the word corresponding to the element. Specifically, the step of identifying the personality keyword from the historical browsing information includes:

[0078] Step S31, filtering out historical attribute keywords from each of the historical browsing information, and forming each of the historical attribute keywords into a first feature vector;

[0079] Step S32, obtaining second feature vectors corresponding to each preset personality keyword, and calculating the first feature vector and each second feature vector respectively to generate a vector similarity value;

[0080] Step S33, comparing each of the vector similarity values ​​with a preset second threshold value, screening out a target vector similarity value greater than the preset second threshold value, and determining a preset personality keyword corresponding to the target vector similarity value as the personality keyword.

[0081] Understandably, users will view and browse the notes posted by other users through Xiaohongshu, and collect, comment on, and like the notes that interest them. The server obtains the note content corresponding to the user's such operation on Xiaohongshu as historical browsing information, and identifies each of the historical browsing information obtained to determine the element keywords and style keywords contained in each. Among them, the identified element keywords are keywords that reflect the style, and each element keyword is classified, and multiple style categories with the largest number of element keywords are screened out, and the corresponding classification keywords are determined for each screened style category, and then the determined classification keywords and style keywords are used together as historical attribute keywords. For example, a certain historical browsing information is a travel-related note content, in which the identified element keyword contains "dopamine", and the style keyword is identified as "sunshine", then the determined historical attribute keywords are "happy" and "sunshine".

[0082] Furthermore, each of the selected historical attribute keywords is vectorized to obtain a first feature vector corresponding to each historical attribute keyword. In addition, a plurality of preset personality keywords representing personality are pre-set in the server, and each preset personality keyword is also pre-vectorized, and each group corresponds to a second feature vector. For each historical attribute keyword, its first feature vector is used to calculate with each second feature vector to obtain a vector similarity value between the first feature vector and each second feature vector, and the calculation formula is shown in the following formula (2).

[0083] Wi=

[0084] (α*p*ln (α*p+(1-α)*qi) ) / ((1-α)*qi*ln (α*qi+(1-α)*p) ) (2);

[0085] Among them, Wi is the vector similarity value, α is the historical key parameter corresponding to the historical attribute keyword, p is the first eigenvector, ln is the natural logarithm, and qi is the second eigenvector corresponding to the i-th preset personality keyword.

[0086] Furthermore, for each historical attribute keyword, the vector similarity values ​​calculated are compared with the preset second threshold representing the degree of similarity, and the target vector similarity value greater than the preset second threshold is determined, and then the second feature vector from which each target vector similarity value comes is found, and the preset personality keyword corresponding to the second feature vector is found. The preset personality keyword found is the personality keyword corresponding to the historical attribute keyword. After each historical attribute keyword has determined its corresponding personality keyword, it is possible to identify all personality keywords that reflect the user's preferred note style from the historical browsing information, so as to form note content that meets the user's content needs and style preferences based on this type of personality keyword and various content keywords and editing keywords.

[0087] Furthermore, content keywords are extended from the note keywords set by the user, reflecting the user's demand for note content, editing keywords are identified from the image to be edited, reflecting the characteristics of the image to be edited, and can be used to generate note content that describes the characteristics of the image, and personality keywords reflect the user's preferred note style. Therefore, based on each content keyword, editing keyword and personality keyword, note content that meets the user's content needs and style preferences can be generated.

[0088] Step S40, generating a note title based on the note content and the display information, and generating the note title, the note content and the display information into a complete note.

[0089] Furthermore, the generated note content and display information are subject-identified to determine the common theme of the two and generate a note title. The server is pre-set with a note template for generating notes. The generated note title, note content, and display information are added to the area corresponding to the note template to generate a complete note. The generated complete note is returned to the display interface of the smart terminal for the user to view. If the user is not satisfied with the generated complete note, he or she can modify it according to his or her needs, such as modifying the style. Of course, he or she can also choose to regenerate the note, and finally generate a Xiaohongshu note that meets the user's needs.

