The future of Internet search. Knowledge network

December 12, 2012 22:07

Google's service Knowledge Graph («Knowledge Network") is now available in Russian. This tool is more like automated encyclopedia — in the search results are given blocks with brief information about the objects that are somehow related to the query.

"We have a large base of knowledge that surrounds us in the world. These are some concepts, people, places, architecture, world, dog breeds. In general, anything that can be described. All these concepts have a number of characteristics. In addition, we know some ties that are between these objects, for example, that a certain actor starred in films, and this breed is related to dogs. And there are so many sites now, there are more than 500 million, and among billions of connections "- is trying to explain the concept of service representative of the Moscow Google Search Quality team Rinat Safin.

For example, at the request of "Peter", this "knowledge network" shows years of life of the king, the names of his parents, the main events in his life, and — very important — the connection of this historical figure with other events and phenomena. Under the short squeeze in Russian is svetloserenkaya, almost invisible link to the source — in this case, "Wikipedia." Interestingly, in the Russian-speaking unit link is to the English-language article.

"The data come from various sources, some of them — it's open source, the license which allows the use of their data, it is among the" Wikipedia ", which is a very important, useful and important source. This FreeBase, we bought several years ago and who, in fact, contains structural information about the objects from the real world. These are sources that contain data about the statistics, for example, is a source of SIA FAQ Book — this is some CIA base, open at all, about the population of the different regions. This World Bank data on the population and the economy in different regions. Information is gathered from various sources, if the sources are open. If it is any secret information, but we know that the users need it, we try to buy it and add to the Knowledge Graph », — says Rinat Safin.

It seems that from the "Wikipedia" in block 70 percent of the information, and I think, Google has complied with all the requirements of the license "Wikipedia" citation. By the way, one would assume that because of the "knowledge network" from "Wikipedia" fall attendance. However, the service was launched in the U.S. in May of this year and no recessions in the audience has not seen.

It should be noted that, on the idea of Google, this tool should be a transition to a semantic search Google. However, the language — the substance multifaceted, and, entering a search query string "Venus", the user may have in mind is not the second planet from the sun, and the ancient Roman goddess of fertility and the famous ancient Greek sculpture. The user can also search for beauty salons and restaurants with the same name. Knowledge Graph essence is to ensure that the search has learned to understand the essence of the real world in their relationship, but not primitive, "read a line of search queries."

It is important to note that Google is also working on authenticity issued by the facts — that is, create a database of reliable sources, as well as the internal mechanism of a comparison or matching information. In addition, the development of the service is difficult and specific features for specific languages: connections between words and objects in English may mean nothing in Russian.

Like this post? Please share to your friends: