Text information correlation analysis system and method based on artificial intelligence

Abstract

The invention discloses a text information correlation analysis system and method based on artificial intelligence, and relates to the technical field of text information correlation analysis, and the method comprises the steps: obtaining the examination and approval condition data of a project, obtaining the historical examination and approval condition data of a historical project, analyzing the consistency degree of the examination and approval conditions between the project and the historical project, and determining the correlation degree. Obtaining an undetermined contrast historical item; obtaining project text information of the project, obtaining historical project text information of an undetermined comparison historical project of the project, and evaluating a project association degree between the project and the undetermined comparison historical project to obtain a target comparison historical project; acquiring project data of the project, acquiring historical project data of a target comparison historical project of the project, and performing data exception assessment on the project to obtain target exception data of the project; and obtaining target abnormal data of the project, pushing the target abnormal data to the approver of the project on the platform, and performing intelligent management on the approval of the project.

Description

Technical Field

[0001] The present invention relates to the technical field of text information association analysis, and in particular to a text information association analysis system and method based on artificial intelligence. Background Art

[0002] Generally speaking, the normal implementation of a project requires approval before it can be implemented. In the conventional project approval process, a general project needs to go through multiple evaluation processes such as project application, preliminary review, expert review, risk assessment, etc. to determine whether the project is feasible. When using conventional methods to approve projects, it is necessary to review and evaluate a lot of text information about the project, which will result in a lot of time being wasted in the project approval process. In addition, the approvers may not be very familiar with the project and may not be comprehensive enough in the project evaluation process.

[0003] Therefore, in the current process of project approval, there is also a reference and assistance to the current project being approved based on the text information of historical projects. However, there are also many disadvantages in using historical projects to refer to and assist the current project being approved. First, there are a large number of historical projects that have passed the approval. How to find projects that are consistent with the current project being approved from a large number of historical projects takes a lot of time. Secondly, the technologies involved in different projects are different, and personnel with relevant professional knowledge are required to screen and analyze historical projects. If a large number of professionals are used, this will lead to a large amount of manpower and material resources being consumed in the screening process. If professionals are not used to screen and analyze historical projects, the historical projects obtained are likely to be quite different from the projects being approved, affecting the normal project approval work and may even cause huge economic losses and safety accidents. Summary of the invention

[0004] The purpose of the present invention is to provide a text information relevance analysis system and method based on artificial intelligence to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solution: a text information relevance analysis method based on artificial intelligence, the method comprising:

[0006] Step S100: Acquire the approval condition data of the project, acquire the historical approval condition data of the historical project, analyze the consistency of the approval conditions between the project and the historical project, and obtain the pending comparison historical project;

[0007] Step S200: Acquire the project text information of the project, acquire the historical project text information of the pending comparison historical project of the project, evaluate the project association degree between the project and the pending comparison historical project, and obtain the target comparison historical project;

[0008] Step S300: Obtain project data of the project, obtain the target of the project and compare it with the historical project data of the historical project, perform data anomaly assessment on the project, and obtain target anomaly data of the project;

[0009] Step S400: Obtain target exception data of the project, push the target exception data to the project approvers on the platform, and perform intelligent management on the project approval.

[0010] Furthermore, step S100 includes:

[0011] Step S101: Acquire the projects to be approved in the current cycle, and acquire the approval condition data of the projects, where the approval condition data includes data corresponding to various approval condition parameters of the projects;

[0012] Step S102: Obtain each historical project from the cloud platform, obtain the historical approval condition data of each historical project, and analyze the consistency between the approval conditions of the project and each historical project. The specific analysis process of the consistency between the approval conditions of the project and the ath historical project includes calculating the characteristic value Z of the bth approval condition parameter of the project. b =(X b -μ b ) / σ b , where X b Represents the value of the b-th approval condition parameter of the project, μ b It is expressed as the average value of the b-th approval condition parameter in each historical project, σ b It is expressed as the standard deviation of the b-th approval condition parameter in each historical project;

