Archive classification management system based on artificial intelligence

Abstract

The invention relates to the technical field of archive management, and particularly discloses an archive classification management system based on artificial intelligence, which comprises the following modules: a multi-modal data acquisition module used for integrating texts, images, audios and metadata to generate fusion feature vectors; the dynamic classification engine is used for dynamically optimizing classification rules based on a Transform model and a reinforcement learning framework; the block chain evidence storage module is used for recording a classification operation hash value by using an intelligent contract; the intelligent life cycle management module is used for automatically triggering archiving and destroying processes according to the classification labels; and the self-adaptive value evaluation module is used for dynamically generating a value report through a time sequence neural network. The method has the beneficial effects that 1, the classification efficiency and accuracy are improved; 2, file values are fully excavated; 3, the system adaptability is enhanced; and 4, file safety is guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of file management, and in particular, to a file classification management system based on artificial intelligence. Background Art

[0002] With the rapid development of information technology, the amount of file data generated by various industries has increased explosively, and the complexity of file management has been continuously improved. There are many problems in the existing file classification management methods: on the one hand, the traditional classification methods based on manual or simple rules are inefficient. When facing a large number of files, the classification speed is slow and errors are prone to occur, making it difficult to meet the needs of quickly retrieving and using files; on the other hand, the potential value in file data cannot be fully explored, and intelligent classification and correlation analysis cannot be performed according to the comprehensive characteristics of files, resulting in a low utilization rate of file resources. In addition, in the scenario of cross-domain and multi-format file management, the existing system lacks the ability to effectively process complex data structures and semantic information, and it is difficult to achieve accurate classification. Summary of the Invention

[0003] The present invention proposes a file classification management system based on artificial intelligence to solve the above problems, which combines multi-modal data analysis, dynamic machine learning and blockchain technology to realize intelligent classification, secure storage and life cycle management of files.

[0004] The technical solution of the present invention is realized as follows:

[0005] A file classification management system based on artificial intelligence, comprising the following modules:

[0006] Multi-modal data acquisition module: used to integrate text, images, audio and metadata, extract features through natural language processing (NLP), convolutional neural network (CNN) and speech recognition technology, and generate fused feature vectors;

[0007] Dynamic classification engine: based on the Transformer model and reinforcement learning framework, dynamically optimizes classification rules according to user feedback, and supports multi-label classification and fuzzy classification;

[0008] Blockchain evidence storage module: uses smart contracts to record key operations in the classification process to ensure data immutability and operation traceability;

[0009] Intelligent life cycle management module: automatically triggers the filing, backup, sharing or destruction process according to classification labels, and generates operation logs;

[0010] Adaptive value evaluation module: dynamically analyzes technical advancement, social value, usage frequency and compliance through time series neural network technology, adjusts index weights and generates value reports.

[0011] Furthermore, the multimodal data acquisition module includes:

[0012] Text analysis unit: extracting semantic features using the BERT model;

[0013] Image processing unit: extracting image features using ResNet-50, supporting cross-modal association;

[0014] Metadata parsing unit: automatically extracting file creation time, author, version information, and fusing with content features.

[0015] Furthermore, the dynamic classification engine optimizes the model in the following ways:

[0016] The correction behavior of the user for the classification result is used as a feedback signal, and the DQN (Deep Q-Network) is used to adjust the classification threshold;

[0017] Introducing domain association rules in knowledge graph construction to enhance the semantic understanding ability of fuzzy classification.

[0018] Furthermore, the blockchain evidence storage module is specifically implemented as: classifying and storing the operation hash value on the chain, and binding it with the timestamp; the smart contract automatically verifies the compliance of the archiving process, and an alarm is triggered for illegal operations.

[0019] Furthermore, the adaptive value evaluation module includes:

[0020] Dynamic weight allocation unit: analyzing historical data based on the LSTM model to predict the future importance of each indicator;

[0021] Destruction decision unit: combining the value score and compliance requirements to generate a destruction recommendation and push it to the life cycle management module.

[0022] Adopting the above technical solutions, the beneficial effects of the present invention are as follows:

[0023] 1. Improving classification efficiency and accuracy: Through the collaborative work of the multimodal data acquisition and the dynamic classification engine, it is possible to quickly and accurately classify a large number of archives. Compared with traditional methods, the classification efficiency is greatly improved, and the error rate is significantly reduced.

[0024] 2. Fully mining the value of archives: The adaptive value evaluation module can discover potential connections and new classification patterns among archives, helping users obtain archive information from different perspectives and giving full play to the value of archive resources.

[0025] 3. Enhancing system adaptability: The dynamic classification engine enables the system to automatically adjust the classification structure as the archive data changes, adapting to the evolving business needs and archive management requirements.

