Zero-code application generation and data integration method based on natural language

Abstract

The present invention discloses a zero-code application generation and data integration method based on natural language, which relates to the technical field of application generation and data integration, including analyzing the similarity of text information between different historical input records of users to obtain approximate historical input records; evaluating the data difference of historical output data between historical input records and approximate historical input records to obtain target approximate historical input records; obtaining keywords in the text information of historical input records, analyzing the semantic similarity of keywords in the text information between historical input records and target approximate historical input records to obtain target keyword replacement data; obtaining natural language information input by users on a platform in the current period to obtain target keyword replacement data of users, optimizing and adjusting the information content of users' input in the current period, generating target data, and displaying it to users on the platform.

Description

Technical Field

[0001] The present invention relates to the technical field of application generation and data integration, and in particular to a natural language-based zero-code application generation and data integration method. Background Art

[0002] Natural language refers to the language used by humans for daily communication, usually expressed in spoken and written form. The use of natural language technology allows non-technical users to perform data operations and generate applications in natural language, reducing the user's need for programming languages ​​and technical knowledge, allowing users to participate more extensively in application development and data processing. By understanding technology through natural language, users can quickly describe their needs, and by parsing these needs, the framework of the application can be automatically generated, further speeding up the application development process. At the same time, through natural language, users can more naturally describe data sources and integration requirements, simplifying the data fusion process.

[0003] Application generation and data integration can be quickly performed through natural language. However, in the actual application generation and data integration process, the platform needs to perform corresponding operations based on the natural language input by the user. However, since natural language itself is very vague, and different users have different expression habits and styles, different users may express the same intention in very different ways. It is difficult for the platform to accurately understand the user's true meaning, which will not only reduce the user experience, but may also cause users to lose trust in the platform, thereby affecting the brand image and user trust. Summary of the invention

[0004] The purpose of the present invention is to provide a natural language-based zero-code application generation and data integration method to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solution: a zero-code application generation and data integration method based on natural language, the method comprising:

[0006] Step S100: constructing a natural language processing cloud platform, obtaining the user's historical input records, extracting text information from the historical input records, analyzing the similarity of text information between different historical input records of the user, and obtaining approximate historical input records;

[0007] Step S200: obtaining historical output data output by the platform according to the historical input record, obtaining an approximate historical input record of the historical input record, evaluating the data difference of the historical output data between the historical input record and the approximate historical input record, and obtaining a target approximate historical input record;

[0008] Step S300: Acquire keywords in the text information of the historical input record, analyze the semantic similarity between the keywords in the text information of the historical input record and the target similar historical input record, and obtain target keyword replacement data;

[0009] Step S400: Acquire the natural language information input by the user on the platform in the current period, obtain the user's target keyword replacement data, optimize and adjust the information content input by the user in the current period, generate target data, and display it to the user on the platform;

[0010] Furthermore, step S100 includes:

[0011] Step S101: construct a natural language processing cloud platform, obtain and record the natural language input by the user on the platform when performing application generation and data integration, obtain the user's historical input record, and extract text information from the historical input record;

[0012] Step S102: obtaining text information of each historical input record of the user, preprocessing the text information in the historical input record, and obtaining a keyword group in the historical input record;

[0013] Step S103: Obtain the feature value DF(a) of the ath keyword in the keyword group;

[0014] Step S104: analyzing the similarity of text information between different historical input records of the user, wherein the similarity of text information between the bth historical input record and the cth historical input record is analyzed to obtain the text information similarity value between the cth historical input record and the bth historical input record;

[0015] Step S105: When the text information approximation value P C,B If the time point of the c-th historical input record is less than the time point of the b-th historical input record, the b-th historical input record is recorded as the approximate historical input record of the c-th historical input record; otherwise, the c-th historical input record is recorded as the approximate historical input record of the b-th historical input record.