[0013] Step S103: Acquire the characteristic values ​​of various approval condition parameters of the project, and calculate the consistency degree P between the approval condition of the project and the ath historical project. a :

[0014]

[0015] Where j represents the total number of project approval condition parameters; X i It is represented as the characteristic value of the parameter of the i-th approval condition of the project; It is represented as the characteristic value of the parameter of the i-th approval condition in the a-th historical project;

[0016] Step S104: When the approval condition consistency between the project and the a-th historical project is greater than the preset feature approval consistency threshold, it is determined that the approval conditions between the project and the a-th historical project are consistent, and the a-th historical project is recorded as a pending comparison historical project of the project;

[0017] The above steps calculate the characteristic values ​​of the approval condition parameters of the project because the values ​​corresponding to different approval condition parameters of different projects and different historical projects are likely to differ greatly, which is likely to cause a serious impact on the final calculation of the degree of consistency, thereby greatly reducing the calculation accuracy. By calculating the degree of consistency of the approval conditions between the project and the historical projects through the characteristic values ​​of the approval condition parameters, the impact caused by the excessive difference in the values ​​of different approval condition parameters is avoided, and the accuracy of the final judgment on the consistency of the approval conditions of the project and the historical projects is greatly increased.

[0018] Further, step S200 includes:

[0019] Step S201: Acquire project text information of a project, the project text information includes several project documents of the project, and acquire historical project text information of a historical project, the historical project text information includes various historical project documents of the historical project;

[0020] Step S202: using a natural language processing tool to perform word segmentation processing on the project document and the historical project document respectively, to obtain word sets of the project document and the historical project document respectively;

[0021] Step S203: construct a word vector model, randomly select a word set of several historical project documents as a training set, obtain the central word, set the window size C, and obtain C / 2 words adjacent to the central word respectively to obtain the context word set of the central word;

[0022] Step S204: For each context word in the context word set, randomly initialize the word vector and calculate the input vector h:

[0023]

[0024] Among them, Vw x Represented as the word vector of the i-th word in the context word set;

[0025] Step S205: Use the Softmax function to calculate the probability distribution r(w) of the central word in the output layer of the word vector model:

[0026] r(w)=Softmax(W T ×h),

[0027] Where W is the preset output weight matrix; W T Represented as the transpose of the weight matrix;

[0028] Step S206: Calculate the loss of the central word using cross entropy loss:

[0029] L=-log(r(w|w1,w2,...,w C ),

[0030] Among them, w1, w2, ..., w C They are represented as the 1st, 2nd, ..., Cth context words in the context word set respectively;

[0031] According to the loss and combined with the training set, the word vector model is optimized and adjusted to obtain the target word vector model;

[0032] Step S207: Evaluate the project association degree between the project and each pending comparison history project, wherein the specific process of evaluating the project association degree between the project and the e-th pending comparison history project is to use the target word vector model to perform vector conversion on the word set of several project documents of the project to obtain the word vector set of several project documents, and obtain the word vector group of each historical project document in the e-th pending comparison history project;

[0033] Step S208: Calculate the document association degree values ​​of several project documents of the project and each historical project document in the e-th pending comparison historical project, wherein the document association degree U of the f-th project document of the project and the g-th historical project document in the e-th pending comparison historical project is e,fg :

[0034]

[0035] Among them, G represents the word vector group of the g-th historical project document; F represents the word vector group of the f-th project document;

[0036] Step S209: Obtain the maximum value of the document association degree between the f-th project document of the project and each historical project document in the e-th pending comparison history project, and record the maximum value as the marked document association degree value U between the f-th project document of the project and the e-th pending comparison history project e,f ;

[0037] Step S210: Calculate the project relevance Y between the project and the e-th pending comparison history project e :

[0038]

[0039] Where m represents the total number of project documents; U e,z It is expressed as the correlation degree value between the zth project document of the project and the marked document of the eth pending comparison history project;

[0040] When the project correlation degree Y eIf the value is greater than a preset project association degree threshold, it is determined that there is an association between the project and the e-th pending comparison history project, and the e-th pending comparison history project is recorded as the target comparison history project of the project.

