Method for integrating and sharing internal resources of digital media in cloud environment

Abstract

The invention belongs to the technical field of cloud computing, and discloses a method for integrating and sharing internal resources of digital media in a cloud environment. The method comprises the following specific steps: S1, multi-modal resource discovery and metadata standardization; s2, distributed resource pool construction and intelligent indexing; s3, a dynamic scheduling algorithm driven by AI; s4, a permission and traceability mechanism endowed by the block chain; s5, opening a containerized micro-service interface; s6, resource optimization driven by user behaviors; and S7, edge-cloud collaborative dynamic caching is carried out. The method has the advantages that on one hand, through multi-objective optimization, delay load regulation and control and reinforcement learning, cost is reduced, response is improved, high-quality tasks are guaranteed, and resource scheduling is optimized; on one hand, a trust foundation is built for resource management and copyright is protected from behavior recording and permission grading to tracking protection by utilizing the characteristics of the block chain; in addition, micro-services are packaged, APIs and SDKs are provided, third-party access and integration are promoted, the development efficiency is improved, and all-around enabling is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of cloud computing, and specifically relates to a method for integrating and sharing internal resources of digital media in a cloud environment. Background Art

[0002] In today's digital wave, the digital media industry is booming but deeply mired in many dilemmas. The phenomenon of resource silos is extremely common, and the data resources of each department and system are independent of each other, making it difficult to interconnect and interoperate. This has created numerous obstacles to cross-departmental collaboration, greatly reducing work efficiency. Low sharing efficiency is also a major problem. The traditional method relying on centralized storage has a complex resource retrieval process, consuming a large amount of time and seriously affecting the speed of business progress. Moreover, the static permission management model is full of loopholes, with numerous security hazards and a high risk of data leakage.

[0003] Looking at existing cloud platforms, although they have a certain foundation, there are still obvious shortcomings in key technical aspects. Their dynamic resource scheduling ability is poor, and they are unable to flexibly allocate resources in the face of peaks and valleys in business volume. Their multi-modal data adaptability is insufficient, making it difficult to compatibly process different types of data such as images, audio, and video. The intelligent sharing mechanism is lacking, unable to meet the current demand for efficient and precise sharing.

[0004] Therefore, improvements and optimizations are needed. By combining blockchain technology to ensure data security and traceability, AI-driven dynamic scheduling algorithms to optimize resource allocation, and a microservices containerized architecture to enhance system flexibility, the traditional model can be broken, leading to a new era of intelligent, elastic, and decentralized resource management. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for integrating and sharing internal resources of digital media in a cloud environment to solve the problems raised in the above background art.

[0006] To achieve the above objective, the present invention provides the following technical solution: The specific steps of the method for integrating and sharing internal resources of digital media in a cloud environment are as follows:

[0007] S1: Multi-modal resource discovery and metadata standardization: Using AI image recognition technology, the convolutional neural network CNN accurately classifies picture scenes and identifies human features to assist in determining the theme; for audio and video, speech recognition, natural language processing NLP, and key frame extraction are used to clarify the content theme; connect to the copyright library, and through watermarking and metadata parsing, extract copyright information, and build a metadata framework based on the JSON-LD specification, covering general and specific attributes to support cross-format retrieval;

[0008] S2: Distributed Resource Pool Construction and Intelligent Indexing: Adopt a hybrid cloud storage architecture, stratify hot and cold data according to popularity. Store hot data in the high-speed layer and migrate cold data to low-cost media. Dynamically adjust the levels based on intelligent strategies. Combine with a knowledge graph to visually model the resource associations in the film and music fields. Use a GNN to optimize the index path, learn its structural features, and predict fast traversal paths. For example, when retrieving an actor's works, it can accurately jump, improving the retrieval efficiency by over 30% and shortening the response time.

[0009] S3: AI-driven Dynamic Scheduling Algorithm: Design a multi-objective optimization model focusing on cost, latency, and load balancing. In terms of cost, select the optimal configuration in combination with the cloud billing model, reduce computing power during off-peak hours, and choose low-cost storage. Optimize latency by real-time monitoring metrics and prediction models, and reserve bandwidth for 4K rendering. Use a distributed algorithm for load balancing to monitor node loads and dynamically allocate tasks. Utilize RL technology, where the agent learns strategies based on interactive feedback and adaptively adjusts to ensure high-priority tasks.

[0010] S4: Blockchain-empowered Permission and Traceability Mechanism: Deploy a private chain to record resource access, modification, and sharing information in encrypted blocks, covering key elements of operations, ensuring that operations are traceable and tamper-proof. Use blockchain smart contracts to achieve dynamic permission grading and time-limited resource authorization. In case of copyright disputes, the chain records can be traced back to clarify ownership. Combine with digital certificates to assign unique identifiers to original works and strengthen copyright protection.