[0016] Furthermore, in step S103, the specific process of obtaining the feature value of the ath keyword includes:

[0017] Calculate the characteristic inverse document frequency of each keyword in the keyword group of the historical input record, where the characteristic inverse document frequency F(a) of the ath keyword in the keyword group is:

[0018]

[0019] Among them, N sum Represents the total number of historical input records; N a It is represented by the total number of historical input records containing the ath keyword;

[0020] Calculate the feature value DF(a) of the ath keyword in the historical input record:

[0021] DF(a)=F(a)×Q a ,

[0022] Among them, Q a It is represented by the frequency of the ath keyword appearing in the text information in the historical input records.

[0023] Furthermore, in step S104, the specific process of obtaining the approximate value of the text information of the cth historical input record and the bth historical input record includes:

[0024] Analyzing the similarity of text information between different historical input records of the user, wherein the specific process of analyzing the similarity of text information between the bth historical input record and the cth historical input record is to obtain the keyword group of the cth historical input record and the bth historical input record;

[0025] Obtaining the characteristic values ​​of each keyword in the keyword group of the cth historical input record and the bth historical input record, and converting the text information in the cth historical input record and the bth historical input record into a numerical vector group according to the characteristic values;

[0026] Calculate the text information approximation P of the cth historical input record and the bth historical input record C,B :

[0027]

[0028] Where y represents the total number of elements in the numerical vector group of the cth historical input record; B z The zth element in the numeric vector group of the bth historical input record; C z The zth element in the numeric vector group represented as the cth historical input record.

[0029] Further, step S200 includes:

[0030] Step S201: When a user inputs text information in a natural language into the platform, the platform generates content according to the text information input by the user, and outputs the generated content to the user on the platform, thereby obtaining output data of the platform;

[0031] Step S202: obtaining historical output data output by the platform according to the historical input record, obtaining an approximate historical input record of the historical input record, evaluating the data difference of the historical output data between the historical input record and a certain approximate historical input record of the historical input record, wherein the specific evaluation process is to extract the code generated by the platform from the historical output data of the historical input record and a certain approximate historical input record respectively;

[0032] Step S203: using a code formatting tool to format the generated code, remove comments and blank lines in the code, and preprocess the code;

[0033] Step S204: Analyze the string similarity between each string in the code of the historical input record and a similar historical input record, and obtain the string similarity between each string in the historical input record and each string in the similar historical input record;

[0034] Step S205: Evaluate the data difference of the historical output data between the historical input record and the approximate historical input record, and obtain the target approximate historical input record of the historical input record;

[0035] In the above steps, the data difference of the historical output data between the historical input record and the approximate historical input record is evaluated because the text information input by the user between the historical input record and a certain approximate historical input record is judged to be similar, but similar content is input multiple times, which means that the content output by the text information platform using natural language input by the user does not meet the actual needs of the user. Therefore, by evaluating the data difference of the historical output data between the historical input record and the approximate historical input record, as well as the time sequence of the historical input records, it is possible to quickly find the content that the user actually wants to input, thereby improving the subsequent optimization effect of the user input content.

[0036] Furthermore, in step S204, the specific process of obtaining the string similarity between each string of the historical input record and each string of a similar historical input record includes:

[0037] Analyze the string similarity between each string in the code of a certain approximate historical input record of the historical input record, wherein the string similarity between the d-th string in the code of the historical input record and the p-th string in the code of a certain approximate historical input record is specifically analyzed as follows: create a two-dimensional array dp, set the size of the two-dimensional array dp to m×n, wherein m represents the total number of characters in the d-th string in the historical input record, n represents the total number of characters in the p-th string in a certain approximate historical input record, and the d-th string and the p-th string are the strings corresponding to the historical input record and the approximate historical input record;

[0038] Among them, the feature edit distance of the element dp[i,j] in the i-th row and j-th column in the two-dimensional array dp is calculated. The specific calculation formula of the feature edit distance of dp[i,j] is:

[0039]

[0040] Among them, d i Represented as the i-th character in the d-th string; p j It represents the jth character in the pth string; dp[i-1,j-1] represents the i-1th row and j-1th column element in the two-dimensional array dp; dp[i,j-1] represents the i-th row and j-1th column element in the two-dimensional array dp; dp[i-1,j] represents the i-1th row and j-1th column element in the two-dimensional array dp;

[0041] Get the feature edit distance of dp[m,n] in the two-dimensional array dp, and calculate the string similarity R between the dth string of the code of the historical input record and the pth string in the code of a similar historical input record d,p :

[0042]

[0043] Where |d| represents the length of the dth string of the code in the historical input record; |p| represents the length of the pth string in the code of a certain approximate historical input record;

[0044] Get the string similarity between each string of the historical input record and each string of a similar historical input record.