[0041] Furthermore, step S300 includes:

[0042] Step S301: Obtain each target of the project and compare it with the historical project to obtain the project data of the project, where the project data includes data corresponding to each indicator of the project;

[0043] Step S302: Acquire historical project data of each target control historical project, the historical project data including data corresponding to various indicators of the target control historical project, and pre-process the project data and the historical project data;

[0044] Step S303: performing data anomaly assessment on various indicators of the project, wherein the specific process of performing data anomaly assessment on the βth indicator of the project is to perform unique hot encoding on the data of various indicators of the project to obtain data points of various indicator data of the project;

[0045] Step S304: Calculate the abnormal value K of the βth index in the project β :

[0046]

[0047] Among them, Q β Represented as the data point of the βth indicator in the project; Q β,γ It is represented as the data point of the γth target of the project against the βth indicator of the historical project; s is represented as the total number of historical projects compared with each target in the project;

[0048] Step S305: When the marked abnormal value of the βth indicator in the project is greater than the set marked abnormal threshold, the βth indicator is determined to be an abnormal indicator of the project, several abnormal indicators of the project are obtained, and the several abnormal indicators are aggregated to obtain the target abnormal data of the project.

[0049] Furthermore, step S400 includes:

[0050] Step S401: acquiring target abnormal data of the project, acquiring several abnormal indicators of the project from the target abnormal data, and acquiring marked abnormal values ​​of the several abnormal indicators;

[0051] Step S402: Based on the marked abnormal values, several abnormal indicators of the project are sorted to obtain the priorities of the several abnormal indicators, and the target abnormal data is pushed to the project approvers on the platform to intelligently manage the project approval.

[0052] In order to better implement the above method, a text information correlation analysis system based on artificial intelligence is also proposed, which includes a pending comparison history project module, a target comparison history project module, a target abnormal data module, and an intelligent approval module;

[0053] The module of pending comparison historical projects is used to analyze the consistency of approval conditions between the project and the historical projects, and obtain the pending comparison historical projects;

[0054] A target control historical project module is used to evaluate the project correlation degree between the project and the pending control historical project to obtain the target control historical project;

[0055] The target abnormal data module is used to evaluate the data abnormality of the project and obtain the target abnormal data of the project;

[0056] The intelligent approval module is used to obtain the target exception data of the project, push the target exception data to the project approvers on the platform, and perform intelligent management of the project approval.

[0057] Further, the pending comparison history project module includes an approval condition consistency degree unit and a pending comparison history project unit;

[0058] The approval condition consistency unit is used to obtain the characteristic values ​​of various approval condition parameters of the project and calculate the consistency degree of approval conditions between the project and various historical projects;

[0059] The pending comparison historical project unit is used to determine the consistency of the approval conditions between the project and each historical project according to the consistency of the approval conditions between the project and each historical project, and obtain the pending comparison historical project.

[0060] Further, the target comparison history project module includes a project correlation degree unit and a target comparison history project unit;

[0061] A project correlation degree unit is used to calculate the project correlation degree between the project and each pending comparison historical project;

[0062] The target comparison history project unit is used to obtain the target comparison history project of the project according to the project association degree with each pending comparison history project.

[0063] Further, the target abnormal data module includes a marking abnormal value unit and a target abnormal data unit;

[0064] The outlier marking unit is used to evaluate the data anomaly of various indicators of the project and calculate the outlier marking values ​​of several indicators of the project;

[0065] The target abnormal data unit is used to obtain several abnormal indicators of the project, and aggregate the several abnormal indicators to obtain the target abnormal data of the project.

[0066] Further, the intelligent approval module includes an intelligent approval unit;

[0067] The intelligent approval unit is used to obtain the target abnormal data of the project, obtain several abnormal indicators of the project from the target abnormal data, push the target abnormal data to the project approvers on the platform, and perform intelligent management of the project approval.