[0011] S5: Containerized Microservice Interface Opening: Package core functions such as transcoding and watermark embedding into microservices, managed by Kubernetes, which can achieve automated deployment, elastic scaling, and load balancing. For example, automatically expand transcoding containers during peak hours and provide a RESTful API that complies with the HTTP specification for external use. It is convenient for third parties to call using common languages, and the API documentation details the interface details to lower the threshold. It also comes with multi-language SDKs. The Java SDK simplifies transcoding and accelerates integration.

[0012] S6: User Behavior-driven Resource Optimization: Based on federated learning, collect multi-source operation logs while protecting privacy. Use text mining to extract high-frequency search terms, gain insights into user interest hotspots, analyze sharing paths, and clarify resource dissemination rules, such as the sharing situation of creative materials among designers. Optimize the storage distribution accordingly. Also, based on user behavior analysis and prediction models, estimate demand and pre-deploy popular resources to edge nodes to improve the experience.

[0013] S7: Edge-Cloud Collaborative Dynamic Caching: Deploy lightweight caching clusters at edge nodes close to user terminals. Leveraging the low-latency advantage of edge computing, set up caching servers beside 5G base stations to serve surrounding intelligent terminals, reducing transmission latency. Using the LSTM algorithm, predict hot resources based on historical access and time series, preload popular content. If the edge cache is hit, directly supply the data to relieve the cloud, and communicate with the cloud in real time for dynamic updates to ensure the response speed.

[0014] Preferably, the multi-modal resource discovery and metadata standardization in S1 refer to: With the help of advanced AI image recognition technology, use convolutional neural networks to accurately classify the scene elements of pictures. For pictures of people, it can also identify facial features and action postures to assist in judging the theme. In the understanding of audio-visual content, use speech recognition to convert audio into text, and combine natural language processing to extract key information. Determine the video theme through key frame extraction and image understanding. In terms of copyright information extraction, connect to the copyright database, use digital watermark recognition and metadata parsing to obtain key copyright details, and issue a warning when suspected infringement is found. When constructing a unified metadata framework, according to the JSON-LD and Schema.org specifications, cover general and specific domain attributes, provide a complete description for resources in different formats such as videos and 3D models, and lay a solid foundation for cross-format retrieval.

[0015] Preferably, the construction of the distributed resource pool and intelligent indexing in S2 refer to: Adopt a hybrid cloud storage architecture, perform hot and cold stratification according to the data access heat. Store hot data in the high-speed cloud storage layer to ensure fast response to popular resources. Migrate cold data to low-cost media, monitor the access frequency through intelligent strategies, and dynamically adjust the storage level. Combine knowledge graph technology to visually model the associations between resources in the film and music fields, connect peripheral nodes with core nodes to form a knowledge network, provide semantic clues, and use graph neural networks (GNNs) to optimize the indexing path. The (GNN) model learns the network structure features and predicts the fast traversal path, significantly shortening the retrieval response time.

[0016] Preferably, the specific steps of the AI-driven dynamic scheduling algorithm in S3 are as follows:

[0017] Step 1: Design a multi-objective optimization model and plan the cost strategy: Comprehensively consider cost, latency, and load balancing, combined with the billing model of cloud service providers, covering storage, computing, and network transmission costs. On the premise of meeting performance, formulate an optimal resource allocation plan. Reduce computing redundancy and select low-cost storage during off-peak hours to avoid resource waste;

[0018] Step 2: Delay optimization and load balancing control: Real-time monitoring of network status and server load indicators, using prediction models to estimate task execution delays, reserving bandwidth for high-real-time tasks and 4K real-time rendering. At the same time, using distributed load balancing algorithms to monitor server node loads, dynamically assign tasks, and timely adjust task assignments for overload risk nodes to ensure system stability.

[0019] Step 3: Implement adaptive scheduling with the help of reinforcement learning: Use reinforcement learning technology to allow the RL agent to interact with the environment, try different resource allocation actions, learn the optimal strategy based on feedback from task completion results, optimize resource allocation in real time, and prioritize the execution of high-priority tasks.

[0020] Preferably, the specific steps of the permission and traceability mechanism enabled by blockchain in S4 are as follows:

[0021] Step 1: Deploy a private chain to record operation behaviors: Deploy a private chain to record the access, modification and sharing of resources in the form of encrypted blocks. Each block covers the operation time, subject, object and key information of details to ensure that the operation track cannot be tampered with and is traceable throughout the process;

[0022] Step 2: Use smart contracts to achieve permission grading: Based on blockchain-based smart contracts, dynamic permission grading is performed. For limited-time promotion resources, time-limited authorization is set through smart contracts to grant users specific access rights within a specified time. When the authorization expires, the permission is automatically revoked to ensure the reasonable use of resources and the rights and interests of copyright holders;

[0023] Step 3: Combine technology to strengthen copyright tracking and protection: When a copyright dispute occurs, quickly trace back the operation records on the blockchain to clearly present the entire process from resource creation to circulation, so as to determine the ownership of the copyright. At the same time, combined with digital certificate technology, a unique copyright mark is issued for the original work to further strengthen copyright protection.