[0090] The note generation method implemented in this embodiment is pre-set with a preset template. The user can enter the keywords for the notes to be generated and upload the pictures or videos to be published in the preset module. Once the keywords, pictures or videos uploaded through the preset template are received, the note keywords and the pictures to be edited for generating the notes are obtained from the information, and the content keywords corresponding to the note keywords are obtained based on the preset vocabulary, and the editing keywords corresponding to the pictures to be edited are identified at the same time, and then the editing keywords and the pictures to be edited are encrypted and merged to form display information; in addition, the user's historical browsing information is obtained, and the personality keywords are identified from the acquired historical browsing information, and then the note content is generated according to each content keyword, each editing keyword and each personality keyword, and the note title is generated according to the note content and the display information, and the note title, note content and display information are jointly generated into a complete note. In this way, the user is allowed to input information including text and pictures or videos through the preset template, so that the obtained note keywords and the pictures to be edited are correlated, which reflects the user's need to generate notes. At the same time, content keywords reflect content related to note keywords, editing keywords reflect content related to pictures or videos, and personality keywords are used to reflect the user's favorite style, so that the note content generated based on each content keyword, each editing keyword, and each personality keyword can accurately reflect user needs from multiple dimensions, avoiding users spending a lot of time to accurately construct keywords. In addition, by encrypting and integrating editing keywords and the pictures to be edited, other users need to enter the correct password when downloading, avoiding theft by other users, thereby achieving fast and accurate generation of notes that meet user needs and are secure.

[0091] For further information, please refer to Figure 2 Based on the first embodiment of the note generation method of the present invention, a second embodiment of the note generation method of the present invention is proposed.

[0092] The difference between the second embodiment of the note generation method and the first embodiment of the note generation method is that the step of identifying the editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information includes:

[0093] Step S21, identifying the person in the picture to be edited, obtaining the editing keyword, and identifying the area to be added and the characteristic elements corresponding to the picture to be edited;

[0094] Furthermore, the characters included in the picture to be edited are identified, and the characters are portraits and / or object images in the picture to be edited. By identifying the characteristic attributes of the portraits and / or object images, the editing keywords corresponding to the picture to be edited are obtained. In addition, the overall layout and overall color tone of the picture to be edited are also identified, and the area to be added suitable for adding editing keywords and the characteristic elements reflecting the overall style and color are determined. Specifically, the steps of identifying the characters in the picture to be edited, obtaining the editing keywords, and identifying the area to be added and the characteristic elements corresponding to the picture to be edited include:

[0095] Step S211, identifying a prominent area containing a person in the picture to be edited, and attributes of the person contained therein, and obtaining a keyword corresponding to the attribute as the editing keyword;

[0096] Step S212, treating the area outside the prominent area that does not contain the person as a non-salient area, and determining the area to be added from the non-salient area according to the note keyword;

[0097] Step S213, obtaining a first pixel mean value of the salient area and a second pixel mean value of the non-salient area, and generating an absolute value of a difference between the first pixel mean value and the second pixel mean value;

[0098] Step S214, determining the characteristic elements of the image to be edited according to the magnitude relationship between the absolute value of the difference and a preset first threshold.

[0099] Furthermore, the contour area of ​​the person included in the image to be edited is identified as a prominent area in the image to be edited, and the features of the person included in the image to be edited are identified based on the note keywords. For example, if the note keywords include tops, the style, color, and style of the tops of the included person are identified to obtain the attributes of the included person. In addition, a mapping relationship library of attributes and keywords is pre-set, and the identified attributes are compared with the various attributes in the mapping relationship library to determine the corresponding attributes, and then the keywords mapped by the corresponding attributes in the mapping relationship library are searched, and the keywords found are the editing keywords that reflect the characteristics of the person.

[0100] Furthermore, the area outside the figure outline area in the image to be edited is regarded as a non-salient area, and the area to be added in the non-salient area is determined based on the content of the user's attention reflected by the note keyword. For example, for the note keyword including tops, the area near the top in the non-salient area is regarded as the area to be added for adding and merging the editing keywords.