[0068] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention realizes intelligent approval of projects, and takes into account two influencing factors, namely, approval conditions and the project itself, during the approval process of the project. Therefore, when obtaining the target of the project and comparing historical projects, the first consideration is whether the approval conditions between the project and the historical projects are consistent, and then a specific analysis is performed on the historical projects with consistent project approval conditions to find historical projects associated with the project, and based on the project situation of the historical projects, an abnormality judgment is made on the project data, and the data judged to be abnormal is found, and then the abnormal data is sent to the approval personnel, which not only greatly speeds up the approval speed, but also greatly improves the approval accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 It is a method flow chart of a text information relevance analysis system and method based on artificial intelligence of the present invention;

[0070] Figure 2 It is a module schematic diagram of a text information relevance analysis system and method based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0071] 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.

[0072] Example: Figure 1-Figure 2 As shown, the present invention provides a technical solution, a text information relevance analysis method based on artificial intelligence, the method comprising:

[0073] Step S100: Acquire the approval condition data of the project, acquire the historical approval condition data of the historical project, analyze the consistency of the approval conditions between the project and the historical project, and obtain the pending comparison historical project;

[0074] Wherein, step S100 includes:

[0075] Step S101: Acquire the projects to be approved in the current cycle, and acquire the approval condition data of the projects, where the approval condition data includes data corresponding to various approval condition parameters of the projects;

[0076] For example, various approval condition parameters include project personnel, material costs, etc.;

[0077] Step S102: Obtain each historical project from the cloud platform, obtain the historical approval condition data of each historical project, and analyze the consistency between the approval conditions of the project and each historical project. The specific analysis process of the consistency between the approval conditions of the project and the ath historical project includes calculating the characteristic value Z of the bth approval condition parameter of the project. b =(X b -μ b ) / σ b , where X b Represents the value of the b-th approval condition parameter of the project, μ b It is expressed as the average value of the b-th approval condition parameter in each historical project, σ b It is expressed as the standard deviation of the b-th approval condition parameter in each historical project;

[0078] Step S103: Acquire the characteristic values ​​of various approval condition parameters of the project, and calculate the consistency degree P between the approval condition of the project and the ath historical project. a :

[0079]

[0080] Where j represents the total number of project approval condition parameters; X i It is represented as the characteristic value of the parameter of the i-th approval condition of the project; It is represented as the characteristic value of the parameter of the i-th approval condition in the a-th historical project;

[0081] For example, if the total number of approval condition parameters of a project is 2, the characteristic value X1 of the first approval condition parameter of the project is 0.6; the characteristic value of the first approval condition parameter of the third historical project is is expressed as 0.4; the characteristic value X1 of the second approval condition parameter of the project is expressed as 0.7; the characteristic value of the second approval condition parameter of the third historical project It is expressed as 0.5;

[0082] Calculate the consistency degree of approval conditions between the project and the third historical project P3:

[0083]

[0084] Step S104: When the approval condition consistency between the project and the a-th historical project is greater than the preset feature approval consistency threshold, it is determined that the approval conditions between the project and the a-th historical project are consistent, and the a-th historical project is recorded as a pending comparison historical project of the project;

[0085] Step S200: Acquire the project text information of the project, acquire the historical project text information of the pending comparison historical project of the project, evaluate the project association degree between the project and the pending comparison historical project, and obtain the target comparison historical project;

[0086] Wherein, step S200 includes:

[0087] Step S201: Acquire project text information of a project, the project text information includes several project documents of the project, and acquire historical project text information of a historical project, the historical project text information includes various historical project documents of the historical project;

[0088] Step S202: using a natural language processing tool to perform word segmentation processing on the project document and the historical project document respectively, to obtain word sets of the project document and the historical project document respectively;

[0089] Step S203: construct a word vector model, randomly select a word set of several historical project documents as a training set, obtain the central word, set the window size C, and obtain C / 2 words adjacent to the central word respectively to obtain the context word set of the central word;

[0090] Step S204: For each context word in the context word set, randomly initialize the word vector and calculate the input vector h:

[0091]