[0045] Furthermore, the specific process of evaluating the data difference of the historical output data between the historical input record and the approximate historical input record in step S205 includes:

[0046] Obtain the average value of the string similarity between each string in the code of the historical input record and each string in the code of a certain approximate historical input record, and use the reciprocal of the average value as the output difference value between the historical input record and the certain approximate historical input record;

[0047] The output difference value between the historical input record and several corresponding historical approximate input records is obtained, and the historical approximate input record corresponding to the maximum value of the output difference value is recorded as the target approximate historical input record of the historical input record.

[0048] Furthermore, step S300 includes:

[0049] Step S301: Acquire a keyword group of the text information of the user's historical input record, and acquire a keyword group of the target historical input record that is similar to the historical input record;

[0050] Step S302: Analyze the semantic similarity between the target approximate historical input record and the different keywords in the keyword group between the historical input record, and obtain the characteristic semantic similarity value of the αth keyword of the historical input record and each keyword in the keyword group of the target approximate historical input record;

[0051] Step S303: When a keyword in the keyword group of the target approximate historical input record has a maximum feature semantic approximation value with the αth keyword among the feature semantic approximation values ​​of each keyword, the keyword is used as the target replacement keyword of the αth keyword;

[0052] Step S304: obtaining a plurality of target replacement keywords of a plurality of keywords, and collecting the plurality of target replacement keywords of a plurality of keywords to obtain the user's target keyword replacement data.

[0053] Furthermore, in step S302, for the αth keyword in the historical input record, the specific process of obtaining the feature semantic approximation value of each keyword in the keyword group of the target similar historical input record includes:

[0054] Convert the historical input records and each keyword in the keyword group of the target similar historical input records into vectors respectively;

[0055] Calculate the feature semantic approximation between the historical input record and each keyword in the target approximate historical input record, where the feature semantic approximation U of the αth keyword in the historical input record and the βth keyword in the target approximate historical input record is α,β :

[0056]

[0057] Among them, Vα Represented as the vector corresponding to the αth keyword; V β It is represented as the vector corresponding to the βth keyword.

[0058] Furthermore, step S400 includes:

[0059] Step S401: acquiring natural language information input by users on the platform in the current period, and converting the natural language information input by users in the current period into text information;

[0060] Step S402: Acquire the user's target keyword replacement data, extract keywords from the text information input by the user in the current period, and replace the keywords in the text information input by the user in the current period with the target replacement keywords corresponding to the keywords based on the target keyword replacement data, optimize and adjust the information content input by the user in the current period, and the platform outputs the target data based on the optimized and adjusted text information, and displays the target data to the user on the platform.

[0061] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention realizes application generation and data integration based on the personalization of natural language input by users, and considers that different users have different expression methods and expression habits during the actual user input process. Therefore, when analyzing the natural language input by users, the present invention uses the user's own historical records as a reference to optimize and adjust the expression of the user input, so that the platform can accurately understand the content input by users, achieve truly personalized analysis of users, improve the user experience, and enhance dependence on the platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a method flow chart of the natural language-based zero-code application generation and data integration method of the present invention;

[0063] Figure 2 It is a data difference assessment flow chart of the zero-code application generation and data integration method based on natural language of the present invention. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] Example: Figure 1-Figure 2As shown, the present invention provides a technical solution, a zero-code application generation and data integration method based on natural language, the method comprising:

[0066] Step S100: constructing a natural language processing cloud platform, obtaining the user's historical input records, extracting text information from the historical input records, analyzing the similarity of text information between different historical input records of the user, and obtaining approximate historical input records;

[0067] Wherein, step S100 includes:

[0068] Step S101: construct a natural language processing cloud platform, obtain and record the natural language input by the user on the platform when performing application generation and data integration, obtain the user's historical input record, and extract text information from the historical input record;

