Project archive management method and system based on text recognition engine

Abstract

The invention provides a project archive management method and system based on a text recognition engine, and relates to the technical field of artificial intelligence machine learning, and the method comprises the steps: carrying out the text recognition of a pre-processed project document through a text recognition engine based on machine learning training; and in the process of library building management of the text recognition result, performing an optimal game between the autonomous processing strategy and the user divergence in the next process, and executing the next process according to the game result. According to the method, the text recognition engine is trained in advance by utilizing a machine learning technology, the preprocessed project document is subjected to text recognition by utilizing the text recognition engine, and a text recognition result is used for library building management, so that the refinement degree and the intelligent level of archive management are improved. The optimal balance between the user demand and the system rule is found, and the optimal execution strategy is executed.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence and machine learning technology, and in particular to a project archive management method and system based on a text recognition engine. Background Art

[0002] Currently, project archive management primarily relies on digitalization, typically converting paper-based project archives into electronic form and then categorizing and storing them. However, this approach lacks in-depth text recognition and intelligent processing capabilities for project archives, resulting in a low level of refinement and intelligence in archive management.

[0003] With the continuous advancement of artificial intelligence machine learning technology, how to effectively apply it to project archive management, thereby improving the sophistication and intelligence level of project archive management, has become a key issue that needs to be solved urgently.

[0004] In addition, most existing digital management methods rely on fixed management rules, such as fixed electronic file conversion modes and classification modes, but in actual applications, there are often differences between user needs and system rules.

[0005] Therefore, how to find the best balance between user needs and system rules and implement the optimal execution strategy is also a problem that needs to be solved urgently. Summary of the Invention

[0006] One of the purposes of the present invention is to provide a project archive management method based on a text recognition engine to solve the problems in the background technology.

[0007] An embodiment of the present invention provides a project archive management method based on a text recognition engine, comprising:

[0008] Use the pre-trained text recognition engine based on machine learning to perform text recognition on the pre-processed project documents;

[0009] In the process of building a database for text recognition results, an optimal game is played between the autonomous processing strategy and user disagreements for the next step, and the next step is executed based on the result of the game.

[0010] Optionally, the step of pre-training the text recognition engine based on machine learning includes:

[0011] A large number of project document recognition samples are used as training samples for machine learning training to obtain a text recognition engine.

[0012] Optionally, the project document preprocessing step includes:

[0013] The project documents are processed at least including image tilt correction, seal removal, table removal and stain removal.

[0014] Optionally, the optimal game between the autonomous processing strategy and user disagreement for the next step includes:

[0015] When the difference between the expected quantitative results of the execution effect of the autonomous processing strategy and the user's disagreement exceeds the difference threshold, the larger quantitative result of the expected execution effect of the autonomous processing strategy and the user's disagreement is used as the result of the game; otherwise, there is no detection of whether the user will accept the autonomous processing strategy;

[0016] When the answer is yes, the autonomous processing strategy is taken as the result of the game; otherwise, the user disagreement is taken as the result of the game.

[0017] Optionally, the non-sensitized detection of whether the user will accept the autonomous processing strategy includes:

[0018] Generate a logical sequence of verification that verifies that the user will accept the autonomous processing strategy;

[0019] When a user reviews any previous step and a continuous overlap occurs between the behavioral logic sequence and the verification logic sequence, if the sum of the first verification weights of the verification logics involved in the continuous overlap exceeds the first weight sum threshold but does not exceed the second weight sum threshold that is greater than the first weight sum threshold, a target local sequence is delineated from the verification logic sequence after the continuous overlap; wherein the second verification weight sums of no more than the threshold number of continuous verification logics contained in the target local sequence are closest to the difference between the second weight sum threshold and the first verification weight sum;

[0020] Try to guide users to actively generate new verification behaviors that conform to the target local sequence;

[0021] When the attempt is successful, it is determined that the user will accept the autonomous processing strategy;

[0022] If the first verification weight sum exceeds the second weight sum threshold, it is determined that the user will accept the autonomous processing strategy.