[0024] Preferably, the specific steps of opening the containerized microservice interface in S5 are as follows:

[0025] Step 1: Encapsulate microservices and perform container orchestration management: Encapsulate the core functions of transcoding and watermark embedding into independently running microservices, and manage them with the help of Kubernetes. Kubernetes implements container orchestration of microservices, achieves automated deployment, elastic scaling, and load balancing, and automatically increases or decreases container instances according to policies during peak video transcoding hours to prevent task backlogs.

[0026] Step 2: Provide RESTful API and lower the access threshold: Provide a RESTful API that complies with the Hypertext Transfer Protocol (HTTP) protocol specification externally. The interface is unified and concise, facilitating third-party developers to call it using common programming languages. At the same time, elaborate on the functions, parameter formats, and return value types of each interface through API documentation to reduce the access difficulty;

[0027] Step 3: Provide a Software Development Kit (SDK) to simplify the development process and accelerate integration: Provide an SDK to encapsulate convenient function libraries for different mainstream programming languages, further simplifying the development process. Taking the Java SDK as an example, it provides a one-click video transcoding call method. Developers only need to input a small number of parameters to achieve complex functions, significantly shortening the development cycle and accelerating the integration of third-party applications with the core functions of the platform.

[0028] Preferably, the user behavior-driven resource optimization in S6 refers to, based on federated learning technology, collecting and analyzing multi-source operation logs while protecting user privacy, extracting high-frequency search terms from retrieval records through text mining, understanding user interest hotspots, analyzing user sharing paths, mastering resource dissemination rules, optimizing storage distribution for creative materials frequently shared by users in the design industry, reducing cross-regional access latency, and based on the results of user behavior analysis, using a prediction model to estimate demand trends, pre-deploying popular resources to the corresponding regional edge nodes in advance so that users can quickly obtain resources, effectively improving the overall user experience.

[0029] Preferably, the edge-cloud collaborative dynamic caching in S7 refers to deploying a lightweight cache cluster at the edge node close to the user terminal. With the low-latency characteristics of edge computing, it can quickly respond to local requests. Set up a cache server beside the 5G base station to cache data for surrounding intelligent terminals nearby, reducing the backhaul transmission latency. Use the LSTM prediction algorithm to analyze user request behaviors, and accurately predict hot resources based on the user's historical access patterns and time series factors. Before the video entertainment peak from 8 to 10 pm every night, pre-load popular TV series into the edge cache. When the edge cache is hit, directly supply the data, reducing the load on the cloud, avoiding waste of bandwidth and computing resources, and the edge node communicates with the cloud in real time to dynamically update the cache according to the actual user access, ensuring the cache popularity and effectiveness, and continuously improving the terminal response speed.

[0030] The beneficial effects of the present invention are as follows:

[0031] 1. The present invention plans costs by designing a multi-objective optimization model, comprehensively considering various cost factors, reasonably adjusting during off-peak hours, greatly reducing resource waste and operating costs. Secondly, through delay optimization and load balancing control, it real-time monitors the network and server status, predicts task delays, reserves bandwidth for critical tasks, and flexibly allocates tasks using distributed algorithms to ensure the stable operation of the system and effectively improve the response speed. Finally, it achieves adaptive scheduling through reinforcement learning, continuously optimizes resource allocation based on task feedback, guarantees high-priority tasks, and improves the overall execution efficiency.

[0032] 2. The present invention, firstly, deploys a private chain to record operation behaviors. With the characteristics of immutability and full traceability, it constructs a reliable trust foundation for resource management, providing evidence for both daily operation and maintenance and problem troubleshooting; secondly, uses smart contracts to achieve permission grading and flexibly control time-limited resources, which not only meets the time-limited needs of users but also ensures the reasonable allocation of resources, safeguards the interests of copyright holders, and improves resource utilization efficiency; thirdly, combines technologies to strengthen copyright tracking and protection. In case of copyright disputes, it can quickly trace back, accurately determine ownership, and the digital certificate further highlights the originality of the work, escorting copyright in all aspects and reducing losses caused by infringement disputes.