[0069] Step S102: obtaining text information of each historical input record of the user, preprocessing the text information in the historical input record, and obtaining a keyword group in the historical input record;

[0070] For example, the preprocessing process includes,cleaning the text, removing useless tokens and stop words, etc.;

[0071] Step S103: Obtain the feature value DF(a) of the ath keyword in the keyword group;

[0072] Step S104: analyzing the similarity of text information between different historical input records of the user, wherein the similarity of text information between the bth historical input record and the cth historical input record is analyzed to obtain the text information similarity value between the cth historical input record and the bth historical input record;

[0073] Step S105: When the text information approximation value P C,B is greater than a preset text information approximation threshold, determining that the text information between the c-th historical input record and the b-th historical input record is approximate, obtaining the time points of the user input information corresponding to the c-th historical input record and the b-th historical input record, and when the time point corresponding to the c-th historical input record is less than the time point of the b-th historical input record, recording the b-th historical input record as an approximate historical input record of the c-th historical input record, otherwise, recording the c-th historical input record as an approximate historical input record of the b-th historical input record;

[0074] In step S103, the specific process of obtaining the feature value of the ath keyword includes:

[0075] Calculate the characteristic inverse document frequency of each keyword in the keyword group of the historical input record, where the characteristic inverse document frequency F(a) of the ath keyword in the keyword group is:

[0076]

[0077] Among them, N sum Represents the total number of historical input records; N a It is represented by the total number of historical input records containing the ath keyword;

[0078] For example, the total number N of each historical input record is expressed as 100; the total number N2 of historical input records containing the second keyword "movie" is expressed as 300; calculate the characteristic inverse document frequency F(2) of the second keyword in the keyword group:

[0079]

[0080] Calculate the feature value DF(a) of the ath keyword in the historical input record:

[0081] DF(a)=F(a)×Q a ,

[0082] Among them, Q a It is represented by the frequency of occurrence of the ath keyword in the text information in the historical input record;

[0083] In step S104, the specific process of obtaining the approximate value of the text information of the cth historical input record and the bth historical input record includes:

[0084] Analyzing the similarity of text information between different historical input records of the user, wherein the specific process of analyzing the similarity of text information between the bth historical input record and the cth historical input record is to obtain the keyword group of the cth historical input record and the bth historical input record;

[0085] Obtaining the characteristic values ​​of each keyword in the keyword group of the cth historical input record and the bth historical input record, and converting the text information in the cth historical input record and the bth historical input record into a numerical vector group according to the characteristic values;

[0086] Calculate the text information approximation P of the cth historical input record and the bth historical input record C,B :

[0087]

[0088] Where y represents the total number of elements in the numerical vector group of the cth historical input record; B z The zth element in the numeric vector group of the bth historical input record; C zThe zth element in the numeric vector group represented by the cth historical input record;

[0089] Step S200: obtaining historical output data output by the platform according to the historical input record, obtaining an approximate historical input record of the historical input record, evaluating the data difference of the historical output data between the historical input record and the approximate historical input record, and obtaining a target approximate historical input record;

[0090] Wherein, step S200 includes:

[0091] Step S201: When a user inputs text information in a natural language into the platform, the platform generates content according to the text information input by the user, and outputs the generated content to the user on the platform, thereby obtaining output data of the platform;

[0092] Step S202: obtaining historical output data output by the platform according to the historical input record, obtaining an approximate historical input record of the historical input record, evaluating the data difference of the historical output data between the historical input record and a certain approximate historical input record of the historical input record, wherein the specific evaluation process is to extract the code generated by the platform from the historical output data of the historical input record and a certain approximate historical input record respectively;

[0093] Step S203: using a code formatting tool to format the generated code, remove comments and blank lines in the code, and preprocess the code;

[0094] Step S204: Analyze the string similarity between each string in the code of the historical input record and a similar historical input record, and obtain the string similarity between each string in the historical input record and each string in the similar historical input record;

[0095] Step S205: Evaluate the data difference of the historical output data between the historical input record and the approximate historical input record, and obtain the target approximate historical input record of the historical input record;

