Information credibility analysis of web content

ABSTRACT

General users write daily news about themselves and post information they consider interesting as digital documents for blogs and SNS. Such digital content includes both valuable information as well as worthless, false, and demagogic information. Ordinary web search engines can display web pages in a particular order. The ranking method evaluates the score of web content and generates a ranked list. The top-ranked web content on search engines is often relevant to the user's query, though, in some cases, the content may not be credible or valuable. Nevertheless, readers often trust the authenticity of the displayed information. Even if users believe that the content is useful, the search engine cannot evaluate the retrieved digital content, and users have to retrieve a variety of content using different keywords. The need for an information analysis technology that helps find credible and valuable information from large amounts of Web content is progressively growing.

In Japan, the NICT (National Institute of Information and Communications Technology) initiated the ''Information Credibility Criteria Project'' in 2oo6, and the MIC (Ministry of Internal Affairs and Communications), too, initiated the ''Research and Development of Information Credibility Verification Technology for Telecommunication Service'' in 2007.

The NICT's project addresses the issue of information credibility by analyzing credibility based on the following criteria: (1) content, (2) sender, (3) appearance, and (4) authenticity of content. We believe that the understanding of texts by a machine is important and that an NLP (Natural Language Processing) approach is very effective in evaluating the credibility criteria. The MIC's project aims to develop methods to analyze not only text information but also multimedia content using NLP, information retrieval and data mining approaches. By using different methods for analyzing the information credibility criteria, credible information can be acquired, which eventually becomes valuable knowledge.

This talk will throw light on the activities of both projects in Japan.

INDEX TERMS

Primary Classification:
  H. Information Systems
  H.3 INFORMATION STORAGE AND RETRIEVAL
      H.3.3 Information Search and Retrieval

General Terms:
Algorithms

Keywords:
information credibility, information retrieval, nlp