Font Size: a A A

Research And Implementation Of Concept Network Techniques Aiming To Filtering Bad Internet Context

Posted on:2008-03-16Degree:MasterType:Thesis
Country:ChinaCandidate:K FangFull Text:PDF
GTID:2178360242476827Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
The massive open Internet brings into our lives bad information as well as the information we need. Due to the openess of the Internet, the information it carries cannot be censored before released as the traditional media, and because of its much faster spread speed than the traditional media, bad information such as erotica ,violence, and political assaults from western countries are exposed to our people. This brings all kinds of negative influence and greatly endangers our society. Hence, it is extre- -mely necessary to develop a way to behold and control the information on the Internet.Context is the main kind of information on the Internet. Research on the Internet bad information control nowadays are usually based on key word filtering or statistics. The advantage of these kinds of methods are easy to implement and calculates fast. Yet mere key word filtering lacks necessary semantic analysis, it approaches the context meaning only in the level of word, not the whole sentence. Two context which contain the same word may in fact entirely differs from each other in the meaning it presents. Therefore context information filtering, especially biased context, need further researching.The main purpose of this essay is to establish a system structure which analyses and filters the bad information on the Internet on the level of semantic analysis. The concept network uses abstract "concept" as its basic nodes instead of words. Concepts are better in this situation than words in that its abstractness evades misunderstanding. Our first step is to extract characteristic sentences from the context which most represent the meaning of the context. Then, we divide the sentences into exact words and take out the "main" words and abstract them into "concept". The concept from a single sentence can be put together and form a concept set, which represents the whole sentence. When sets from the same kind of context are put together, we call it a concept network.This essay first introduces the security problem of the Internet and the latest research results on bad information control. Then after analyz- -ing the advantage and insufficiency the these results, we put forward the advantage of using concept network on network control. Next, we introduce two existing concept dictionaries, one of which is crucial to establishing the concept network. The next part of this essay is to introduce the outstanding research results which are of crucial significance to our building the concept network, related context processing methods and weight calculation are also discussed. Finally we describe the model of the concept network and run some tests to prove its effectiveness.
Keywords/Search Tags:concept, concept network, semantic analysis, similarity calculation, information filtering, case grammar
PDF Full Text Request
Related items