[0096] The specific process of obtaining the string similarity between each string of the historical input record and each string of a similar historical input record in step S204 includes:

[0097] Analyze the string similarity between each string in the code of a certain approximate historical input record of the historical input record, wherein the string similarity between the d-th string in the code of the historical input record and the p-th string in the code of a certain approximate historical input record is specifically analyzed as follows: create a two-dimensional array dp, set the size of the two-dimensional array dp to m×n, wherein m represents the total number of characters in the d-th string in the historical input record, n represents the total number of characters in the p-th string in a certain approximate historical input record, and the d-th string and the p-th string are the strings corresponding to the historical input record and the approximate historical input record;

[0098] Among them, the feature edit distance of the element dp[i,j] in the i-th row and j-th column in the two-dimensional array dp is calculated. The specific calculation formula of the feature edit distance of dp[i,j] is:

[0099]

[0100] Among them, d i Represented as the i-th character in the d-th string; p j It represents the jth character in the pth string; dp[i-1,j-1] represents the i-1th row and j-1th column element in the two-dimensional array dp; dp[i,j-1] represents the i-th row and j-1th column element in the two-dimensional array dp; dp[i-1,j] represents the i-1th row and j-1th column element in the two-dimensional array dp;

[0101] Get the feature edit distance of dp[m,n] in the two-dimensional array dp, and calculate the string similarity R between the dth string of the code of the historical input record and the pth string in the code of a similar historical input record d,p :

[0102]

[0103] Where |d| represents the length of the dth string of the code in the historical input record; |p| represents the length of the pth string in the code of a certain approximate historical input record;

[0104] Get the string similarity between each string of the historical input record and each string of a similar historical input record;

[0105] The specific process of evaluating the data difference of the historical output data between the historical input record and the similar historical input record in step S205 includes:

[0106] Obtain the average value of the string similarity between each string in the code of the historical input record and each string in the code of a certain approximate historical input record, and use the reciprocal of the average value as the output difference value between the historical input record and the certain approximate historical input record;

[0107] Obtain output difference values ​​between the historical input record and a number of corresponding historical approximate input records, and record the historical approximate input record corresponding to the maximum value of the output difference value as the target approximate historical input record of the historical input record;

[0108] Step S300: Acquire keywords in the text information of the historical input record, analyze the semantic similarity between the keywords in the text information of the historical input record and the target similar historical input record, and obtain target keyword replacement data;

[0109] Wherein, step S300 includes:

[0110] Step S301: Acquire a keyword group of the text information of the user's historical input record, and acquire a keyword group of the target historical input record that is similar to the historical input record;

[0111] Step S302: Analyze the semantic similarity between the target approximate historical input record and the different keywords in the keyword group between the historical input record, and obtain the characteristic semantic similarity value of the αth keyword of the historical input record and each keyword in the keyword group of the target approximate historical input record;

[0112] Step S303: When a keyword in the keyword group of the target approximate historical input record has a maximum feature semantic approximation value with the αth keyword among the feature semantic approximation values ​​of each keyword, the keyword is used as the target replacement keyword of the αth keyword;

[0113] Step S304: obtaining a plurality of target replacement keywords of a plurality of keywords, and collecting the plurality of target replacement keywords of the plurality of keywords to obtain the user's target keyword replacement data;

[0114] Among them, in step S302, for the αth keyword of the historical input record, the specific process of obtaining the feature semantic approximation value of each keyword in the keyword group of the target similar historical input record includes:

[0115] Convert the historical input records and each keyword in the keyword group of the target similar historical input records into vectors respectively;

[0116] Calculate the feature semantic approximation between the historical input record and each keyword in the target approximate historical input record, where the feature semantic approximation U of the αth keyword in the historical input record and the βth keyword in the target approximate historical input record is α,β :

[0117]

[0118] Among them, V α Represented as the vector corresponding to the αth keyword; V β It is represented as the vector corresponding to the βth keyword;

[0119] Step S400: Acquire the natural language information input by the user on the platform in the current period, obtain the user's target keyword replacement data, optimize and adjust the information content input by the user in the current period, generate target data, and display it to the user on the platform;

[0120] Wherein, step S400 includes:

[0121] Step S401: acquiring natural language information input by users on the platform in the current period, and converting the natural language information input by users in the current period into text information;

[0122] Step S402: Acquire the user's target keyword replacement data, extract keywords from the text information input by the user in the current period, and replace the keywords in the text information input by the user in the current period with the target replacement keywords corresponding to the keywords based on the target keyword replacement data, optimize and adjust the information content input by the user in the current period, and the platform outputs the target data based on the optimized and adjusted text information, and displays the target data to the user on the platform.