[0023] Optionally, generating a verification logic sequence for verifying that the user will accept the autonomous processing strategy includes:

[0024] Analyze the different types of differences between autonomous processing strategies and user disagreements in terms of their respective execution effect expectations;

[0025] Traverse each difference type in turn;

[0026] During each traversal, obtain the process scenario that supports the user to express the execution effect tendency of the different types traversed through behavior in any previous step;

[0027] Taking the execution effect tendency that is more inclined to the autonomous processing strategy corresponding to the traversed difference type as the verification target, a local verification logic sequence is generated according to the scenario configuration of the process scenario;

[0028] After traversing each difference type, the corresponding local verification logic sequence is sorted and spliced ​​according to the maximum possible review order of the process scene obtained in each traversal to obtain the verification logic sequence.

[0029] Optionally, the attempt to guide the user to actively generate a new verification behavior that conforms to the target local sequence includes:

[0030] Obtain the set of behaviors generated by the user within the last preset time when reviewing any previous step;

[0031] Generate guidance information; wherein the guidance information is relevant to each behavior in the behavior set and has the function of guiding the user to actively generate new confirmation behavior;

[0032] Output guidance information to the user.

[0033] Optionally, the steps for obtaining user differences include:

[0034] Generate a quick-select table of disagreements based on the predicted possible disagreements among users regarding autonomous processing strategies;

[0035] Based on the possible disagreements selected by the user from the disagreement quick selection table and the newly input disagreement, the user disagreement is determined.

[0036] Optionally, executing the next step based on the game result includes:

[0037] When the result of the game is an autonomous processing strategy, the next step is executed based on the autonomous processing strategy;

[0038] When the result of the game is user disagreement, the autonomous processing strategy is revised with the acceptance of user disagreement as the correction goal, and the next step is executed based on the revised autonomous processing strategy.

[0039] An embodiment of the present invention provides a project archive management system based on a text recognition engine, comprising:

[0040] The text recognition module is used to perform text recognition on pre-processed project documents using a text recognition engine pre-trained based on machine learning;

[0041] The game execution module is used to conduct an optimal game between the autonomous processing strategy and user disagreement for the next step in the process of building a database for the text recognition results, and execute the next step based on the result of the game.

[0042] The present invention has achieved the following beneficial effects:

[0043] This invention utilizes machine learning technology to pre-train a text recognition engine, which is then used to perform text recognition on pre-processed project documents. The results of this recognition are then used for database management, improving the sophistication and intelligence of archive management. Secondly, during the database management of the text recognition results, an optimal negotiation is performed between the autonomous processing strategy and user disagreements for the next step. The next step is then executed based on the results of this negotiation, achieving the optimal balance between user needs and system rules, and implementing the optimal execution strategy.

[0044] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0045] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0047] Figure 1 Schematic diagram of a project archive management method based on a text recognition engine in an embodiment of the present invention;

[0048] Figure 2 Schematic diagram of a project archive management system based on a text recognition engine in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0050] Example 1:

[0051] The embodiment of the present invention provides a project archive management method based on a text recognition engine, such as Figure 1 As shown, including:

[0052] S1. Use the pre-trained text recognition engine based on machine learning to perform text recognition on the pre-processed project documents;

[0053] S2. In the process of building a database for the text recognition results, an optimal game is conducted between the autonomous processing strategy and user differences for the next step, and the next step is executed based on the result of the game.

[0054] In the above technical solution, a text recognition engine with the ability to recognize text in project documents is pre-trained based on machine learning. In order to improve the efficiency of text recognition, the project documents are pre-processed to improve the document quality. The text recognition engine is used to perform text recognition on the pre-processed project documents, and the identified text content of each type is used as the result of text recognition.

[0055] When building a database to manage the results of text recognition, you can follow the preset classification rules and classify the text content according to type; you can also build a secure database to store key information safely; you can also analyze the results of text recognition, generate analysis reports, and build a database for storage.