[0101] Furthermore, in order to make the added editing keywords more compatible with the overall style of the image to be edited, the RGB three-color values ​​of each pixel in the prominent area are also obtained, and the RGB three-color values ​​are averaged to obtain the first pixel mean; at the same time, the RGB three-color values ​​of each pixel in the non-salient area are also obtained, and the obtained RGB three-color values ​​are also averaged to obtain the second pixel mean. Then, a difference operation is performed between the first pixel mean and the second pixel mean to obtain the absolute value of the difference between the two. In order to characterize the size of the absolute value of the difference, a preset first threshold is pre-set, and the absolute value of the difference is compared with the preset first threshold to determine whether the absolute value of the difference is greater than the preset first threshold. If it is greater than the preset first threshold, it means that the difference between the first pixel mean and the second pixel mean is large. At this time, in order to reduce the difference between the two, the two can be averaged, and the result obtained by the average operation is used as the characteristic element of the image to be edited, so that the editing keyword is added to the to-be-added area of ​​the image to be edited according to the RGB three-color value corresponding to the characteristic element, so that the overall style of the image to be edited is more integrated. On the contrary, if the absolute value of the difference is determined to be not greater than the preset first threshold value through comparison, it means that the difference between the first pixel mean and the second pixel mean is not large. At this time, the first pixel mean can be used as a characteristic element of the image to be edited, so that the editing keyword can be added to the to-be-added area of ​​the non-significant area according to the color tone style of the significant area, which can also make the overall style of the image to be edited more integrated.

[0102] Step S22, determining the pixel parameters of the editing keyword according to the characteristic elements, and adding the editing keyword having the pixel parameters to the to-be-added area of ​​the to-be-edited picture to form the to-be-encrypted picture;

[0103] Step S23, encrypting the picture to be encrypted to form display information.

[0104] Furthermore, according to the RGB three-color values ​​​​embodied by the characteristic elements, the pixel parameters of the editing keyword are determined, and the pixel parameters at least include R value, G value and B value. Then, the editing keyword is set to this type of pixel parameter, and the editing keyword with the pixel parameter is added to the to-be-added area of ​​the to-be-edited picture to form a picture to be encrypted. Thereafter, the picture to be encrypted is encrypted and formed into display information. The encrypted display information can be viewed by other Xiaohongshu users, but the correct password needs to be entered to prevent theft. Among them, the step of encrypting the picture to be encrypted to form display information includes:

[0105] Step S231, forming the picture to be encrypted into multiple sub-pictures, and obtaining characteristic parameters of each of the sub-pictures;

[0106] Step S232, calculating the signature parameter of the image to be encrypted based on the pixel mean of each sub-image and each characteristic parameter;

[0107] Step S233, randomly obtaining pixel points from the image to be encrypted, and determining the coordinate values ​​of the pixel points;

[0108] Step S234, calculating an encryption password based on the signature parameter and the coordinate value, and encrypting the image to be encrypted based on the encryption password to form display information.

[0109] Furthermore, the image to be encrypted is divided into a plurality of sub-images, and the specific number can be set according to the demand, such as 4, 9, 12, 16, etc. Then, the image features contained in each sub-image are identified to obtain respective feature parameters. Among them, the identified image features can be to randomly select any color block from the sub-image, or to use the image contained in the sub-image as the selected color block, and then randomly select pixels from the color block or select pixels according to a certain rule, such as an arithmetic number or a geometric number, and form the selected pixels into a vector as the feature parameter of the sub-image. In addition, the pixel points of each sub-image are averaged to obtain the pixel mean of each sub-image, and then the signature parameters of the image to be encrypted are calculated based on the pixel mean and feature parameters of each sub-image. The calculation formula is shown in the following formula (3).

[0110]

[0111] Among them, S is the signature parameter, ∑ is the accumulation operator, n is the number of word images, f(i) is the pixel mean function, is the (i-1)th characteristic parameter.

[0112] Furthermore, a pixel point is randomly selected from the image to be encrypted, and the coordinate value of the pixel point is identified. Among them, one corner can be selected from the four corners of the image to be encrypted as the coordinate origin, the side extending from the coordinate origin in the horizontal direction is the X axis, and the side extending in the vertical direction is the Y axis. The coordinate value of the selected pixel point is identified through the two-dimensional coordinate system. Then, the encryption password is calculated based on the calculated signature parameters and the coordinate value. The calculation formula is shown in the following formula (4).

