Research Of Reliability On Multi-Representations Military Information

With the rapid growth of the internet, Web2.0 technology brought a brand new model for the internet. There were amount of social software applications on the internet which was open and interactive such as Blogs, BBS, TAG, SNS, RSS, WIKI and so on. Each Internet-user had their own blogs, WIKI, social tag and Podcast to contact with each other by software. Thanks to these applications being widely used, the Internet became more and more diversiform and complicated. On one hand the Internet made us feel convenient and happy but on other hand it brought us lots of troubles such as the reliability of the information on the Internet.Sun Tzu said:The art of war is of vital importance to the State. It is a matter of life and death, a road either to safety or to ruin. Hence it is a subject of inquiry which can on no account be neglected. The Internet military information was of the art of war which was paid more attention by each country's military researchers. How to discriminate the reliable information from the mass was a big question for those researchers. The opening of the Internet not only brought convenience but it also cost us plenty of time to get the reliable information. Thus, how to get reliable military information became the primary research in military.This paper proposed a multi-representations method to acquire reliable military information by considering the characteristics of the internet military information.First, we structurally stored the military-related information crawled from the internet. And then we got the themes which was clustered by LDA model and then got reliability each document of the same theme considering the information source, information content and the reviews. Among of the three considerations, information source consisted of website, web page and information publisher. The reliability of information source was based on the HITS and link-reachability algorithm. The text and military features was applied on acquiring reliability of information content. We analyzed the sentiment of the reviewers and applied improved probability model to got the got the reliability of reviews. At last, we ranked them by their reliability in the same theme according to learning to rank. The experimental results showed that the proposed method did military information acquiring and discriminating good to satisfy the demand of military researchers. It also showed that multi-representations method performance better than single-representation method.