[0056] The process of building a database and managing the results of text recognition is divided into multiple steps. The autonomous processing strategy of the next process is the processing strategy that the system independently plans to execute in the next process. User disagreement is a disagreement between the user and the system. The user actually wants the system to execute the processing strategy in the next process. The optimal game between the two is carried out, and the result of the game is the optimal execution strategy, which is used to execute the next process.

[0057] This invention utilizes machine learning technology to pre-train a text recognition engine, which is then used to perform text recognition on pre-processed project documents. The results of this recognition are then used for database management, improving the sophistication and intelligence of archive management. Secondly, during the database management of the text recognition results, an optimal negotiation is performed between the autonomous processing strategy and user disagreements for the next step. The next step is then executed based on the results of this negotiation, achieving the optimal balance between user needs and system rules, and implementing the optimal execution strategy.

[0058] Example 2:

[0059] In an embodiment of the present invention, the step of pre-training the text recognition engine based on machine learning includes:

[0060] A large number of project document recognition samples are used as training samples for machine learning training to obtain a text recognition engine.

[0061] In this technical solution, we pre-collect image samples of various project documents, including different formats, fonts, layouts, and languages. We annotate each collected image sample with text content and text recognition areas, ultimately generating a large number of project document recognition samples. These serve as training samples for machine learning, enabling the trained text recognition engine to possess the ability to recognize text in project documents.

[0062] Example 3:

[0063] In the process of processing project archive images, the recognition of image quality is affected by many factors, resulting in unsatisfactory extraction of text content, which in turn affects the acquisition of key information.

[0064] To address this issue, in an embodiment of the present invention, the project document preprocessing step includes:

[0065] The project documents are processed at least including image tilt correction, seal removal, table removal and stain removal.

[0066] In the above technical solution, the tilted document image is first corrected using Hough transform and perspective transform techniques, thereby correcting the tilt angle of the document image, removing unnecessary background parts, and retaining the main content of the document. For images containing seals and tables, the luminance component in the YUV color space is used to separate the seal and text, removing the seal part; at the same time, the Hough transform is used to detect straight lines, and the table area is filled with the background color to eliminate table interference. When dealing with image quality degradation issues, a local adaptive threshold method is used for binarization to address non-uniform illumination; for damaged images, the Criminisi algorithm is applied to repair the damaged areas; for motion blur, the improved DeblurGAN algorithm effectively removes blur and significantly improves image quality. After these processes, the quality of document images will be significantly improved, and the recognition effect of text content will also be significantly improved.

[0067] Example 4:

[0068] In the embodiment of the present invention, the optimal game between the autonomous processing strategy and user disagreement for the next step includes:

[0069] When the difference between the expected quantitative results of the execution effect of the autonomous processing strategy and the user's disagreement exceeds the difference threshold, the larger quantitative result of the expected execution effect of the autonomous processing strategy and the user's disagreement is used as the result of the game; otherwise, there is no detection of whether the user will accept the autonomous processing strategy;

[0070] When the answer is yes, the autonomous processing strategy is taken as the result of the game; otherwise, the user disagreement is taken as the result of the game.

[0071] In the above technical solution, the autonomous processing strategy and user disagreement each have an expected effect. The expected effect is the expected effect of project archive management that can be achieved by executing the next step of the autonomous processing strategy or user disagreement as a result of the game. The quantitative result refers to the result of quantifying and weighting the degree of effect of the expected effect using preset quantitative indicators. The quantification process is exemplified as follows:

[0072] Assume that the expected effect, various quantitative indicators and their quantitative results, and weights are as shown in the following table:

[0073]

[0074] Then the expected quantitative result of this effect = (30×0.6)+(10×0.4)=18+4=22.

