For a long time, Google has kept the formulas for its search algorithms in strict confidence. The reasons are clear: first, to protect against competitors; secondly, that this information could not be used by “black” optimizers for their own purposes. And then, finally, the policy of secrecy begins to change. The head of the Search Quality division, Udi Manber, posted a
message on the corporate blog, which he calls the first step to discover previously secret information.
In the first post, Udi Manber talks about the basics of the Google search engine ranking system. He explains that the well-known PageRank algorithm is part of a much larger system that includes language models (processing phrases, synonyms, typos, etc.), query processing models (this is not linguistics, but an attempt to understand the essence of search queries) , temporary models (for some queries it is better to issue the most recent pages indexed in the last 30 minutes, and for others - old pages with proven quality), as well as personalization models.
In the first message of Manbera, we do not see anything “secret” that we would not know before. But this is only the beginning, and in the future, Manber may publish new interesting information, so that it makes sense to closely follow the Google corporate blog.
')
Independent experts
comment with irony that Google has always been a company no less closed than Microsoft. Even worse, more hypocritical, because the system of Google’s work is this: they force people to open information for free access, collect it into their own closed “black box”, with the help of which they earn money. It is clear that such actions cause some dissatisfaction with the authors of the content, so the new strategy of Google to open its algorithms is, to some extent, a forced step.