• by Laurie Sullivan  @lauriesullivan, March 3, 2015

What if searchers could tell fact from fiction on the Internet? A team of computer scientists at Google has proposed a way to rank search results. Rather than rank by authority or popularity,  scientists agree that the Internet could become a much more reliable place for information if it were ranked by factual accuracy. The move would change search engine optimization by increasing the importance of editorial content.

In a research paper published by Google, the scientists call the ranking factor a "trustworthiness score" computed by "Knowledge-Based Trust (KBT)," where data can reliably compute the true trustworthiness levels of the indexed sources. The researchers apply the score to a database of 2.8 billion facts extracted from the Web to estimate the trustworthiness of 119 million Web pages.

Rather than count incoming links, the system counts the number of incorrect facts within a page. The results computed for each page are known as the Knowledge-Based Trust score, determined through an algorithm that dynamically decides the level of truth for each source.  

The paper, highlighted by New Scientist, proposes a new metric for evaluating Web-source quality. The Google scientists propose a probabilistic model that jointly estimates the correctness of extractions and source data, and the trustworthiness of sources. Experimental results have shown both promise in evaluating Web source quality and improvement over existing techniques, per the research, which also calls attention to existing flaws in ranking factors.