[0075] The quantitative result difference is the absolute value of the difference between the quantitative results of the expected execution effects of the autonomous handling strategy and the user disagreement. The difference threshold is the threshold value that indicates that the quantitative result difference is too large. When the quantitative result difference exceeds the difference threshold, it indicates that the expected degree of effectiveness of the autonomous handling strategy and the user disagreement is too different. Taking the larger quantitative result as the result of the game can ensure the execution of the optimal execution strategy. Otherwise, it means that the expected degree of effectiveness of the autonomous handling strategy and the user disagreement is relatively close. To ensure the consistency of the system's autonomous decision-making as much as possible, it is necessary to prioritize the autonomous handling strategy as the result of the game whenever possible. Then, a non-sensory detection is performed to determine whether the user will accept the autonomous handling strategy. If so, the autonomous handling strategy is used as the result of the game; otherwise, the user disagreement is used as the result of the game. Non-sensory detection means that the detection process is not perceived by the user.

[0076] When performing the optimal game, the embodiment of the present invention quantifies the expected execution effects of the autonomous processing strategy and the user disagreement. When the difference in the quantified results exceeds the difference threshold, the one with the larger quantified result is used as the result of the game. Otherwise, a non-sensical detection is performed to determine whether the user will accept the autonomous processing strategy. If so, the autonomous processing strategy is used as the result of the game. Otherwise, the user disagreement is used as the result of the game, fully ensuring the execution of the optimal execution strategy in many cases. In particular, when detecting whether the user will accept the autonomous processing strategy, a non-sensical detection is performed, which not only avoids interference with the user, but also prevents the user from developing subjective biases (such as believing that the system deliberately forces them to accept the autonomous processing strategy) or discomfort due to knowing the detection process, or even affecting the subjective accuracy of the disagreement raised in the subsequent database construction and management process.

[0077] Example 5:

[0078] In an embodiment of the present invention, the non-sensical detection of whether the user will accept the autonomous processing strategy includes:

[0079] Generate a logical sequence of verification that verifies that the user will accept the autonomous processing strategy;

[0080] When a user reviews any previous step and a continuous overlap occurs between the behavioral logic sequence and the verification logic sequence, if the sum of the first verification weights of the verification logics involved in the continuous overlap exceeds the first weight sum threshold but does not exceed the second weight sum threshold that is greater than the first weight sum threshold, a target local sequence is delineated from the verification logic sequence after the continuous overlap; wherein the second verification weight sums of no more than the threshold number of continuous verification logics contained in the target local sequence are closest to the difference between the second weight sum threshold and the first verification weight sum;

[0081] Try to guide users to actively generate new verification behaviors that conform to the target local sequence;

[0082] When the attempt is successful, it is determined that the user will accept the autonomous processing strategy;

[0083] If the first verification weight sum exceeds the second weight sum threshold, it is determined that the user will accept the autonomous processing strategy.

[0084] In the above technical solution, the term "previous step" refers to any step in the process of building and managing the database of text recognition results. When a user reviews these steps, they will generate actions, which will then express action logic, forming a action logic sequence. A continuous overlapping portion refers to the occurrence of multiple consecutive action logics in the action logic sequence that are identical to multiple consecutive verification logics in the verification logic sequence. This identical portion is considered a continuous overlapping portion.

[0085] Each confirmation logic in the confirmation logic sequence has a confirmation weight, which represents the degree to which the user will accept the autonomous handling strategy if their behavior aligns with the confirmation logic. The first weight and threshold represent the confirmation weight and threshold that indicate the user is likely to accept the autonomous handling strategy, while the second weight and threshold represent the confirmation weight and threshold that indicate the user will accept the autonomous handling strategy. If the sum of the first confirmation weights of each confirmation logic in the continuous overlap exceeds the first weight and threshold but does not exceed the second weight and threshold, further confirmation of the likelihood is required. If the first confirmation weight and threshold exceed the second weight and threshold, the user is directly determined to accept the autonomous handling strategy.

[0086] When further confirming the likelihood, it's necessary to attempt to guide the user in generating new behaviors that can be used for further confirmation. Because the various confirmation logics within the confirmation logic sequence have an inherent sequential logical relationship, users are more likely to express behavioral logic that aligns with the confirmation logic following the continuous overlapping portion of the confirmation logic sequence. Therefore, the target local sequence is defined after the continuous overlapping portion of the confirmation logic sequence, ensuring that the target local sequence contains continuous confirmation logic. Furthermore, to improve the efficiency of further confirmation, the target local sequence must contain no more than a threshold number of consecutive confirmation logics, and its second confirmation weight sum must be closest to the difference between the second weight sum threshold and the first confirmation weight sum. This threshold can be pre-set. Finally, an attempt is made to guide the user in actively generating new confirmation behaviors that align with the target local sequence. If the attempt is successful, it is determined that the user has accepted the autonomous processing strategy.