[0033] 3. The present invention, firstly, encapsulates microservices and container orchestration management, independently encapsulates key functions, and uses Kubernetes to achieve automated deployment and elastic scaling to flexibly handle peak periods of video transcoding, eliminate task backlogs, and improve efficiency; secondly, provides RESTful APIs and lowers the threshold, standardizes interfaces and details documents to facilitate third parties to access using common languages, breaking development barriers and promoting cooperation; thirdly, supports SDKs to simplify processes and accelerate integration. Function libraries in different languages are in place, and the Java SDK makes transcoding easy to start. With a small number of parameters, complex operations can be completed, shortening the cycle and helping third-party applications quickly integrate core functions. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a flowchart of the method for integrating and sharing internal resources of digital media in the cloud environment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0036] As Figure 1 shown, the embodiments of the present invention provide the following specific steps for the method of integrating and sharing internal resources of digital media in the cloud environment:

[0037] S1: Multimodal Resource Discovery and Metadata Standardization: Utilize AI image recognition technology. The convolutional neural network (CNN) accurately classifies picture scenes and identifies human features to assist in determining the theme. For audio and video, speech recognition, natural language processing (NLP), and key frame extraction are used to clarify the content theme. Connect to the copyright library, parse and extract copyright information through watermarks and metadata, and build a metadata framework based on the JSON-LD specification, covering general and specific attributes to support cross-format retrieval.

[0038] S2: Distributed Resource Pool Construction and Intelligent Indexing: Adopt a hybrid cloud storage architecture. Stratify hot and cold data according to popularity. Store hot data in the high-speed layer and migrate cold data to low-cost media. Dynamically adjust the levels based on intelligent strategies. Combine with a knowledge graph to visually model the associations of resources in the film and television and music fields. Use a graph neural network (GNN) to optimize the indexing path. It learns the structural features and predicts the fast traversal path. For example, when retrieving an actor's works, it can accurately jump, improving the retrieval efficiency by more than 30% and shortening the response time.

[0039] S3: AI-Driven Dynamic Scheduling Algorithm: Design a multi-objective optimization model focusing on cost, latency, and load balancing. In terms of cost, select the optimal configuration in combination with the cloud billing model, reduce computing power during off-peak hours, and choose low-cost storage. Optimize latency by real-time monitoring metrics and prediction models, and reserve bandwidth for 4K rendering. Use a distributed algorithm for load balancing to monitor node loads and dynamically allocate tasks. Utilize reinforcement learning (RL) technology. The agent learns strategies based on interactive feedback and adaptively adjusts to ensure high-priority tasks.

[0040] S4: Blockchain-Enabled Permission and Traceability Mechanism: Deploy a private chain to record resource access, modification, and sharing information in encrypted blocks, covering key elements of operations, ensuring that operations are traceable and tamper-proof. Use blockchain smart contracts to achieve dynamic permission grading and time-limited resource authorization. In case of copyright disputes, the chain records can be traced back to clarify the ownership. Combine with digital certificates to assign a unique identifier to original works and strengthen copyright protection.

[0041] S5: Containerized Microservice Interface Opening: Package core functions such as transcoding and watermark embedding into microservices, which are managed by Kubernetes. It can achieve automated deployment, elastic scaling, and load balancing. For example, automatically expand transcoding containers during peak periods. Provide a RESTful API that complies with the HTTP specification for external use. It is convenient for third parties to call using common languages. Moreover, the API documentation details the interface details to lower the threshold and also provides multi-language SDKs. The Java SDK simplifies transcoding and accelerates integration.

[0042] S6: User Behavior-driven Resource Optimization: Based on federated learning, multi-source operation logs are collected while protecting privacy. High-frequency retrieval words are extracted using text mining to gain insights into users' interest hotspots, analyze sharing paths, clarify resource dissemination patterns, and understand the sharing situation of creative materials among designers. Based on this, the storage distribution is optimized. Additionally, through user behavior analysis and using a prediction model to estimate demand, popular resources are pre-deployed to edge nodes to enhance the experience.

[0043] S7: Edge-Cloud Collaborative Dynamic Caching: A lightweight caching cluster is deployed at edge nodes close to user terminals. Leveraging the low-latency advantage of edge computing, a caching server is set up beside 5G base stations to serve surrounding intelligent terminals, reducing transmission latency. Using the LSTM algorithm, hot resources are predicted based on historical access and time series, and popular content is pre-loaded. If the edge cache is hit, data is directly supplied, relieving the cloud, and it communicates with the cloud in real-time for dynamic updates to ensure the response speed.

[0044] 2. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1, characterized in that: the multi-modal resource discovery and metadata standardization in S1 refers to using advanced AI image recognition technology, employing a convolutional neural network to accurately classify the scene elements of pictures, and for pictures of people, it can also identify facial features and action postures to assist in judging the theme. In the understanding of audio-visual content, speech recognition is used to convert audio into text, and key information is extracted by combining natural language processing. The video theme is determined through key frame extraction and image understanding. In terms of copyright information extraction, the copyright database is accessed, and key copyright details are obtained using digital watermark recognition and metadata parsing. If suspected infringement is found, an early warning is issued. When constructing a unified metadata framework, according to the JSON-LD and Schema.org specifications, it covers general and specific domain attributes, providing a complete description for resources in different formats such as videos and 3D models, and laying a solid foundation for cross-format retrieval.