[0113]

[0114] Among them, k1 and k2 are passwords, π is the symbol of pi, S is the signature parameter, sin is the sine function, cos is the dashed line function, and (x, y) is the coordinate value.

[0115] Furthermore, the encrypted image is encrypted with an encryption password, and the encrypted image forms display information. The display information supports other Xiaohongshu users to view through the Xiaohongshu client, but if the display information is to be saved locally, it is necessary to enter the decryption password corresponding to the encryption password to decrypt the display information, so as to avoid arbitrary theft of images in Xiaohongshu notes.

[0116] It should be noted that the decryption password corresponding to the encryption password can be the inverse operation of the encryption password, which will not be described in detail here. In addition, multiple pixels can be randomly selected for the calculation of the encryption password, and the calculation is performed through the coordinate values ​​of the multiple pixels and the signature parameters, so that the calculated encryption password is more secure, thereby making the encryption of the image to be encrypted more secure.

[0117] This embodiment determines the area to be added from the non-salient area of ​​the image to be edited, and determines the characteristic elements that reflect the overall tone style of the image to be edited, and then adds the editing keywords to the area to be added in the image to be edited according to the characteristic elements, so that the added editing keywords are more integrated with the overall tone style of the image to be edited, while also avoiding the occlusion of the significant area in the image to be edited, and better achieving the graphic effect. In addition, an encryption mechanism is set for the image to be edited with the editing keywords added, and the signature parameters are calculated through the pixel information contained in the image, and then the encryption password is calculated for encryption operation, so as to avoid the generated display information from being arbitrarily stolen, thereby improving the security of generating Xiaohongshu notes.

[0118] For further information, please refer to Figure 3 Based on the first and second embodiments of the note generation method of the present invention, a third embodiment of the note generation method of the present invention is proposed.

[0119] The difference between the third embodiment of the note generation method and the first and second embodiments of the note generation method is that the step of generating note content based on each of the content keywords, each of the editing keywords and each of the personality keywords includes:

[0120] Step S34, generating content texts corresponding to each of the content keywords and editing texts corresponding to each of the editing keywords based on a preset generation model;

[0121] Step S35, combining the content text and the edit text into a text to be adjusted, and adjusting the text to be adjusted based on the personality keywords to obtain the note content.

[0122] Furthermore, a preset generation model for generating text is pre-trained in the server, and each content keyword is identified through the preset generation model, and texts corresponding to each content keyword are generated as content texts. At the same time, each editing keyword is identified, and texts corresponding to each editing keyword are generated as editing texts. Then, the content text and the editing file are combined to determine whether there are logical contradictions or sentences that are not smoothly connected between the two. If so, they are corrected based on the correct sentences to obtain the modified text to be adjusted. Thereafter, the text to be adjusted is adjusted according to the user preferences reflected by the personality keywords. For example, if the personality keywords reflect the user's preference for cuteness, then modal particles such as "ah, oh, yeah" are added to the text to be adjusted, or reduplication is added, such as "take a car, walk a walk", etc. The adjusted text to be adjusted is formed into the note content.

[0123] Furthermore, in order to generate high-quality texts through the preset generation model, this embodiment uses a large number of popular notes as training samples and optimizes the preset generation model at intervals of a preset period. Specifically, the step of generating content texts corresponding to each of the content keywords and editing texts corresponding to each of the editing keywords based on the preset generation model includes:

[0124] Step a1, obtaining popularity information of notes published within a preset period, and screening out target notes from the published notes according to the popularity information;

[0125] Step a2, dividing each of the target notes into training samples and test samples, and transmitting each training sample to a preset generation model for optimization training;

[0126] Step a3, after the training times reach the preset times, the preset generation model is tested based on the test sample, and the loss function value of the preset generation model is calculated, and the preset generation model is optimized based on the loss function value.

[0127] Furthermore, at every preset period, the Xiaohongshu notes posted by other users within the preset period are taken as published notes, and the popularity information of the published notes is obtained, and the popularity information includes but is not limited to the number of views, the number of likes, the number of reposts, and the number of comments. Then, based on the popularity information, the published notes posted within the preset period are screened, and the published notes with high popularity are screened out as target notes. Among them, the higher the number of views and / or the higher the number of likes and / or the higher the number of reposts and / or the higher the number of comments, the higher the popularity. In addition, a preset number threshold can be set, and when the above number is greater than the preset number threshold, the relevant published notes are screened as target notes. Alternatively, a ranking mechanism can be set. For example, if the ranking number is set to 10,000, the above number of each published note can be compared respectively, and the top 10,000 published notes can be screened out as target notes.