[0092] Among them, Vw x Represented as the word vector of the i-th word in the context word set;

[0093] Step S205: Use the Softmax function to calculate the probability distribution r(w) of the central word in the output layer of the word vector model:

[0094] r(w)=Softmax(W T ×h),

[0095] Where W is the preset output weight matrix; W T Represented as the transpose of the weight matrix;

[0096] Step S206: Calculate the loss of the central word using cross entropy loss:

[0097] L=-log(r(w|w1,w2,...,w C )),

[0098] Among them, w1, w2, ..., w C They are represented as the 1st, 2nd, ..., Cth context words in the context word set respectively;

[0099] According to the loss and combined with the training set, the word vector model is optimized and adjusted to obtain the target word vector model;

[0100] Step S207: Evaluate the project association degree between the project and each pending comparison history project, wherein the specific process of evaluating the project association degree between the project and the e-th pending comparison history project is to use the target word vector model to perform vector conversion on the word set of several project documents of the project to obtain the word vector set of several project documents, and obtain the word vector group of each historical project document in the e-th pending comparison history project;

[0101] Step S208: Calculate the document association degree values ​​of several project documents of the project and each historical project document in the e-th pending comparison historical project, wherein the document association degree U of the f-th project document of the project and the g-th historical project document in the e-th pending comparison historical project is e,fg :

[0102]

[0103] Among them, G represents the word vector group of the g-th historical project document; F represents the word vector group of the f-th project document;

[0104] Step S209: Obtain the maximum value of the document association degree between the f-th project document of the project and each historical project document in the e-th pending comparison history project, and record the maximum value as the marked document association degree value U between the f-th project document of the project and the e-th pending comparison history project e,f ;

[0105] Step S210: Calculate the project relevance Y between the project and the e-th pending comparison history project e :

[0106]

[0107] Where m represents the total number of project documents; U e,z It is expressed as the correlation degree value between the zth project document of the project and the marked document of the eth pending comparison history project;

[0108] For example, if m is 3, the correlation value between the first project document of the project and the marked document of the second pending comparison history project is U2,1 is 0.6; the correlation value U between the second project document of the project and the marked document of the second pending comparison history project 2,1 is 0.7; the correlation value U between the third project document of the project and the marked document of the second pending comparison history project 2,1 is 0.8;

[0109] Calculate the project's correlation degree Y2 with the second pending comparison history project:

[0110]

[0111] When the project correlation degree Y e If the value is greater than a preset project correlation threshold, it is determined that there is a correlation between the project and the e-th pending comparison history project, and the e-th pending comparison history project is recorded as the target comparison history project of the project;

[0112] Step S300: Obtain project data of the project, obtain the target of the project and compare it with the historical project data of the historical project, perform data anomaly assessment on the project, and obtain target anomaly data of the project;

[0113] Wherein, step S300 includes:

[0114] Step S301: Obtain each target of the project and compare it with the historical project to obtain the project data of the project, where the project data includes data corresponding to each indicator of the project;

[0115] For example, various indicators include project defect rate, project resource utilization rate, etc.;

[0116] Step S302: Acquire historical project data of each target control historical project, the historical project data including data corresponding to various indicators of the target control historical project, and pre-process the project data and the historical project data;

[0117] For example, the preprocessing process includes handling missing values, outliers, standardizing data formats, and ensuring that the characteristics between historical data and current project data are consistent;

[0118] Step S303: performing data anomaly assessment on various indicators of the project, wherein the specific process of performing data anomaly assessment on the βth indicator of the project is to perform unique hot encoding on the data of various indicators of the project to obtain data points of various indicator data of the project;

[0119] Step S304: Calculate the abnormal value K of the βth index in the project β :

[0120]

[0121] Among them, Qβ Represented as the data point of the βth indicator in the project; Q β,γ It is represented as the data point of the γth target of the project against the βth indicator of the historical project; s is represented as the total number of historical projects compared with each target in the project;