[0045] On the one hand, leveraging cutting-edge AI technologies, such as using a convolutional neural network to subdivide picture elements, identify human features and postures to assist in determining the theme, and for audio-visual content, using speech, natural language processing, and key frame and image understanding technologies to clarify the content theme. On the other hand, accessing the copyright library, using watermark and metadata parsing to detect infringement, and constructing a unified metadata framework according to specifications such as JSON-LD to accurately profile various resources and empower cross-format retrieval.

[0046] 3. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1 is characterized in that: the construction of the distributed resource pool and intelligent indexing in S2 refers to adopting a hybrid cloud storage architecture, performing hot and cold stratification according to the data access heat, storing hot data in the high-speed cloud storage layer to ensure fast response of popular resources, migrating cold data to low-cost media, monitoring the access frequency through intelligent policies, dynamically adjusting the storage level, combining knowledge graph technology, visually modeling the associations between resources in the film and television and music fields, connecting peripheral nodes with core nodes to form a knowledge network, providing semantic clues, optimizing the indexing path by using a graph neural network (GNN), the (GNN) model learning the network structure features, predicting the fast traversal path, and significantly shortening the retrieval response time.

[0047] First, adopt hybrid cloud storage, stratify hot and cold according to heat, place hot data in the high-speed layer to ensure instant response of popular resources, transfer cold data to low-cost media, intelligently monitor the frequency, and dynamically adjust the level; then combine the knowledge graph to visually associate film and television and music resources, build a knowledge network to provide semantic assistance; finally, use GNN to optimize indexing, let it learn the structure to find the fastest path, greatly improve the retrieval efficiency, and significantly shorten the response time. GNN is the abbreviation of Graph Neural Network.

[0048] 4. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1 is characterized in that: the specific steps of the AI-driven dynamic scheduling algorithm in S3 are as follows:

[0049] Step 1: Design a multi-objective optimization model and plan the cost strategy: comprehensively consider cost, latency, and load balancing, combine the billing model of cloud service providers, cover storage, computing, and network transmission costs, and formulate an optimal resource allocation plan on the premise of meeting performance. When it is off-peak, reduce computing power redundancy and select low-cost storage to avoid resource waste;

[0050] Step 2: Perform latency optimization and load balancing control: real-time monitor the network status and server load indicators, use a prediction model to estimate the task execution latency, reserve bandwidth for high-real-time tasks, such as 4K real-time rendering. At the same time, adopt a distributed load balancing algorithm to monitor the load conditions of server nodes, dynamically allocate tasks, and timely adjust the task allocation for nodes at risk of overload to ensure system stability;

[0051] Step 3: Achieve adaptive scheduling by means of reinforcement learning: use reinforcement learning technology to enable the RL agent to interact with the environment, try different resource allocation actions, learn the optimal strategy based on the feedback of the task completion effect, and optimize the resource allocation in real time to prioritize the execution of high-priority tasks.

[0052] The AI-driven dynamic scheduling algorithm in S3 consists of three steps: first, design a multi-objective optimization model and cost strategy, integrate cost, latency, and load balancing factors, and formulate a resource allocation plan based on the cloud billing model, reduce computing power during off-peak hours, and select low-cost storage; second, latency optimization and load balancing regulation, real-time monitoring of network and server loads, reserve bandwidth for high-real-time tasks, and use distributed algorithms to dynamically allocate tasks and deal with overloads; third, with the help of reinforcement learning adaptive scheduling, RL agents interactively learn, optimize allocation based on feedback, and give priority to high-priority tasks.

[0053] 5. The integrated sharing method of digital media internal resources in a cloud environment according to claim 1 is characterized in that: the specific steps of the permission and traceability mechanism enabled by blockchain in S4 are as follows:

[0054] Step 1: Deploy a private chain to record operation behaviors: Deploy a private chain to record the access, modification and sharing of resources in the form of encrypted blocks. Each block covers the operation time, subject, object and key information of details to ensure that the operation track cannot be tampered with and is traceable throughout the process;

[0055] Step 2: Use smart contracts to achieve permission grading: Based on blockchain-based smart contracts, dynamic permission grading is performed. For limited-time promotion resources, time-limited authorization is set through smart contracts to grant users specific access rights within a specified time. When the authorization expires, the permission is automatically revoked to ensure the reasonable use of resources and the rights and interests of copyright holders;

[0056] Step 3: Combine technology to strengthen copyright tracking and protection: When a copyright dispute occurs, quickly trace back the operation records on the blockchain to clearly present the entire process from resource creation to circulation, so as to determine the ownership of the copyright. At the same time, combined with digital certificate technology, a unique copyright mark is issued for the original work to further strengthen copyright protection.

