Font Size: a A A

Research And Application On On-line News Topic Detection In Food Security

Posted on:2016-03-11Degree:MasterType:Thesis
Country:ChinaCandidate:L TangFull Text:PDF
GTID:2308330479482173Subject:Software engineering
Abstract/Summary:PDF Full Text Request
In recent years, incidents on food security one after another, and has great impact on people’s health and their quality of life and the long-term development of the economy. With the rapid development of the Internet, the mass of information is growing explosively. It is important to quickly and accurately integrate the information about food security on the Internet in order to gain useful information and discover food security news topics, which can help food regulatory authorities and the related companies to learn the problems in the first time and react, and it has practical significance on strengthening our food security supervision.In this paper, we apply Topic Detection and Tracking(TDT) on food security news, and use On-line Topic Detection(OTD) technology to discover food security news topics on the Internet. The main tasks of this paper are as follows:(1) In the preprocessing of OTD, based on the characteristics of food security news, this paper proposes an improved preprocessing strategy on good security news. Through identifying the main elements of news and name entities of food security field and the key words appearing in the title of the news, using the improve TF-IDF algorithm to perform feature weighting, can improve the quality of preprocessing.(2) In the text clustering of OTD, by analyzing the classic Single-Pass clustering algorithm, this paper put forward an improved Single-Pass clustering algorithm by preprocessing a certain number or a certain period of time of text to lighten the disadvantage of dealing one text for one time. Furthermore, this paper introduces the concept of seeds documents, which better present a topic model and lighten topic center drift. Finally, the improvement will be verified by the experiments.(3) This paper designs and implements a food security news online topic detection system. The system consists of analysis subsystem and presentation platform. Analysis subsystem is topic detection module to discover new topic and gather follow-up news. Presentation platform contains five modules which are Latest News, Hot Topics, Special Report, Public Opinion Bulletin, Bulletin Management and User Management.In summary, based on the incremental clustering algorithm this paper studies the technology of OTD on food security news, and applies it into the design and implementation of the food security news online topic detection system, finally provides a visual analysis and information retrieval service, which has realistic value.
Keywords/Search Tags:Food Security, news public sentiment, Topic Detection, Topic Model, Incremental Clustering
PDF Full Text Request
Related items