[0122] Step S305: when the marked abnormal value of the βth indicator in the project is greater than the set marked abnormal threshold, the βth indicator is determined to be an abnormal indicator of the project, and several abnormal indicators of the project are obtained, and the several abnormal indicators are aggregated to obtain the target abnormal data of the project;

[0123] Step S400: Obtain target abnormal data of the project, push the target abnormal data to the project approvers on the platform, and perform intelligent management on the project approval;

[0124] Wherein, step S400 includes:

[0125] Step S401: acquiring target abnormal data of the project, acquiring several abnormal indicators of the project from the target abnormal data, and acquiring marked abnormal values ​​of the several abnormal indicators;

[0126] Step S402: Based on the marked abnormal values, several abnormal indicators of the project are sorted to obtain the priorities of the several abnormal indicators, and the target abnormal data is pushed to the project approvers on the platform to perform intelligent management of the project approval;

[0127] In order to better implement the above method, a text information correlation analysis system based on artificial intelligence is also proposed, which includes a pending comparison history project module, a target comparison history project module, a target abnormal data module, and an intelligent approval module;

[0128] The module of pending comparison historical projects is used to analyze the consistency of approval conditions between the project and the historical projects, and obtain the pending comparison historical projects;

[0129] A target control historical project module is used to evaluate the project correlation degree between the project and the pending control historical project to obtain the target control historical project;

[0130] The target abnormal data module is used to evaluate the data abnormality of the project and obtain the target abnormal data of the project;

[0131] The intelligent approval module is used to obtain the target abnormal data of the project, push the target abnormal data to the project approvers on the platform, and perform intelligent management of the project approval;

[0132] Among them, the pending comparison history project module includes the approval condition consistency degree unit and the pending comparison history project unit;

[0133] The approval condition consistency unit is used to obtain the characteristic values ​​of various approval condition parameters of the project and calculate the consistency degree of approval conditions between the project and various historical projects;

[0134] The pending comparison historical project unit is used to determine the consistency of the approval conditions between the project and each historical project according to the consistency of the approval conditions between the project and each historical project, and obtain the pending comparison historical project;

[0135] Among them, the target comparison historical project module includes a project correlation degree unit and a target comparison historical project unit;

[0136] A project correlation degree unit is used to calculate the project correlation degree between the project and each pending comparison historical project;

[0137] A target comparison history project unit is used to obtain a target comparison history project of a project according to the project association degree with each pending comparison history project;

[0138] Among them, the target abnormal data module includes a marking abnormal value unit and a target abnormal data unit;

[0139] The outlier marking unit is used to evaluate the data anomaly of various indicators of the project and calculate the outlier marking values ​​of several indicators of the project;

[0140] The target abnormal data unit is used to obtain several abnormal indicators of the project and aggregate the several abnormal indicators to obtain the target abnormal data of the project;

[0141] Among them, the intelligent approval module includes an intelligent approval unit;

[0142] The intelligent approval unit is used to obtain the target abnormal data of the project, obtain several abnormal indicators of the project from the target abnormal data, push the target abnormal data to the project approvers on the platform, and perform intelligent management of the project approval.

[0143] 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 text information relevance analysis method based on artificial intelligence, characterized in that: The method comprises: Step S100: Acquire the approval condition data of the project, acquire the historical approval condition data of the historical project, analyze the consistency of the approval conditions between the project and the historical project, and obtain the pending comparison historical project; Step S200: Acquire the project text information of the project, acquire the historical project text information of the pending comparison historical project of the project, evaluate the project association degree between the project and the pending comparison historical project, and obtain the target comparison historical project; Step S300: Acquire the project data of the project, obtain the historical project data of the target comparison historical project of the project, perform data anomaly assessment on the project, and obtain the target anomaly data of the project; Step S400: Obtain target abnormal data of the project, push the target abnormal data to the project approver on the platform, and perform intelligent management on the project approval.

