Research On The Organization Of Text User Generated Content Based On Linked Data

With the development of the Internet,the amount of text user generated content on the Internet has exploded.These messages are intricated linked.By organizing them effectively,we can find a wealth of knowledge.However,these messages are highly arbitrary and complex.It is difficult to organize them effectively by using traditional information organization methods.As a lightweight semantic implementation technology,linked data has advantages in many aspects such as machine readable,semantic association,network data sharing,interoperability and so on.In view of its advantages,the paper proposes a linked data mashup system for text user generated content.The system includes four layers:data layer,query layer,integration layer and application layer.And the paper carries out a case study and the data about movies is grabed from DBPedia,LinkedMDB,GeoNames and Douban.We provide a new way to organize the text user generated content.The main research work is as follows:(1)The paper proposes a linked data mashup system for text user generated content,which consist of four layers:data layer,query layer,integration layer,and application layer.We aim to achieve organizating the text user generated content effectively,and enrich related knowledge by using named entity recognition,language conversion,linked datasets query,datasets integration and mashup,visual presentation and so on.(2)The solution to the key problem is provided in the process of system construction.There are many key issues in the model construction process,such as named entity recognition,associated datasets query,datasets integration and mashup,and visual presentation.The paper uses existing natural language processing tools to solve the question of common named entity recognition.For specific types of named entity recognition,the paper uses Apache's OpenNLP open source framework to train special types of named entity recognition models.The local datasets is associated with external datasets to realize the linked datasets query and mashup.And we select the D3.js visualization technology to realize the visual presentation.(3)The paper processes the system by using film review information on Douban,and implements linked data mashup system of text user generated content by using Java language.The movie information and its review on Douban is grabed by using data collection tools.Meanwhile,the dataset is associated with DBpedia datasets.LinkedMDB datasets,and GeoNames datasets.The results show that the proposed system can solve the problem of organizing the user generated content effectively by providing open linked datasets.At the same time,it can enable users to get plenty of external links,and expand related knowledge.