[0123] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A zero-code application generation and data integration method based on natural language, characterized in that: The method comprises: Step S100: constructing a natural language processing cloud platform, obtaining historical input records of the user, extracting text information from the historical input records, analyzing the similarity of text information between different historical input records of the user, and obtaining approximate historical input records, wherein the time point of the user input information corresponding to the historical input record is less than the time point of the user input information corresponding to the approximate historical input record of the historical input record; Step S200: acquiring historical output data output by the platform according to the historical input record, acquiring an approximate historical input record of the historical input record, evaluating the data difference of the historical output data between the historical input record and the approximate historical input record, and obtaining a target approximate historical input record; Step S300: obtaining keywords in the text information of the historical input record, analyzing the semantic similarity of the keywords in the text information between the historical input record and the target approximate historical input record, and obtaining target keyword replacement data, wherein the target keyword replacement data includes target replacement keywords corresponding to the keywords in the historical input record, and the target replacement keywords are obtained from the target approximate historical input record; Step S400: Acquire the natural language information input by the user on the platform during the current cycle, obtain the target keyword replacement data of the user, optimize and adjust the information content input by the user during the current cycle, generate target data, and display it to the user on the platform.

2. The natural language-based zero-code application generation and data integration method according to claim 1, characterized in that: The step S100 includes: Step S101: construct a natural language processing cloud platform, obtain and record the natural language input by the user on the platform when performing application generation and data integration, obtain the user's historical input record, and extract text information from the historical input record; Step S102: obtaining text information of each historical input record of the user, preprocessing the text information in the historical input record, and obtaining a keyword group in the historical input record; Step S103: Obtain the feature value DF(a) of the ath keyword in the keyword group; Step S104: Analyze the similarity of text information between different historical input records of the user, wherein the similarity of text information between the bth historical input record and the cth historical input record is analyzed to obtain the text information similarity value P between the cth historical input record and the bth historical input record. C,B ; Step S105: When the text information approximation value P C,B is greater than a preset text information approximation threshold, determining that the text information between the c-th historical input record and the b-th historical input record is approximate, obtaining the time points of the user input information corresponding to the c-th historical input record and the b-th historical input record, and when the time point corresponding to the c-th historical input record is less than the time point of the b-th historical input record, recording the b-th historical input record as an approximate historical input record of the c-th historical input record; otherwise, recording the c-th historical input record as an approximate historical input record of the b-th historical input record.

3. The natural language-based zero-code application generation and data integration method according to claim 2, characterized in that: In step S103, the specific process of obtaining the feature value of the ath keyword includes: Calculate the characteristic inverse document frequency of each keyword in the keyword group of the historical input record, wherein the characteristic inverse document frequency F(a) of the ath keyword in the keyword group is: Among them, N sum N is the total number of the historical input records. a It is represented by the total number of historical input records containing the a-th keyword; Calculate the feature value DF(a) of the ath keyword in the historical input record: DF(a)=F(a)×Q a , Among them, Q a It is represented by the frequency of occurrence of the ath keyword in the text information in the historical input record.

4. The natural language-based zero-code application generation and data integration method according to claim 2, characterized in that: In step S104, the specific process of obtaining the approximate value of the text information of the cth historical input record and the bth historical input record includes: Analyzing the similarity of text information between different historical input records of the user, wherein the specific process of analyzing the similarity of text information between the bth historical input record and the cth historical input record is to obtain the keyword group of the cth historical input record and the bth historical input record; Obtaining feature values ​​of each keyword in the keyword group of the cth historical input record and the bth historical input record, and converting text information in the cth historical input record and the bth historical input record into a numerical vector group according to the feature values; Calculate the text information approximation P of the cth historical input record and the bth historical input record C,B : Wherein, y represents the total number of elements in the numerical vector group of the c-th historical input record; B z The zth element in the numerical vector group of the bth historical input record is represented by C z The zth element in the numeric vector group represented as the cth historical input record.