2. The method for analyzing text information relevance based on artificial intelligence according to claim 1, characterized in that: The step S100 includes: Step S101: Acquire the projects to be approved in the current cycle, and acquire the approval condition data of the projects, wherein the approval condition data includes data corresponding to various approval condition parameters of the projects; Step S102: Obtain each historical project from the cloud platform, obtain the historical approval condition data of each historical project, and analyze the consistency of the approval conditions between the project and each historical project. The specific analysis process of the consistency of the approval conditions between the project and the ath historical project includes calculating the characteristic value Z of the bth approval condition parameter of the project. b =(X b -μ b ) / σ b , where X b Represents the value of the b-th approval condition parameter of the project, μ b It is expressed as the average value of the b-th approval condition parameter in each historical project, σ b Expressed as the standard deviation of the approval condition parameter of item b in each of the historical items; Step S103: Acquire the characteristic values ​​of various approval condition parameters of the project, and calculate the consistency degree P between the approval condition of the project and the ath historical project. a : Wherein, j represents the total number of approval condition parameters of the project; X i It is represented as the characteristic value of the i-th approval condition parameter of the project; X i a is represented as the characteristic value of the i-th approval condition parameter in the a-th historical item; Step S104: When the degree of consistency between the approval conditions of the project and the ath historical project is greater than the preset feature approval consistency threshold, it is determined that the approval conditions between the project and the ath historical project are consistent, and the ath historical project is recorded as the pending comparison historical project of the project.

3. The method for analyzing text information relevance based on artificial intelligence according to claim 2, characterized in that: The step S200 includes: Step S201: acquiring project text information of the project, the project text information including several project documents of the project, acquiring historical project text information of the historical project, the historical project text information including various historical project documents of the historical project; Step S202: using a natural language processing tool to perform word segmentation processing on the project document and the historical project document respectively, to obtain word sets of the project document and the historical project document respectively; Step S203: construct a word vector model, randomly select a word set of several historical project documents as a training set, obtain a central word, set a window size C, and respectively obtain C / 2 words adjacent to the central word to obtain a context word set of the central word; Step S204: For each context word in the context word set, randomly initialize the word vector and calculate the input vector h: Among them, Vw x Represented as the word vector of the i-th word in the context word set; Step S205: Using the Softmax function, calculate the probability distribution r(w) of the central word in the output layer of the word vector model: r(w)=Softmax(W T ×h), Where W is the preset output weight matrix; W T is represented as the transpose of the weight matrix; Step S206: Calculate the loss of the central word using cross entropy loss: L=-log(r(w|w1,w2,...,w c )), Among them, w1, w2, ..., w C They are represented as the 1st, 2nd, ..., Cth context words in the context word set respectively; According to the loss and in combination with the training set, the word vector model is optimized and adjusted to obtain a target word vector model; Step S207: Evaluate the project association degree between the project and each pending comparison history project, wherein the specific process of evaluating the project association degree between the project and the e-th pending comparison history project is to use the target word vector model to perform vector conversion on the word set of several project documents of the project to obtain the word vector set of the several project documents, and obtain the word vector group of each historical project document in the e-th pending comparison history project; Step S208: Calculate the document association degree values ​​of several project documents of the project and each historical project document in the e-th pending comparison historical project, wherein the document association degree of the f-th project document of the project and the g-th historical project document in the e-th pending comparison historical project is Wherein, G represents the word vector group of the g-th historical project document; F represents the word vector group of the f-th project document; Step S209: Obtain the maximum value of the document association degree between the f-th project document of the project and each historical project document in the e-th pending comparison history project, and record the maximum value as the marked document association degree value U between the f-th project document of the project and the e-th pending comparison history project. e,f ; Step S210: Calculate the project correlation degree Y between the project and the e-th pending comparison history project e : Wherein, m represents the total number of project documents of the project; U e,z It is represented as the association degree value between the zth project document of the project and the marked document of the eth pending comparison history project; When the project association degree Y e If the value is greater than a preset project association degree threshold, it is determined that there is an association between the project and the e-th pending comparison history project, and the e-th pending comparison history project is recorded as the target comparison history project of the project.

