Font Size: a A A

Research On Location-based Information Association Retrieval Technology

Posted on:2020-03-20Degree:MasterType:Thesis
Country:ChinaCandidate:M Y YanFull Text:PDF
GTID:2518306548993629Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
The development of the Internet has brought about the explosive growth of information and diversification of expressions,making information transmission more efficient,convenient and fast.The endless stream of digital intelligent information technology is also changing people's lives.Internet news is one of the most frequently searched content by users because of its timeliness and rich information.In the face of massive news,people also encountered some unavoidable problems,that is,due to the excessive expansion of information,it is difficult for users to find the content they need,which also promotes the gradual development of text visualization and related information retrieval technologies.How to analyze and mine various information in the Internet environment,form the interconnection of information,and provide efficient search service of knowledge chain,and gradually become one of the effective means for users to understand complex text content.At the same time,the user's attention to spatial information in the text is continuously improved.Incorporating text retrieval into a certain space or scene framework,fully demonstrating the characteristics of text in the geospatial dimension,will further satisfy the user's need for a spatial location-aware retrieval method.This thesis focuses on geospatial-based news analysis and correlation methods,builds a correlation model for news elements,refines the aggregation granularity of existing information correlation models,and further proposes a systematic geocoding method for geographical elements.The core location information was stripped from the news text,and a location-based association retrieval method was proposed based on the element network to expand the information retrieval mode.The main contributions of this thesis include:1.News element association model.This thesis improves the existing chapter-level information association model.Based on the meaning of the event,a feature-centric news association model is defined.The model is dynamically updated in real time with the news text,and has strong scalability.Experiments show that the model can achieve efficient and effective retrieval of news through the element level and expand the traditional information retrieval model.At the end of the thesis,it explains in detail how to apply the above method to the "location-based information association retrieval" system,introduces the prototype system architecture and the design and implementation of each module,and proves the feasibility of the theory and method of this thesis.An example of an associated retrieval application for various information is provided.2.Geographical mapping of news text.Aiming at the characteristics of geographic information contained in text,this thesis proposes a three-stage text space-to-geospatial mapping method,which includes place name entity recognition,place name disambiguation based on hierarchical tree,and multi-level location focus to blur and uncertain qualitative characteristics The spatial information is converted into structured coordinate data.Experiments show that this method can effectively extract the core position of news,and thus incorporate text information into the spatial reference frame.3.Relevant retrieval of news information based on geographic location.This thesis makes full use of the geographic information contained in the news and based on the news element association model,and proposes a geographic community mining method that combines spatial and semantic features to form multiple spatial structures that consider spatial distance and semantic impact.Aiming at user retrieval,in the formed geographic community,the relevance ranking of geographic entities based on path search is studied.Experiments show that the method can effectively implement location-based news information retrieval and recommendation,and increase the regional characteristics of news retrieval.
Keywords/Search Tags:Web news, element association model, text geocoding, association search
PDF Full Text Request
Related items