[0128] Furthermore, the screened published notes are divided into training samples and test samples according to a preset ratio, for example, 70% of the training samples and 30% of the test samples. The divided training samples are then transferred to the preset generation model to train the preset generation model. In addition, a preset number of times representing the end of training is preset. When it is detected that the number of optimization trainings reaches the preset number, keywords are extracted from each test sample, and the extracted keywords are transferred to the preset generation model, and the preset generation model generates test notes based on the keywords. Then, the test features of the test notes are identified, and the test features reflect the key information points of the test notes. The test samples are also pre-annotated with label features that reflect their key information points. The loss function value of the preset generation model is calculated based on the label features and the test features, and the preset generation model is optimized by the calculated loss function value. The specific calculation formula is shown in the following formula (5).

[0129]

[0130] Among them, L is the loss function value, exp is the exponential function, h is the model parameter of the preset generation model, m is the number of test samples, γ is the label feature corresponding to the test sample, and γi is the test feature generated based on the i-th test. By superimposing the error between the test feature generated by each test sample and its corresponding label feature, the calculation of the loss function value is more accurate.

[0131] A preset function value characterizing the size of the loss function value is pre-set, and the calculated loss function value is compared with the preset function value to determine whether the loss function value is less than the preset function threshold. If it is less than, it means that the difference between the test features of the test notes generated by the preset generation model and the label features annotated by the test sample is small, and the optimized training of the preset generation model has achieved a good learning effect on the published notes within the preset period, so the optimization training of the preset generation model is completed. On the contrary, if it is determined that the loss function value is not less than the preset function threshold, it means that the difference between the test features and the label features is large, and optimization training needs to be continued. After the number of optimization training reaches the preset number again, the loss function value is calculated again for judgment, and this cycle is repeated until the calculated loss function value is less than the preset function threshold, and the optimization training of the preset generation model is completed, so that the optimized preset generation model can be used to process each content keyword, each edge keyword and each personality keyword to generate a note content with high popularity.

[0132] In this embodiment, the preset generation model is used to identify content keywords and editing keywords, generate text that meets the user's content needs, and adjust the style of the generated text according to the personality keywords to obtain note content that meets the user's style needs. At the same time, an optimization mechanism for the preset generation model is set, and notes with high popularity are obtained at preset intervals to optimize the preset generation model, and after testing that the generation effect of the optimized and trained preset generation model is improved, the preset generation model optimization training is completed. In this way, the generated note content not only meets the user's content and style needs, but also makes the generated note content popular.

[0133] In addition, an embodiment of the present invention also provides a note generation system. Figure 4 , Figure 4 This is a schematic diagram of the structure of the device hardware operating environment involved in the note generation system implementation scheme of the present invention.

[0134] like Figure 4As shown, the note generation system may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the user interface 1003 may optionally include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 1005 may be a high-speed RAM memory, or it may be a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may optionally be a storage device independent of the aforementioned processor 1001.

[0135] Those skilled in the art will understand that Figure 4 The hardware structure of the note generation system shown in the figure does not constitute a limitation of the note generation system, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0136] like Figure 4 As shown, the memory 1005 as a readable storage medium may include an operating system, a network communication module, a user interface module and a control program. The operating system is a program for managing and controlling the note generation system and software resources, and supports the operation of the network communication module, the user interface module, the control program and other programs or software; the network communication module is used to manage and control the network interface 1004; the user interface module is used to manage and control the user interface 1003.

