Research And Implementation Of Data Retrieval On Large Data Center

In recent years,with the rapid development of information access technology,data generated by various applications grow exponentially.As new data sources gradually emerge,the content of the data becomes richer and richer,indicating that we have entered the era of big data.Faced with such huge amount of data,numerous data types and complex data structures,the efficient and unified management of data and valuable information access appears imminent.To solve this problem,various types of data management system emerges including the large data center system is one of them,providing a comprehensive retrieval and management capabilities for existing data resources.In this paper,the implementation methods of data retrieval is systematically studied from spatial and non-spatial data aspects,combines with application requirements of large data center systems.The aim is to accomplish the fast and accurate acquisition of target data and provide reference and basis for user's decision.(1)Spatial data retrieving section.Based on the analysis of basic features of spatial data and Oracle Spatial technology,this paper selects the spatial characteristics,attribute characteristics as well as spatial relation characteristics of spatial data as the basic direction of spatial data retrieval.The principles and implementation methods of the spatial data retrieval function are studied based on Oracle Spatial.The topological relationship query method is proposed based on the 9-intersection model while the unique expression of all topological relations is achieved.This paper describes the method of establishing spatial object buffer and accomplishes a typical distance relationship query--buffer query.Finally,the description information of spatial data is effectively organized and the efficient retrieval of massive spatial data is achieved.(2)Non-spatial data retrieving section.This paper chooses text data as the main study object and proposes the data retrieval method based on abstract.Among them,the key elements include:within Automatic Word,the hierarchical design architecture of word dictionary is utilized to optimize the Forward Maximum Matching algorithm,which further improves the automatic segmentation efficiency;within automatic abstracting,the automatic summarization method based on a combination of structure and statistics is proposed based on the existing abstracting method,which makes up for the shortcomings of existing methods in terms of scope and abstract readability;finally the full-text retrieval technology based on vocabulary is elaborated,breaking the limitations of previous keyword search.The idea of adding additional abstract to improve the efficiency of data retrieval lays the foundation for massive non-ideological data retrieval.(3)The dissertation of research into practice.The data retrieval functionality is developed regarding big data center systems,which confirms the accuracy and usefulness of the research.