How did the search results form?

February 28--New Scientist in the UK is looking at a new trend in Google's search: Google tries to determine the ranking of Web pages in search results based on credibility rather than number of links. Google is changing the way it used to measure the quality of Web pages by the number of imported links, and a new system that a team of researchers are working on will no longer compute the number of imported links, but compute the amount of false information in the Web page, and give the page a confidence score, This determines how the pages are positioned in search results.

The research team argues that the internet is littered with rubbish, and that some fake medical ads may make headlines, and that some fabrications of "news" are spreading like wildfire. That is, if the number of pages is linked to enough, even if flooded with a large amount of false information, will also appear in the search results in front of the position. In the new software designed by this team, the fewer error messages will be considered more believable.

How did the search results form?

On the search results of the shelling, the domestic Internet users are not unfamiliar, "bidding door" in a domestic search engine's head has always been difficult to swing. Planing but moral criticism, as long disseminated in the Internet, how do we get the results of the search on weekdays?

On such as Googlebot (commonly known as Google Crawler), such as the inclusion of search engines, scientific fan here not to discuss. Search engine results are good or bad, the industry has a word for the degree of equity: relevance (relevance), in simple terms, is the search results are not the user to query the content. If you want to query a phone number, a lot of people will usually call 114, in the process of questioning, the user would never want 114 to give them the empty number or wrong answer.

Search engine to do the job, is to respond quickly to the user's search, to meet the needs of users to feedback the results of the user. It is the search engine's core work to quickly find the answers the user needs to retrieve, or, more generally, to include the highest quality content that is most relevant to the user's search requirements.

At present, different search engines use different sorting methods. More is "the number of pages are linked more and more authoritative site links, the higher the quality of the ranking," the "Hyperlink Analysis" and "the frequency of the query word, the higher the order of its ranking on the more than the former" statistical method of frequency. It also includes a Pay-per-view method based on the number of clicks the page is clicked on and how much to pay for the site.

PageRank, a Google search engine, was created at Stanford University in 1998 by its two founder Larry Page and Sergey, a prominent representative of the hyper-link analysis method. PageRank the link to the page as a vote on the page and is a token of its importance.

How important are search results sorted?

In the search results in front of the position is very important, because this is directly related to the user's search engine experience problem, the user can be in the appropriate length of the search list to get their own answer, search engine when the user wants the results of cleverly presented in front of the user, all in the search results of the ranking

A study of user clicks for search engines shows that:

1 60-65% query clicked the top 10 search results page;

2 percent will consider clicking on a page that is ranked 11 through 20;

3 only sputter will click on the list of the 21st to 30th page in the search results.

In other words, most users are unwilling to turn the page to see the results of the search engine. Google's hot-click maps and a eye tracking experiment at Cornell University are good evidence of this.

The spectrum represents the click heat of users at different locations, the closer the color is to the red, the higher the intensity of the clicks.

From Google's Heat map, the top 3 of the search results are the most hot and attract a lot of clicks. In other words, for a search engine, the first few results are the most critical;

Cornell University's eye tracking experiment obtained more accurate user behavior analysis of Google search results. Analysis results show that the first three results of the total click to almost split the search flow of 80%, the first result obtained 56.38% of the search traffic, the second and third results ranked lower, but far lower than the first results, the top three results of the click ratio of about 11:3:2.

Credit rating: can also search engine a blue sky?

With regard to search quality, there is its own evaluation system. The Cranfield evaluation system, which originated in Cranfield University (Cranfield University), is widely used in all major search engine companies, as well as precision-recall (accuracy-recall rate) evaluation index, MAP (average accuracy rate method), A /b testing, etc. search engine results Quantitative evaluation method of quality. At the same time, there have been many applications that focus on search results, such as Lazytruth plug-ins that can weed out false information, collect rumors from spam sites, and emergent projects that are verified or disproved by cross-referencing other sources. Poor their assessment systems and applications are focusing on the same problem: providing users with better search results.

But objectively speaking, this Google claims to be eager to "facts" to determine the Web page in search results rankings, but how many "facts" can be crawler (Googlebot) fast and high-quality recognition? This fact is feared to be depressing, but just as powerful as Google, cannot change the "fact". At the same time don't forget that sorting is both a scepter of search engines and a talk of search engines asking for their food and clothing parents-advertisers. Credit rating, more like Google to the user to pass a signal: As a search engine, I have been committed to the best search results present to you! Even the search results, such as the Dogskin sort, are more likely to be problematic for our sorting methods.

The higher quality of the search results presented in front of users, is the search engine of the body and the necessary. Pay more attention to the authenticity of the past data, to do a rating, the credibility of the Web page "credibility value" to measure. At this point, Google is still very admirable and expected.