5. The natural language-based zero-code application generation and data integration method according to claim 2, characterized in that: The step S200 includes: Step S201: When the user inputs text information in natural language into the platform, the platform generates content according to the text information input by the user, and outputs the generated content to the user on the platform, thereby obtaining output data of the platform; Step S202: obtaining historical output data output by the platform according to the historical input record, obtaining an approximate historical input record of the historical input record, and evaluating the data difference of the historical output data between the historical input record and a certain approximate historical input record of the historical input record. The specific evaluation process is to extract the code generated by the platform from the historical output data of the historical input record and the certain approximate historical input record respectively; Step S203: using a code formatting tool to format the generated code, remove comments and blank lines in the code, and preprocess the code; Step S204: analyzing the string similarity between each string in the code of the historical input record and a certain similar historical input record, and obtaining the string similarity between each string in the historical input record and each string in the certain similar historical input record; Step S205: Evaluate the data difference of the historical output data between the historical input record and the approximate historical input record to obtain the target approximate historical input record of the historical input record.

6. The natural language-based zero-code application generation and data integration method according to claim 5, characterized in that: The specific process of evaluating the data difference of the historical output data between the historical input record and the similar historical input record in step S205 includes: Obtaining an average value of string similarities between each string of the code of the historical input record and each string in the code of the approximate historical input record, and using the reciprocal of the average value as the output difference value between the historical input record and the approximate historical input record; The output difference value between the historical input record and a plurality of corresponding approximate historical input records is obtained, and the approximate historical input record corresponding to the maximum value of the output difference value is recorded as the target approximate historical input record of the historical input record.

7. The natural language-based zero-code application generation and data integration method according to claim 5, characterized in that: The step S300 includes: Step S301: Acquire a keyword group of text information of the user's historical input record, and acquire a keyword group of a target similar historical input record of the historical input record; Step S302: Analyze the semantic similarity between the target approximate historical input record and different keywords in the keyword group between the historical input record, and obtain the characteristic semantic similarity value between the αth keyword of the historical input record and each keyword in the keyword group of the target approximate historical input record; Step S303: When a certain keyword in the keyword group of the target approximate historical input record has a maximum feature semantic approximation value with the αth keyword among the feature semantic approximation values ​​of the various keywords, the certain keyword is used as a target replacement keyword for the αth keyword; Step S304: obtaining a plurality of target replacement keywords of a plurality of keywords, and collecting the plurality of target replacement keywords of the plurality of keywords to obtain the target keyword replacement data of the user.

8. The natural language-based zero-code application generation and data integration method according to claim 7, characterized in that: In step S302, the specific process of obtaining the feature semantic approximation value of each keyword in the keyword group of the target similar historical input record for the αth keyword of the historical input record includes: Converting the historical input record and each keyword in the keyword group of the target similar historical input record into a vector respectively; Calculate the feature semantic approximation value between the historical input record and each keyword in the target approximate historical input record, where the feature semantic approximation value U of the αth keyword in the historical input record and the βth keyword in the target approximate historical input record is α,β : Among them, V α V is represented by the vector corresponding to the αth keyword; β Represented as the vector corresponding to the βth keyword.

9. The natural language-based zero-code application generation and data integration method according to claim 8, characterized in that: The step S400 includes: Step S401: acquiring the natural language information input by the user on the platform in the current period, and converting the natural language information input by the user in the current period into text information; Step S402: Acquire the target keyword replacement data of the user, extract keywords from the text information input by the user in the current period, and replace the keywords in the text information input by the user in the current period with the target replacement keywords corresponding to the keywords based on the target keyword replacement data, optimize and adjust the information content input by the user in the current period, and the platform outputs target data based on the optimized and adjusted text information, and displays the target data to the user on the platform.

Images