4. The method for analyzing text information relevance based on artificial intelligence according to claim 1, characterized in that: The step S300 includes: Step S301: Obtain each target of the project and compare it with the historical project to obtain the project data of the project, wherein the project data includes data corresponding to each indicator of the project; Step S302: acquiring historical project data of each target control historical project, wherein the historical project data includes data corresponding to various indicators of the target control historical project, and preprocessing the project data and the historical project data; Step S303: performing data anomaly assessment on various indicators of the project, wherein the specific process of performing data anomaly assessment on the βth indicator of the project is to perform unique hot encoding on the data of various indicators of the project to obtain data points of various indicator data of the project; Step S304: Calculate the abnormal value K of the βth index in the project β : Among them, Q β Represented as the data point of the βth indicator in the project; Q β,γ denoted as the data point of the βth indicator in the γth target control history item of the project; s represents the total number of each target control history item in the project; Step S305: When the marked abnormal value of the βth indicator in the project is greater than the set marked abnormal threshold, the βth indicator is determined to be an abnormal indicator of the project, several abnormal indicators of the project are obtained, and the several abnormal indicators are aggregated to obtain the target abnormal data of the project.

5. The method for analyzing text information relevance based on artificial intelligence according to claim 4, characterized in that: The step S400 includes: Step S401: acquiring target abnormal data of the project, acquiring several abnormal indicators of the project from the target abnormal data, and acquiring marked abnormal values ​​of the several abnormal indicators; Step S402: Based on the marked abnormal value, several abnormal indicators of the project are sorted to obtain the priority of the several abnormal indicators, and the target abnormal data is pushed to the project approver on the platform to intelligently manage the approval of the project.

6. An artificial intelligence-based text information relevance analysis system, used to execute an artificial intelligence-based text information relevance analysis method according to any one of claims 1 to 5, characterized in that: The system includes a pending comparison history project module, a target comparison history project module, a target abnormal data module, and an intelligent approval module; The pending comparison historical project module is used to analyze the consistency of approval conditions between the project and historical projects to obtain pending comparison historical projects; The target comparison history project module is used to evaluate the project association degree between the project and the pending comparison history project to obtain the target comparison history project; The target abnormal data module is used to perform data abnormality assessment on the project to obtain target abnormal data of the project; The intelligent approval module is used to obtain target abnormal data of the project, push the target abnormal data to the project approvers on the platform, and perform intelligent management on the project approval.

7. The text information relevance analysis system based on artificial intelligence according to claim 6 is characterized in that: The pending comparison history project module includes an approval condition consistency degree unit and a pending comparison history project unit; The approval condition consistency unit is used to obtain the characteristic values ​​of various approval condition parameters of the project and calculate the consistency degree of approval conditions between the project and various historical projects; The pending comparison historical project unit is used to determine the consistency of the approval conditions between the project and each historical project according to the consistency of the approval conditions between the project and each historical project, so as to obtain the pending comparison historical project.

8. The text information relevance analysis system based on artificial intelligence according to claim 6 is characterized in that: The target comparison history project module includes a project correlation degree unit and a target comparison history project unit; The project association degree unit is used to calculate the project association degree between the project and each pending comparison historical project; The target comparison history item unit is used to obtain the target comparison history item of the item according to the item association degree between the item and each pending comparison history item.

9. The text information relevance analysis system based on artificial intelligence according to claim 6, characterized in that: The target abnormal data module includes a marking abnormal value unit and a target abnormal data unit; The outlier marking unit is used to perform data anomaly evaluation on various indicators of the project and calculate the outlier marking values ​​of several indicators of the project; The target abnormal data unit is used to obtain several abnormal indicators of the project, and to aggregate the several abnormal indicators to obtain the target abnormal data of the project.

10. The text information relevance analysis system based on artificial intelligence according to claim 6, characterized in that: The intelligent approval module includes an intelligent approval unit; The intelligent approval unit is used to obtain target abnormal data of the project, obtain several abnormal indicators of the project from the target abnormal data, push the target abnormal data to the approver of the project on the platform, and intelligently manage the approval of the project.

Images