[0087] When the embodiment of the present invention detects whether a user will accept an autonomous processing strategy without any sense of awareness, a verification logic sequence is generated to confirm that the user will accept the autonomous processing strategy. When the user reviews the behavioral logic sequence of any previous process and the verification logic sequence, a continuous overlap occurs. If the sum of the first verification weights of each verification logic involved in the continuous overlap exceeds the first weight and threshold but does not exceed the second weight and threshold, further possibility confirmation is performed. If the sum of the first verification weight exceeds the second weight and threshold, it is directly determined that the user will accept the autonomous processing strategy. The user will not perceive the detection process throughout the process, achieving a non-sensing detection, which greatly improves the intelligence level and applicability of the system. Further possibility confirmation is performed, the target local sequence is accurately delineated, and an attempt is made to guide the user to actively generate new verification behaviors that conform to the target local sequence. When the attempt is successful, it is determined that the user will accept the autonomous processing strategy, thereby improving the accuracy, comprehensiveness, and efficiency of detecting whether the user will accept the autonomous processing strategy.

[0088] Example 6:

[0089] In an embodiment of the present invention, generating a verification logic sequence for verifying that the user will accept the autonomous processing strategy includes:

[0090] Analyze the different types of differences between autonomous processing strategies and user disagreements in terms of their respective execution effect expectations;

[0091] Traverse each difference type in turn;

[0092] During each traversal, obtain the process scenario that supports the user to express the execution effect tendency of the different types traversed through behavior in any previous step;

[0093] Taking the execution effect tendency that is more inclined to the autonomous processing strategy corresponding to the traversed difference type as the verification target, a local verification logic sequence is generated according to the scenario configuration of the process scenario;

[0094] After traversing each difference type, the corresponding local verification logic sequence is sorted and spliced ​​according to the maximum possible review order of the process scene obtained in each traversal to obtain the verification logic sequence.

[0095] In the above technical solution, the difference type refers to the respective types of multiple difference items in the expected execution effects of the autonomous processing strategy and the user's disagreement. The difference type has an execution effect tendency, which is the execution effect tendency of the processing strategy for database management of the difference type. The process scenario is the behavior scenario that supports the user to express this tendency. For example, if the difference type is retrieval capability, the execution effect tendency is to reduce retrieval time, reduce retrieval response time, etc., and the process scenario is a scenario related to archive retrieval.

[0096] The autonomous processing strategy has an execution effect corresponding to the traversed difference type, with the execution effect tendency being more inclined to the effect as the verification target, and generates a local verification logic sequence according to the scenario configuration of the process scenario. Scenario configuration refers to the configuration that supports the generation of behavior in the process scenario. When generating a local verification logic sequence, for example: the verification target is that the execution effect tendency is more inclined to reduce the retrieval time, the process scenario is a security database that has saved key text content, and the scenario configuration includes the display of the content retrieval time in the security database and the display of the content retrieval time in other databases. Then, the continuous verification logic in the local verification logic sequence is that the behavior reflects the user's attention to the content retrieval time in the security database and the behavior reflects the user's comparison of the content retrieval time in other databases.

[0097] The maximum possibility is based on the historical prediction of the user reviewing different process scenarios when reviewing any previous step of the process. The most likely order of the process scenarios obtained in each traversal is reviewed, and the corresponding local verification logic sequences are sorted and spliced ​​to obtain the verification logic sequence.