[0057] The permission and traceability mechanism enabled by blockchain in S4 is divided into three steps: first, deploy a private chain to record various resource operations in encrypted blocks, including key information, to ensure that operations are traceable and cannot be tampered with; second, dynamically grade permissions through smart contracts, such as time-limited resources can be authorized with a time limit, and the rights will be automatically revoked upon expiration, balancing resource utilization and the interests of the copyright owner; finally, in the event of copyright disputes, the blockchain operation records can be traced back to restore the entire resource transfer process to assist in determining ownership, and digital certificates can be combined to give original works a unique identifier to enhance copyright protection.

[0058] 6. The method for integrating and sharing internal resources of digital media in a cloud environment according to claim 1, characterized in that: the specific steps of opening the containerized microservice interface in S5 are as follows:

[0059] Step 1: Package microservices and perform container orchestration management: Package the transcoding and watermark embedding core functions into independently running microservices and manage them with the help of Kubernetes. Kubernetes realizes the container orchestration of microservices, achieving automated deployment, elastic scaling, and load balancing. When the video transcoding peak occurs, it automatically increases or decreases container instances according to the policy to prevent task backlogs.

[0060] Step 2: Provide RESTful API and lower the access threshold: Provide a RESTful API that follows the Hypertext Transfer Protocol (HTTP) protocol specification externally. The interface is unified and concise, facilitating third-party developers to call with common development languages. At the same time, the functions, parameter formats, and return value types of each interface are elaborated in detail through API documentation to reduce the access difficulty.

[0061] Step 3: Provide a Software Development Kit (SDK) to simplify the development process and accelerate integration: Provide an SDK, which encapsulates convenient function libraries for different mainstream development languages to further simplify the development process. Taking the Java SDK as an example, it provides a one-click video transcoding call method. Developers only need to input a small number of parameters to achieve complex functions, greatly shortening the development cycle and accelerating the integration of third-party applications with the platform's core functions.

[0062] Kubernetes is an open-source container orchestration engine used for automating the deployment, scaling, and management of containerized applications. RESTful API (Representational State Transfer API), that is, the Representational State Transfer application programming interface, is an API designed based on the REST architectural style.

[0063] 7. The method for integrated sharing of digital media internal resources in a cloud environment according to claim 1, characterized in that: the user behavior-driven resource optimization in S6 refers to, based on federated learning technology, collecting and analyzing multi-source operation logs while protecting user privacy, extracting high-frequency search terms from retrieval records through text mining, insight into user interest hotspots, analyzing user sharing paths, mastering resource dissemination rules, for creative materials frequently shared by users in the design industry, optimizing storage distribution, reducing cross-regional access latency, according to the results of user behavior analysis, using a prediction model to estimate demand trends, pre-deploying popular resources to the corresponding regional edge nodes in advance, enabling users to quickly obtain resources, and effectively improving the overall user experience.

[0064] In S6, user behavior-driven resource optimization uses federated learning. On the premise of ensuring privacy, it collects multi-source operation logs. On the one hand, it mines text retrieval records to obtain high-frequency words, gains insights into interest hotspots, analyzes sharing paths to master dissemination rules, and optimizes the storage distribution of specific materials accordingly. On the other hand, based on behavior analysis, it uses a prediction model to estimate demand, pre-deploys popular resources to edge nodes, improves the user acquisition speed, and optimizes the overall experience.

[0065] 8. The integrated sharing method of digital media internal resources in the cloud environment according to claim 1, wherein: the edge-cloud collaborative dynamic caching in S7 refers to deploying a lightweight caching cluster at the edge node close to the user terminal. With the low-latency characteristic of edge computing, it can quickly respond to local requests. A caching server is set up beside the 5G base station to cache data for surrounding intelligent terminals nearby, reducing the backhaul transmission delay. The LSTM prediction algorithm is used to analyze user request behavior. Based on the user's historical access pattern and time series factors, it accurately predicts hot resources. Before the video entertainment peak from 8 to 10 pm every night, popular TV series are pre-loaded into the edge cache. When the edge cache is hit, data is directly provided, reducing the load on the cloud, avoiding waste of bandwidth and computing resources. Moreover, the edge node communicates with the cloud in real time and updates the cache dynamically according to the actual user access, ensuring the cache popularity and effectiveness, and continuously improving the terminal response speed.

[0066] LSTM (Long Short-Term Memory) is a special recurrent neural network (RNN), designed specifically to solve the problem of gradient vanishing or gradient explosion that general RNNs have when dealing with long sequence data. It can effectively learn and remember long-distance dependencies and has been widely used in many fields.

[0067] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or elements inherent to such process, method, article or device.