[0026] 4. Ensure the security of archives: The blockchain evidence storage module provides multi-level security protection for archives, effectively preventing the leakage of archive information and ensuring the security and integrity of archive data. Brief Description of the Drawings

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0028] Figure 1 is the system architecture diagram of the present invention;

[0029] Figure 2 is the reinforcement learning optimization flow chart of the dynamic classification engine;

[0030] Figure 3 is the schematic diagram of the operation hash generation of the blockchain evidence storage module. Detailed Embodiments

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0032] Embodiment 1: Multimodal Archive Classification

[0033] The user uploads an archive containing a "bridge design report" (text) and "steel structure drawings" (image).

[0034] Implementation of the multimodal data acquisition module: The user uploads an archive containing a text report and engineering drawings. The system extracts text keywords (such as "seismic design") and image features (such as "truss structure") through the multimodal module to generate a fused feature vector.

[0035] Implementation of the dynamic classification engine: The dynamic classification engine dynamically optimizes the classification rules based on the fused feature vector and user feedback, marks the archive as "infrastructure project - design phase", and triggers the filing process.

[0036] Implementation of the blockchain evidence storage module: The classification operation hash value is stored on the blockchain and bound to the timestamp to ensure data immutability and operation traceability.

[0037] Implementation of the intelligent lifecycle management module: Automatically trigger the filing, backup, sharing, or destruction process according to the classification label and generate an operation log.

[0038] Implementation of the Adaptive Value Evaluation Module: Through the dynamic analysis technology of time-series neural networks for technological advancement, social value, usage frequency, and compliance, adjust the index weights and generate a value report.

[0039] Example 2: Dynamic Optimization and Security Audit

[0040] The user corrects the classification results of "financial audit" type files multiple times.

[0041] The dynamic classification engine adjusts the threshold through DQN to improve the recall rate of this category.

[0042] The blockchain module records the correction operations and generates an audit trail.

[0043] After analyzing the usage frequency, the value evaluation module recommends retaining high-value files and destroying redundant copies.

[0044] The beneficial effects of the file classification and management system based on artificial intelligence are as follows:

[0045] 1. Efficient multimodal processing: Supports mixed data of text, images, and audio, and the classification accuracy is increased by 30%.

[0046] 2. Dynamic adaptive optimization: Through user feedback and reinforcement learning, the model iteration efficiency is increased by 50%.

[0047] 3. Secure and trustworthy evidence storage: Blockchain technology ensures that operations cannot be tampered with, meeting the requirements of compliance audits.

[0048] 4. Precise value management: Dynamically evaluate the value of files and reduce the storage cost by more than 20%.

[0049] The components not described in detail in this article are prior art.

[0050] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. An artificial intelligence-based archive classification management system, characterized by: Includes the following modules: Multimodal data acquisition module: used to integrate text, images, audio and metadata, extract features through natural language processing, convolutional neural network and speech recognition technology, and generate fused feature vectors; Dynamic classification engine: Based on the Transformer model and reinforcement learning framework, it dynamically optimizes classification rules according to user feedback and supports multi-label classification and fuzzy classification; Blockchain evidence storage module: Use smart contracts to record key operations in the classification process to ensure that data cannot be tampered with and operations are traceable; Intelligent lifecycle management module: automatically triggers archiving, backup, sharing or destruction processes based on classification tags and generates operation logs; Adaptive value assessment module: Dynamically analyzes technological advancement, social value, usage frequency and compliance through time-series neural networks, adjusts indicator weights and generates value reports.

2. The artificial intelligence-based archive classification management system according to claim 1, characterized in that: The multimodal data acquisition module comprises: Text analysis unit: uses the BERT model to extract semantic features; Image processing unit: Use ResNet-50 to extract image features and support cross-modal association; Metadata parsing unit: automatically extracts file creation time, author, and version information, and integrates them with content features.

3. The artificial intelligence-based archive classification management system according to claim 1, characterized in that: The dynamic classification engine optimizes the model by: The user's correction behavior on the classification result is used as a feedback signal, and a deep Q network is used to adjust the classification threshold; Introduce knowledge graphs to construct domain association rules and enhance the semantic understanding ability of fuzzy classification.

4. The artificial intelligence-based archive classification management system according to claim 1, characterized in that: The blockchain evidence storage module is specifically implemented as follows: classified operation hash values ​​are stored on the chain and bound to timestamps; smart contracts automatically verify the compliance of the archiving process, and illegal operations trigger alarms.

5. The artificial intelligence-based archive classification management system according to claim 1 is characterized by: The adaptive value assessment module includes: Dynamic weight allocation unit: Analyze historical data based on LSTM model and predict the future importance of each indicator; Destruction decision unit: Combines value scoring with compliance requirements to generate destruction recommendations and push them to the lifecycle management module.

Images