[0137] exist Figure 4 In the hardware structure of the note generation system shown, the network interface 1004 is mainly used to connect to the backend server and communicate data with the backend server; the user interface 1003 is mainly used to connect to the client (user end) and communicate data with the client; the processor 1001 can call the control program stored in the memory 1005 and perform the following operations:

[0138] Obtaining note keywords and pictures to be edited from information uploaded by the user based on a preset template, and obtaining content keywords corresponding to the note keywords based on a preset vocabulary;

[0139] Identifying an editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information;

[0140] Acquire the user's browsing history information, identify personality keywords from the browsing history information, and generate note content based on each of the content keywords, each of the editing keywords, and each of the personality keywords;

[0141] A note title is generated based on the note content and the display information, and the note title, the note content and the display information are generated as a complete note.

[0142] Furthermore, the step of identifying the editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information includes:

[0143] Identify the person in the picture to be edited, obtain the editing keyword, and identify the area to be added and the characteristic elements corresponding to the picture to be edited;

[0144] Determine the pixel parameters of the editing keyword according to the characteristic elements, and add the editing keyword having the pixel parameters to the to-be-added area of ​​the to-be-edited picture to form the to-be-encrypted picture;

[0145] The picture to be encrypted is encrypted to form display information.

[0146] Furthermore, the step of encrypting the picture to be encrypted to form display information includes:

[0147] Forming the picture to be encrypted into multiple sub-pictures, and obtaining characteristic parameters of each of the sub-pictures;

[0148] Based on the pixel mean of each sub-image and each characteristic parameter, the signature parameter of the image to be encrypted is calculated. The formula for calculating the signature parameter is:

[0149]

[0150] Randomly obtain pixel points from the image to be encrypted, and determine the coordinate values ​​of the pixel points;

[0151] An encryption password is calculated based on the signature parameter and the coordinate value, and the image to be encrypted is encrypted based on the encryption password to form display information. The formula for calculating the encryption password is:

[0152]

[0153] Among them, S is the signature parameter, ∑ is the accumulation operator, n is the number of word images, f(i) is the pixel mean function, is the (i-1)th characteristic parameter, k1 and k2 are passwords, π is the symbol of pi, sin is the sine function, cos is the dotted line function, and (x, y) are the coordinate values.

[0154] Furthermore, the steps of identifying the person in the picture to be edited, obtaining the editing keyword, and identifying the area to be added and the characteristic elements corresponding to the picture to be edited include:

[0155] Identify a prominent area containing a person in the picture to be edited, and the attributes of the person contained therein, and obtain a keyword corresponding to the attribute as the editing keyword;

[0156] The area outside the prominent area that does not contain the person is regarded as a non-salient area, and the area to be added is determined from the non-salient area according to the note keyword;

[0157] Acquire a first pixel mean value of the salient area and a second pixel mean value of the non-salient area, and generate an absolute value of a difference between the first pixel mean value and the second pixel mean value;

[0158] The characteristic elements of the image to be edited are determined according to the magnitude relationship between the absolute value of the difference and a preset first threshold.

[0159] Furthermore, the step of identifying personality keywords from the historical browsing information includes:

[0160] Filtering out historical attribute keywords from each of the historical browsing information, and forming each of the historical attribute keywords into a first feature vector;

[0161] The second feature vectors corresponding to the preset personality keywords are obtained, and the first feature vector and each of the second feature vectors are calculated respectively to generate a vector similarity value. The calculation formula is:

[0162] Wi=

[0163] (α*p*ln (α*p+(1-α)*qi) ) / ((1-α)*qi*ln (α*qi+(1-α)*p) );

[0164] Each of the vector similarity values ​​is compared with a preset second threshold value, a target vector similarity value greater than the preset second threshold value is screened out, and a preset personality keyword corresponding to the target vector similarity value is determined as the personality keyword.

[0165] Among them, Wi is the vector similarity value, α is the historical key parameter corresponding to the historical attribute keyword, p is the first eigenvector, In is the natural logarithm, and qi is the second eigenvector corresponding to the i-th preset personality keyword.

[0166] Furthermore, the step of generating note content based on each of the content keywords, each of the editing keywords and each of the personality keywords includes:

[0167] Generate content texts corresponding to each of the content keywords and edit texts corresponding to each of the edit keywords based on a preset generation model;

[0168] The content text and the edit text are combined into a text to be adjusted, and the text to be adjusted is adjusted based on the personality keywords to obtain the note content.

