What is relevancy

When you enter a query on a search engine page or in the search box (at the top of the screen), the browser starts receiving events that are hidden from ordinary users. How, then, is the information in the vast World Wide Web, and what is search relevance? The very word originated from the Latin "relevo" and is translated into Russian as "raise" or "facilitate." The English equivalent of the term is "relevance". In clause 3.5.1 of GOST 7.73-96 this concept is deciphered, and "relevance" and "relevant" have the same definition, which is in accordance with the information received by the user's information request.

This state standard also contains other useful standardized terms and definitions relating to the field of information retrieval and its dissemination by means of information automated systems. The document helps to understand the meaning of many words in information, publishing or library activities, and not just such a unit of language as "relevance". The definition of the terms "per- tentness" and "continental" is also interpreted in clause 3.5.2 of GOST 7.73-96. They should be interpreted as the consistency of the information information data received. Often this concept is confused with the relevance, which is fundamentally not true, since in one case it is a question of matching the query, and in the other case it is about the correspondence of the need. It is also wrong to answer the question: "what is relevancy?" To answer that this term is close to the notion of "adequacy", that is, the practical applicability of the result.

But how is information? The search engine uses an army of bots - stand-alone programs that visit websites and store information about them. The name is derived from the English word "bot", that is, abbreviated from "robot". These programs find new sites, and then go to the main page and browse its contents. Bots learn pages related to home (in the browser it is loaded first) internal links. Also, bots look at links that point to other sites, that is, external links. By storing this information, they go over them, discover all the new sites and learn how the web resources are linked. Thus, they tirelessly study the "map" of the World Wide Web. But what is relevance? What does it have to do with bots?

On a concrete example, let's see what the search engines Yandex and Google offer. To the question "what is relevance?" In the issuance of Yandex, there were 28 million answers, and Google's 155,000. Here you can say that they all meet the request, and the needs are not always, that is, not all of them will be permanent. For example, pages where the Wikipedia CEO is given an explanation of this concept will be undoubtedly useful, and the site about relevant and irrelevant costs in decision making is not relevant to this topic. A more clear example is that when Yandex issued 31 million answers to a request for "hands up", which meant getting information about a musical group, and among them there is an article calling for hands to be raised, as this, according to the physiotherapists, is very useful. This answer can not be considered as a perpetual one, although out of 31 million answers, he takes the 586th position in extradition.

Thus, the higher the relevance of the text, the more likely it is that it will enter the first positions of the search answers. This indicator has a quantitative evaluation, but the algorithms for its implementation in different search engines differ and are unknown to users, and they change with time, because the search engines, trying to rise in the ratings, try to provide the pertinent information, shortening the time for its search. Highly relevant are pages where the occurrence of a keyword that coincides with a query can be in the range of 3 to 7%. But this is not the only criterion of the relevance of the document. These include a number of internal and external indicators. Internal indicators (except for the density of keywords) are their position in the text (entering the title, their proximity to the title and each other, with the formation of probable search phrases), as well as the presence of synonyms, confirming the correspondence of the content of the text to the query. The main external indicator is considered to be popularity in the global network, which is indicated by PageRank from Google, TIC from Yandex and Tor100 from the Rambler system. That is, how often the page is transferred from other sites, it is better if they are with high ratings. That's what relevance is.