[0098] When generating a verification logic sequence, the embodiment of the present invention generates a local verification logic sequence for each difference type in the expected execution effect between the autonomous processing strategy and the user. The corresponding local verification logic sequences are sorted and concatenated according to the maximum possible order of review of the process scenarios acquired during each traversal, thereby obtaining a verification logic sequence. This significantly improves the ability, accuracy, and efficiency of the verification logic sequence in verifying user acceptance of the autonomous processing strategy. When generating a local verification logic sequence, the verification target is the execution effect that tends to be more inclined toward the execution effect of the traversed difference type in the autonomous processing strategy. The sequence is generated based on the scenario configuration of any previous process scenario that supports the user's behavioral expression of the execution effect tendency for the traversed difference type, thereby improving the accuracy of the generated local verification logic sequence.

[0099] Example 7:

[0100] In the embodiment of the present invention, the attempt to guide the user to actively generate a new verification behavior that conforms to the target partial sequence includes:

[0101] Obtain the set of behaviors generated by the user within the last preset time when reviewing any previous step;

[0102] Generate guidance information; wherein the guidance information is relevant to each behavior in the behavior set and has the function of guiding the user to actively generate new confirmation behavior;

[0103] Output guidance information to the user.

[0104] In the above technical solution, the preset time can be 200 seconds. The generated guidance information is related to each behavior in the behavior set, making the user feel that the system is outputting information based on their recent behavior, avoiding the user's perception of deliberate guidance. The guidance information also has the function of guiding the user to actively generate new verification behaviors. For example, if the behavior set includes checking the security configuration of the security database, the content types in the library, etc., and the verification behavior is to stop and check the content retrieval time in the security database, the guidance information will be "You have recently checked the security configuration of the security database, the content types in the library, etc., but there are still content retrieval times in the security database that have not been checked. Do you want to check them?" Outputting the guidance information to the user attempts to guide the user to actively generate new verification behaviors that conform to the target local sequence, further realizing the non-invasive detection of whether the user will accept the autonomous processing strategy.

[0105] Example 8:

[0106] In this embodiment of the present invention, the step of obtaining user differences includes:

[0107] Generate a quick-select table of disagreements based on the predicted possible disagreements among users regarding autonomous processing strategies;

[0108] Based on the possible disagreements selected by the user from the disagreement quick selection table and the newly input disagreement, the user disagreement is determined.

[0109] In the above technical solution, when obtaining user disagreements, the possible disagreements on the autonomous processing strategy can be predicted based on the user's previous database management habits, and a quick-select table of disagreements can be generated. The user disagreements can be determined based on the possible disagreements selected by the user from the quick-select table of disagreements and the newly input disagreements.

[0110] Example 9:

[0111] In the embodiment of the present invention, the process of executing the next step based on the result of the game includes:

[0112] When the result of the game is an autonomous processing strategy, the next step is executed based on the autonomous processing strategy;

[0113] When the result of the game is user disagreement, the autonomous processing strategy is revised with the acceptance of user disagreement as the correction goal, and the next step is executed based on the revised autonomous processing strategy.

[0114] In the above technical solution, user disagreement refers to user disagreement on the autonomous processing strategy. Therefore, it is necessary to accept user disagreement as the correction goal, correct the autonomous processing strategy, and execute the next step based on the corrected autonomous processing strategy.

[0115] Example 10:

[0116] The embodiment of the present invention provides a project archive management system based on a text recognition engine, such as Figure 2 As shown, including:

[0117] A text recognition module 1 is used to perform text recognition on pre-processed project documents using a text recognition engine pre-trained based on machine learning;

[0118] The game execution module 2 is used to conduct an optimal game between the autonomous processing strategy and user disagreement for the next step in the process of building a database for the text recognition results, and execute the next step based on the result of the game.

[0119] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A project file management method based on a text recognition engine, characterized in that: include: Use the pre-trained text recognition engine based on machine learning to perform text recognition on the pre-processed project documents; In the process of building a database for text recognition results, an optimal game is played between the autonomous processing strategy and user disagreements for the next step, and the next step is executed based on the result of the game.

2. The project file management method based on text recognition engine according to claim 1, characterized in that: The steps of pre-training the text recognition engine based on machine learning include: A large number of project document recognition samples are used as training samples for machine learning training to obtain a text recognition engine.