[0169] Furthermore, the step of generating the content text corresponding to each of the content keywords and the editing text corresponding to each of the editing keywords based on the preset generation model includes:

[0170] Obtaining popularity information of published notes within a preset period, and filtering out target notes from the published notes according to the popularity information;

[0171] Dividing each of the target notes into training samples and test samples, and transmitting each training sample to a preset generation model for optimization training;

[0172] After the number of training times reaches the preset number of times, the preset generation model is tested based on the test sample, and the loss function value of the preset generation model is calculated, and the preset generation model is optimized based on the loss function value. The calculation formula is:

[0173]

[0174] Among them, L is the loss function value, exp is the exponential function, h is the model parameter of the preset generation model, m is the number of test samples, γ is the label feature corresponding to the test sample, and γi is the test feature generated based on the i-th test.

[0175] Furthermore, the step of acquiring content keywords corresponding to the note keywords based on a preset vocabulary includes:

[0176] Identify the subject words in the note keywords, and identify the sub-word library corresponding to the subject words in the preset word library;

[0177] The note keywords are converted into a first feature matrix, and the second feature matrix corresponding to each feature phrase in the sub-lexicon is obtained, and the overlap value between the first feature matrix and each second feature matrix is ​​calculated. The calculation formula is:

[0178]

[0179] Determine a target coincidence value with the largest value among the coincidence values, and determine a characteristic phrase corresponding to the target coincidence value as the content keyword;

[0180] Among them, G represents the overlap value, max is the maximum value, B is the first feature matrix of note keyword conversion, Di is the i-th second feature matrix corresponding to the i-th feature phrase, and cos is the cosine function.

[0181] The specific implementation of the note generation system of the present invention is basically the same as the various embodiments of the above-mentioned note generation method, and will not be repeated here.

[0182] The embodiment of the present invention further provides a readable storage medium having a control program stored thereon, and when the control program is executed by a processor, the steps of the note generation method described above are implemented.

[0183] The readable storage medium of the present invention may be a computer-readable storage medium, and its specific implementation is basically the same as the above-mentioned embodiments of the note generation method, and will not be repeated here.

[0184] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims. All equivalent structures or equivalent process changes made using the contents of the specification and drawings of the present invention, or directly or indirectly used in other related technical fields, are protected by the present invention.

Claims

1. A note generation method, characterized in that: The note generation method comprises: Obtaining note keywords and pictures to be edited from information uploaded by the user based on a preset template, and obtaining content keywords corresponding to the note keywords based on a preset vocabulary; Identifying an editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information; Acquire the user's browsing history information, identify personality keywords from the browsing history information, and generate note content based on each of the content keywords, each of the editing keywords, and each of the personality keywords; generating a note title based on the note content and the display information, and generating a complete note by combining the note title, the note content and the display information; The step of acquiring content keywords corresponding to the note keywords based on a preset vocabulary includes: Identify the subject words in the note keywords, and identify the sub-word library corresponding to the subject words in the preset word library; The note keywords are converted into a first feature matrix, and the second feature matrix corresponding to each feature phrase in the sub-lexicon is obtained, and the overlap value between the first feature matrix and each second feature matrix is ​​calculated. The calculation formula is: Determine a target coincidence value with the largest value among the coincidence values, and determine a characteristic phrase corresponding to the target coincidence value as the content keyword; Where G represents the overlap value, min represents the minimum difference between the first feature matrix and each second feature matrix, n represents the number of elements in the first feature matrix or the second feature matrix, and B k The kth element value in the first feature matrix representing the note keyword conversion, Di k represents the kth element value in the i-th second feature matrix corresponding to the i-th feature phrase, and cos is the cosine function.

2. The note generation method according to claim 1, wherein: The step of identifying the editing keyword corresponding to the picture to be edited, and encrypting and fusing the editing keyword with the picture to be edited to form display information comprises: Identify the person in the picture to be edited, obtain the editing keyword, and identify the area to be added and the characteristic elements corresponding to the picture to be edited; Determine the pixel parameters of the editing keyword according to the characteristic elements, and add the editing keyword having the pixel parameters to the to-be-added area of ​​the to-be-edited picture to form the to-be-encrypted picture; The picture to be encrypted is encrypted to form display information.

