| With the continuous development and progress of the enterprise,the enterprise will accumulate a large number of documents,pictures,video and other related information,which is a huge amount of information.When an enterprise needs a certain document,it will take a lot of time to retrieve it.If you encounter hard disk damage and other issues,you have to face the risk of loss of information.These conditions exposed the limitations and deficiencies of the traditional system architecture,such as data from hard disk use restrictions,regular backup;slow speed retrieval of a large number of documents,unable to retrieve the contents of the document;could not find all the valuable files at once.Under the big data environment of enterprise knowledge management requires faster retrieval,storage scheme is more stable,the original architecture and technology put forward higher requirements,therefore,this paper presents and realization method for knowledge management of the small and medium enterprise cloud retrieval platform architecture,distributed and hybrid storage system solutions,implementation the document full-text retrieval,and the retrieval of document content,intelligent recommendation.The platform supports full format files,supports OFFICE,PDF,TXT,HTML and other formats of text extraction,supports online decompression of RAR and ZIP files,supports online preview of images,videos and other documents.Cloud retrieval platform uses intelligent retrieval,users can accurately retrieve the required files in the retrieval,can improve the success rate of the next search.When users use the retrieval function,they can recommend the relevant search terms,the hottest search terms,etc.,and can find the possible documents in the search results,and online preview.The retrieval tool supports the full format document title,description information and so on,supports the picture,the video on-line preview function.Users can view documents similar to the document on the right,which can be viewed and downloaded.The cloud retrieval platform developed by object-oriented method,the concrete research contents include:1.For SMEs cloud retrieval platform architecture designSMEs can not invest too much money in enterprise knowledge management,while customized,open source components,architecture and platform can effectively reduce the investment of enterprises.Therefore,this platform is designed for small and medium-sized enterprise cloud retrieval platform architecture,through the current situation of the enterprise,to build a meet the business requirements of the website architecture,cloud platform architecture customization,to realize the enterprise knowledge management.2.Hybrid solution for document storageThis platform mainly solves the problem of differential storage.At present,most of the cloud storage platforms are unable to effectively support the coexistence of large files and small files.Therefore,this system proposes a heuristic algorithm to solve the problem.Due to the support of HDFS for small files is not good,too much storage of small files will occupy a large amount of memory space,resulting in a decline in the speed of the machine processing.HBase is suitable for small file storage,storage mode for the file into the BASE64 encoding,and through the code conversion to download.Therefore,the system choose to use HBase to store small files,use HDFS to store large files,improve the efficiency of the system,reduce the consumption of unnecessary resources.3.Research and implementation of Intelligent Recommendation SystemIn order to enable users to find relevant and similar documents when searching,the platform needs to provide an effective intelligent recommendation system.The system needs to use Spark to run machine learning algorithm,Elasticsearch to achieve the retrieval needs of the system.The algorithm mainly includes the LDA clustering algorithm and the Elasticsearch retrieval algorithm,etc.the algorithm is applied to the related search terms recommendation,the most hot search word recommendation,the article clustering analysis and recommendation module.Then the machine learning algorithm can be used to optimize the retrieval accuracy of users,improve the retrieval success rate and improve the product quality.On the basis of the above study,the cloud platform construction,feasibility verification of distributed knowledge document storage solutions,using machine learning and data analysis to achieve intelligent knowledge recommendation and effectiveness of the system,and the performance of the hybrid scheme,recommendation algorithm is verified and real-time.At the theoretical level,a large amount of multi-source knowledge retrieval system is proposed,which is considered to read and write personalization and hybrid storage requirements in large data environment.In the application level,guide enterprises to successfully implement a similar project.Enterprise internal cloud retrieval system compared to Baidu cloud,360 cloud disk and so on,has the full text retrieval,the same name file save and other functions;compared to the number of enterprise cloud disk,enterprise internal cloud retrieval system can be deployed in accordance with existing equipment,reducing the initial investment of enterprises;from the security point of view,some enterprises will not consider using public cloud as the storage platform of enterprises.Therefore,the enterprise cloud retrieval platform is a read-write personalized,massive multi-source knowledge retrieval system considering mixed storage demand;enterprise cloud retrieval platform is a recommendation system based on large data analysis,real-time intelligent machine learning algorithm;enterprise cloud retrieval is a platform for enterprise users,enterprise knowledge management platform cloud platform.Enterprises can reduce operating costs,improve enterprise work efficiency of a framework based on B/S. |
PDF Full Text Request