[0068] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made therein without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. Method for integrated sharing of internal resources of digital media in cloud environment, characterized in that: The specific steps for the integrated sharing of digital media internal resources in the cloud environment are as follows: S1: Multimodal resource discovery and metadata standardization: Utilize AI image recognition technology, and the convolutional neural network CNN accurately classifies picture scenes and identifies human features to assist in determining the theme; for audio and video, speech recognition, natural language processing NLP, and key frame extraction are used to clarify the content theme; connect to the copyright library, and watermark and metadata parsing are used to extract copyright information. Based on the JSON-LD specification, a metadata framework is constructed, covering general and specific attributes to support cross-format retrieval; S2: Construction of a distributed resource pool and intelligent indexing: Adopt a hybrid cloud storage architecture, stratify hot and cold data according to popularity, store hot data in the high-speed layer, and migrate cold data to low-cost media. Rely on intelligent strategies to dynamically adjust the levels. Combine with a knowledge graph to visually model the resources in the film and television and music fields, and use GNN to optimize the indexing path, which learns the structural features, predicts the fast traversal path. For example, when retrieving an actor's works, it can accurately jump, improving the retrieval efficiency by more than 30% and shortening the response time; S3: AI-driven dynamic scheduling algorithm: Design a multi-objective optimization model focusing on cost, latency, and load balancing. In terms of cost, combine with the cloud billing model to select the optimal configuration, reduce computing power during off-peak hours, and select low-cost storage. Optimize latency by real-time monitoring indicators and prediction models, and reserve bandwidth for 4K rendering; for load balancing, use a distributed algorithm to monitor node loads and dynamically allocate tasks. Utilize RL technology, and the agent learns strategies based on interactive feedback and adapts automatically to ensure high-priority tasks; S4: Permission and traceability mechanism empowered by blockchain: Deploy a private chain, record resource access, modification, and sharing information in encrypted blocks, covering key elements of operations, ensuring that operations are traceable and tamper-proof. Use blockchain smart contracts to achieve dynamic permission grading and time-limited resource authorization. In case of copyright disputes, the chain records can be traced back to clarify the ownership. Combine with digital certificates to assign a unique identifier to original works to strengthen copyright protection; S5: Opening of containerized microservice interfaces: Package core functions such as transcoding and watermark embedding into microservices, which are managed by Kubernetes. It can achieve automated deployment, elastic scaling, and load balancing. For example, automatically expand transcoding containers during peak periods, and provide a RESTful API that follows the HTTP specification externally, facilitating third parties to call in common languages. Moreover, the API documentation details the interface details to lower the threshold, and also provides multi-language SDKs. The Java SDK simplifies transcoding and accelerates integration; S6: Resource optimization driven by user behavior: Based on federated learning, collect multi-source operation logs while protecting privacy, use text mining to extract high-frequency search terms, understand user interest hotspots, analyze sharing paths, clarify resource dissemination rules, and the sharing situation of creative materials among designers. Accordingly, optimize the storage distribution, and also rely on user behavior analysis and use a prediction model to estimate demand, pre-deploy popular resources to edge nodes to improve the experience; S7: Edge-Cloud Collaborative Dynamic Caching: Deploy lightweight caching clusters at edge nodes close to user terminals. Leveraging the low-latency advantage of edge computing, cache servers are set up beside 5G base stations to serve surrounding intelligent terminals, reducing transmission latency. Using the LSTM algorithm, predict hot resources based on historical access and time series, preload popular content. If the edge cache is hit, directly supply the data to relieve the cloud, and communicate with the cloud in real time for dynamic updates to ensure the response speed.

2. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1, characterized in that: The multi-modal resource discovery and metadata standardization in S1 refer to leveraging advanced AI image recognition technology, using convolutional neural networks to accurately classify the scene elements of pictures. For pictures of people, it can also identify facial features and action postures to assist in judging the theme. In the understanding of audio-visual content, use speech recognition to convert audio to text, and combine natural language processing to extract key information. Determine the video theme through key frame extraction and image understanding. In terms of copyright information extraction, connect to the copyright database, use digital watermark recognition and metadata parsing to obtain key copyright details, and issue a warning when suspected infringement is found. When constructing a unified metadata framework, based on the JSON-LD and Schema.org specifications, cover general and specific domain attributes, provide a complete description for different format resources such as videos and 3D models, and lay a solid foundation for cross-format retrieval.

3. The integrated sharing method for internal resources of digital media in a cloud environment according to claim 1, characterized in that: The construction of a distributed resource pool and intelligent indexing in S2 refer to adopting a hybrid cloud storage architecture, performing hot and cold stratification according to the data access heat. Store hot data in the high-speed cloud storage layer to ensure fast response to popular resources, and migrate cold data to low-cost media. Monitor the access frequency through intelligent strategies and dynamically adjust the storage level. Combine knowledge graph technology to visually model the associations between resources in the film and television and music fields, connect the core nodes to the surrounding nodes to form a knowledge network, provide semantic clues, and use the graph neural network GNN to optimize the indexing path. The GNN model learns the network structure features and predicts the fast traversal path, significantly shortening the retrieval response time.

4. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1, characterized in that: The specific steps of the AI-driven dynamic scheduling algorithm in S3 are as follows: Step 1: Design a multi-objective optimization model and plan the cost strategy: Comprehensively consider cost, latency, and load balancing, combined with the billing model of cloud service providers, covering storage, computing, and network transmission costs. On the premise of meeting performance, formulate the optimal resource allocation plan. When it is off-peak, reduce computing redundancy and select low-cost storage to avoid resource waste. Step 2: Conduct latency optimization and load balancing control: Real-time monitor the network status and server load indicators, use the prediction model to estimate the task execution latency, reserve bandwidth for high-real-time tasks such as 4K real-time rendering. At the same time, adopt a distributed load balancing algorithm to monitor the load conditions of server nodes, dynamically allocate tasks, and timely adjust the task allocation for nodes at risk of overload to ensure system stability. Step 3: Achieve adaptive scheduling through reinforcement learning: Use reinforcement learning technology to let the RL agent interact with the environment, try different resource allocation actions, and learn the optimal strategy based on the feedback of the task completion effect, and optimize the resource allocation in real time to give priority to ensuring the execution of high-priority tasks.

5. The integrated sharing method for internal resources of digital media in a cloud environment according to claim 1, characterized in that: The specific steps of the permission and traceability mechanism enabled by blockchain in S4 are as follows: Step 1: Deploy a private chain to record operation behaviors: Deploy a private chain to record the access, modification and sharing of resources in the form of encrypted blocks. Each block covers the operation time, subject, object and key information of details to ensure that the operation track cannot be tampered with and is traceable throughout the process; Step 2: Use smart contracts to achieve permission grading: Based on blockchain-based smart contracts, dynamic permission grading is performed. For limited-time promotion resources, time-limited authorization is set through smart contracts to grant users specific access rights within a specified time. When the authorization expires, the permission is automatically revoked to ensure the reasonable use of resources and the rights and interests of copyright holders; Step 3: Combine technology to strengthen copyright tracking and protection: When a copyright dispute occurs, quickly trace back the operation records on the blockchain to clearly present the entire process from resource creation to circulation, so as to determine the ownership of the copyright. At the same time, combined with digital certificate technology, a unique copyright mark is issued for the original work to further strengthen copyright protection.

6. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1, characterized in that: The specific steps for opening the containerized microservice interface in S5 are as follows: Step 1: Encapsulate microservices and perform container orchestration management: Encapsulate the core functions of transcoding and watermark embedding into independently running microservices, and manage them with the help of Kubernetes. Kubernetes implements container orchestration of microservices, achieves automated deployment, elastic scaling, and load balancing, and automatically increases or decreases container instances according to policies during peak video transcoding hours to prevent task backlogs. Step 2: Provide RESTful API and lower the access threshold: Provide RESTful API that complies with the Hypertext Transfer Protocol (HTTP) protocol specification. The interface is unified and concise, which is convenient for third-party developers to call using common development languages. At the same time, the API document elaborates on the function, parameter format and return value type of each interface in detail to reduce the difficulty of access. Step 3: Supporting software development kit SDK simplifies the development process and accelerates integration: Supporting SDK is provided to encapsulate convenient function libraries for different mainstream development languages to further simplify the development process. Taking JavaSDK as an example, it provides a one-click video transcoding calling method. Developers only need to enter a small number of parameters to implement complex functions, which greatly shortens the development cycle and speeds up the integration of third-party applications and platform core functions.

7. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1, characterized in that: The user behavior-driven resource optimization in S6 refers to the collection and analysis of multi-source operation logs based on federated learning technology while protecting user privacy, extracting high-frequency search terms from search records through text mining, gaining insights into user interest hotspots, analyzing user sharing paths, and mastering resource dissemination rules. For creative materials frequently shared by users in the design industry, for example, storage distribution is optimized, cross-regional access latency is reduced, and based on the results of user behavior analysis and with the help of prediction models, demand trends are estimated, and popular resources are pre-deployed to edge nodes in the corresponding regions in advance, so that users can quickly obtain resources and effectively improve the overall user experience.

8. The method for integrated sharing of internal resources of digital media in a cloud environment according to claim 1, characterized in that: The dynamic caching of edge-cloud collaboration in S7 refers to deploying a lightweight caching cluster at the edge nodes close to user terminals. With the low-latency characteristics of edge computing, it can quickly respond to local requests. A caching server is set up beside the 5G base station to cache data for surrounding intelligent terminals nearby, reducing the backhaul transmission latency. The LSTM prediction algorithm is used to analyze user request behaviors. Based on the user's historical access patterns and time series factors, hot resources are accurately predicted. Before the video entertainment peak from 8 to 10 pm every night, popular TV series are pre-loaded into the edge cache. When the edge cache is hit, data is directly provided, reducing the load on the cloud and avoiding waste of bandwidth and computing resources. Moreover, the edge nodes communicate with the cloud in real time, and the cache is dynamically updated according to the actual user access, ensuring the cache popularity and effectiveness, and continuously improving the terminal response speed.

Images