3. The note generation method according to claim 2, wherein: The step of encrypting the picture to be encrypted to form display information comprises: Forming the picture to be encrypted into multiple sub-pictures, and obtaining characteristic parameters of each of the sub-pictures; Based on the pixel mean of each sub-image and each characteristic parameter, the signature parameter of the image to be encrypted is calculated. The formula for calculating the signature parameter is: Randomly obtain pixel points from the image to be encrypted, and determine the coordinate values ​​of the pixel points; An encryption password is calculated based on the signature parameter and the coordinate value, and the image to be encrypted is encrypted based on the encryption password to form display information. The formula for calculating the encryption password is: Among them, S is the signature parameter, ∑ is the accumulation operator, n is the number of sub-images, f(i) is the pixel mean function, is the i-1th characteristic parameter, k1 and k2 are passwords, π is the symbol of pi, sin is the sine function, cos is the cosine function, and (x, y) is the coordinate value.

4. The note generation method according to claim 2, wherein: The steps of identifying the person in the picture to be edited, obtaining the editing keyword, and identifying the area to be added and the characteristic elements corresponding to the picture to be edited include: Identify a prominent area containing a person in the picture to be edited, and the attributes of the person contained therein, and obtain a keyword corresponding to the attribute as the editing keyword; The area outside the prominent area that does not contain the person is regarded as a non-salient area, and the area to be added is determined from the non-salient area according to the note keyword; Acquire a first pixel mean value of the salient area and a second pixel mean value of the non-salient area, and generate an absolute value of a difference between the first pixel mean value and the second pixel mean value; The characteristic elements of the image to be edited are determined according to the magnitude relationship between the absolute value of the difference and a preset first threshold.

5. The note generation method according to any one of claims 1 to 4, characterized in that: The step of identifying personality keywords from the historical browsing information includes: Filtering out historical attribute keywords from each of the historical browsing information, and forming each of the historical attribute keywords into a first feature vector; The second feature vectors corresponding to the preset personality keywords are obtained, and the first feature vector and each of the second feature vectors are calculated respectively to generate a vector similarity value. The calculation formula is: Comparing each of the vector similarity values ​​with a preset second threshold value, selecting a target vector similarity value greater than the preset second threshold value, and determining a preset personality keyword corresponding to the target vector similarity value as the personality keyword; Among them, Wi is the vector similarity value, α is the historical key parameter corresponding to the historical attribute keyword, p is the first eigenvector, ln is the natural logarithm, qi is the second eigenvector corresponding to the i-th preset personality keyword, and |·| is the modulus of the vector.

6. The note generation method according to any one of claims 1 to 4, characterized in that: The step of generating note content based on each of the content keywords, each of the editing keywords and each of the personality keywords comprises: Generate content texts corresponding to each of the content keywords and edit texts corresponding to each of the edit keywords based on a preset generation model; The content text and the edit text are combined into a text to be adjusted, and the text to be adjusted is adjusted based on the personality keywords to obtain the note content.

7. The note generation method according to claim 6, wherein: The step of generating content texts corresponding to the content keywords and editing texts corresponding to the editing keywords based on the preset generation model includes: Obtaining popularity information of published notes within a preset period, and filtering out target notes from the published notes according to the popularity information; Dividing each of the target notes into training samples and test samples, and transmitting each training sample to a preset generation model for optimization training; After the number of training times reaches the preset number of times, the preset generation model is tested based on the test sample, and the loss function value of the preset generation model is calculated, and the preset generation model is optimized based on the loss function value. The calculation formula is: Among them, L is the loss function value, exp is the exponential function, h is the model parameter of the preset generation model, m is the number of test samples, γ is the label feature corresponding to the test sample, and γi is the test feature generated based on the i-th test sample.

8. A note generation system, characterized in that: The note generation system includes: a memory, a processor, a communication bus, and a control program stored in the memory: The communication bus is used to realize the connection and communication between the processor and the memory; The processor is used to execute the control program to implement the steps of the note generation method as described in any one of claims 1-7.

9. A readable storage medium, characterized in that: The readable storage medium stores a control program, and when the control program is executed by the processor, the steps of the note generation method according to any one of claims 1 to 7 are implemented.

Images