Font Size: a A A

Research On The Unstructured Data Ontology And Relevant Algorithms

Posted on:2016-06-21Degree:MasterType:Thesis
Country:ChinaCandidate:X W LiuFull Text:PDF
GTID:2308330464970751Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
With the emergency of the social networks, cloud computing and big data concepts and technology, there are large amounts of data are generated all the time, and in which the ratio of text, images, audio, video and other types of unstructured data by the total amount of data is gradually increasing. Meanwhile, under the background of big data, the real-time sharing and analysis of data will not only bring immeasurable economic value, but also greatly promote the progress of society, and all this based on the premise that effective management of the data, which accounts for a large reasonable proportion of unstructured data management is particularly important. The management for unstructured data is always a big problem, mainly with different management techniques in the diversification of unstructured data, it’s difficult to unified representation.Aiming at the above problems, mainly in research on the unstructured data ontology key technology, the automatic annotation issue on unstructured image semantic, and the establishment of unstructured data ontology model, all in order to achieve the efficient storage and retrieval of unstructured data.Firstly, for the problem of the semantic features acquire and auto-annotation from unstructured image data, we propose a method of the semantic on image auto-annotation, by using the mapping between the image low-level features and its scene semantics, the low-level features will be converted to the binary strings, thus achieving an method of semantic on image auto-annotation by the way of image retrieval, and the experimental results show that this method is effective.In the process of establishing unstructured data ontology model, starting from the data itself features, unstructured data can be broken into their own basic attributes, semantic attributes and feature attribute, and which constitute to describe the characteristics, which will transform the unstructured data management issues to the relatively structured data property management issue, and then by introducing a suitable successor tree indexing model, making for the management method of unstructured data, with support for massive data,real-time, and dynamically updated to meet the superiority complex search condition. Eventually by several complex retrieval examples, verify the availability of the method in the management of unstructured data.
Keywords/Search Tags:Big data, Unstructured data, Ontology data model, Image semantic auto-annotation
PDF Full Text Request
Related items