3. The project file management method based on text recognition engine according to claim 1, characterized in that: The pre-processing steps of the project documents include: The project documents are processed at least including image tilt correction, seal removal, table removal and stain removal.

4. The project file management method based on text recognition engine according to claim 1, characterized in that: The optimal game between the autonomous processing strategy for the next step and user disagreement includes: When the difference between the expected quantitative results of the execution effect of the autonomous processing strategy and the user's disagreement exceeds the difference threshold, the larger quantitative result of the expected execution effect of the autonomous processing strategy and the user's disagreement is used as the result of the game; otherwise, there is no detection of whether the user will accept the autonomous processing strategy; When the answer is yes, the autonomous processing strategy is taken as the result of the game; otherwise, the user disagreement is taken as the result of the game.

5. The project file management method based on text recognition engine according to claim 4, characterized in that: The non-sensitized detection user will accept the autonomous processing strategy, including: Generate a logical sequence of verification that verifies that the user will accept the autonomous processing strategy; When a user reviews any previous step and a continuous overlap occurs between the behavioral logic sequence and the verification logic sequence, if the sum of the first verification weights of the verification logics involved in the continuous overlap exceeds the first weight sum threshold but does not exceed the second weight sum threshold that is greater than the first weight sum threshold, a target local sequence is delineated from the verification logic sequence after the continuous overlap; wherein the second verification weight sums of no more than the threshold number of continuous verification logics contained in the target local sequence are closest to the difference between the second weight sum threshold and the first verification weight sum; Try to guide users to actively generate new verification behaviors that conform to the target local sequence; When the attempt is successful, it is determined that the user will accept the autonomous processing strategy; If the first verification weight sum exceeds the second weight sum threshold, it is determined that the user will accept the autonomous processing strategy.

6. The project file management method based on text recognition engine according to claim 5, characterized in that: The generation of a verification logic sequence for verifying that the user will accept the autonomous processing strategy includes: Analyze the different types of differences between autonomous processing strategies and user disagreements in terms of their respective execution effect expectations; Traverse each difference type in turn; During each traversal, obtain the process scenario that supports the user to express the execution effect tendency of the different types traversed through behavior in any previous step; Taking the execution effect tendency that is more inclined to the autonomous processing strategy corresponding to the traversed difference type as the verification target, a local verification logic sequence is generated according to the scenario configuration of the process scenario; After traversing each difference type, the corresponding local verification logic sequence is sorted and spliced ​​according to the maximum possible review order of the process scene obtained in each traversal to obtain the verification logic sequence.

7. The project file management method based on text recognition engine according to claim 5, characterized in that: The attempt to guide the user to actively generate new verification behaviors that conform to the target local sequence includes: Obtain the set of behaviors generated by the user within the last preset time when reviewing any previous step; Generate guidance information; wherein the guidance information is relevant to each behavior in the behavior set and has the function of guiding the user to actively generate new confirmation behavior; Output guidance information to the user.

8. The project file management method based on text recognition engine according to claim 1, characterized in that: The steps to obtain user differences include: Generate a quick-select table of disagreements based on the predicted possible disagreements among users regarding autonomous processing strategies; Based on the possible disagreements selected by the user from the disagreement quick selection table and the newly input disagreement, the user disagreement is determined.

9. The project file management method based on text recognition engine according to claim 1, characterized in that: The process of executing the next step based on the result of the game includes: When the result of the game is an autonomous processing strategy, the next step is executed based on the autonomous processing strategy; When the result of the game is user disagreement, the autonomous processing strategy is revised with the acceptance of user disagreement as the correction goal, and the next step is executed based on the revised autonomous processing strategy.

10. A project archive management system based on a text recognition engine, characterized in that: include: The text recognition module is used to perform text recognition on pre-processed project documents using a text recognition engine pre-trained based on machine learning; The game execution module is used to conduct an optimal game between the autonomous processing strategy and user disagreement for the next step in the process of building a database for the text recognition results, and execute the next step based on the result of the game.

Images