Google developer on what people are looking for on the Internet
Every day we look for Answers to questions about how to become happy, or why the water under the ice does not freeze. Not the last role in the formation of our approach to the data played by search engines, including Google, which in Russia for eight years. Wonderzine talked with developer Anna Kondratieva, who is now working in the search division at the Moscow office of Google. She told how the language we speak with search engines changed over the decade, why you shouldn’t be afraid of contextual advertising, and why people use voice search only in solitude.
FROM A MOMENT, LIKE GOOGLE WENT TO RUSSIA, EIGHT YEARS. YOU CHECK OUT, HOW AND ALL SEARCHING SYSTEMS, INDEX THE CONTENT OF THE INTERNET. AND YOU have a DECISION TO HAVE SOME DATA, HOW YOU CHANGED REQUESTS FROM THE MOMENT OF A GOOGLE WEEK AND IN OUR TIME. WHAT NOW LOOKING FOR MORE THAN ALL?
First of all, not only search queries have changed - search engines have changed quite a lot. These changes are always interrelated: some functions appear in a search engine, people understand that they are there, and they start to use it. Sometimes changes come from the other side, that is, at the request of the user. This system is constantly evolving. At the dawn of the development of search engines, everything was very simple: you entered the words you wanted to see, and you found documents that contain these words. Now the search engines are much more complicated: we have maps, we have video, we have direct answers to questions. We try to understand exactly what the user asked.
At the end of 2012, Google launched a fundamentally new service in Russia, the Knowledge Graph, in Russian it is called the Knowledge Network. This is a database, but not an index of keywords and addresses, but a set of articles linked by semantic links, a graph of facts and links between them. Thanks to this structure, Google is not just trying to find information that literally matches the request, but is able to directly answer the user's question. If, for example, a user asks about an actor, we try to give as much information as possible about that actor - to tell what he is famous for, what is his marital status, what his films are most popular with, what actors he was shooting. That is, it is a much more complex system than the simple correspondence of one word to another.
And users, in turn, begin to communicate more freely with the search engine — they start asking questions as if they were asking their friend, who knows a lot. They do not speak as before ...
BUY SAUSAGE.
Buy sausage is also a very difficult request! When you say “buy sausage”, the system should understand that “sausage” is a product, and that by saying “buy” you mean some kind of action and, most likely, you want to see some store. It is unlikely that you want to buy sausage on the Internet, most likely you want to know where there is a grocery store nearby, which is open right now. That is, it is not as easy as it might seem.
IS THAT YOU ARE FITTING UP TO USER REQUESTS OR EVERYTHING HAPPENS HERE? FOR EXAMPLE, IF I WANT TO SEE HOW TO LOOK AT RYAN'S Daughter, Gosling and Yeva Mendez, it is enough for me to enter their name in the search engine.
That's right, and if you enter them, then in the drop-down tips you will see that a huge number of people are interested in this.
BEFORE, AT THE ZARE OF DEVELOPMENT OF THE PURCHASERS OR, FOR EXAMPLE, DURING THE EXISTENCE OF THE RA CLEAR AREA AND ITS BEING ALL THE SEARCH ENGINE OUTCOOK AREA REQUIRED TO FULLY COMPLETE WITH ALL MY REQUEST: RYAN AGRI CURRENT OUTPUT
And with great probability before you would see only the Wikipedia page, and not the news of the birth of a child. Although she was not there, so it would be some IMDB. That is, now we are adjusting to users, trying more and more to understand their intentions. Do not just give out the simplest correspondences, but understand what a person wants to see, and solve the problems that arise for him. At the same time, the reverse process is also true. If people understand that we have maps and you can search all the cafes in the area on these maps, they start doing it more often. This is a situation where supply and demand find each other.
There is a joke about the fact that you can climb on YouTube to see how to cook soup, and after half an hour to be in the universe of Stephen Hawking
Now in the media environment, there is a trend such as explanatory journalism. In fact, journalists take away from Google its function. If you now want to understand a difficult situation, it’s enough to read one article and not ask 150 queries to Google. Are you not afraid that due to this trend, over time, search engines will turn into just a collection of everyday knowledge? That is, after ten years, people will come to Google only to buy something and find something specific.
Any system that finds the answer to a question is to some extent a search engine. The opposite is the situation when you have no question, and you are offered information, believing that it may be of interest to you. For example, the magazine itself gives us knowledge, we do not initiate this process. The journalist offers information without knowing whether the person needs it or not. The search is the opposite - there the user is the active side, he begins to initiate this process with his request. This is a very personal and very multifunctional system. Because even one person has a huge range of requests for a small amount of time.
There is a joke about the fact that you can climb on YouTube to see how to cook borscht, and after half an hour to be in the universe of Stephen Hawking and watch a video about how black holes are made. Each person has a set of things that he is interested in, just huge - like a whole nation and all of humanity. Accordingly, journalism, which offers information for a certain group of users, cannot fully satisfy the interests of people from other groups. The search query provides instant information on what interests you right now, at this second.
We have already lost the habit and do not remember how it was when there was no search engine or Internet. We do not remember this moment: when we wanted to find out who was the wife of a famous person or historical character, we had to go to the library, look for this book - it was a very difficult and very long process. Now we do it so fast that we don’t even think about what technology is behind it. So it has become commonplace that we do not understand how important it really is, how complementary our world is, how much it speeds up our ability to learn and receive information.
How many times a day does a person google on average?
Google processes an average of 40 thousand requests per second. Of course, we can’t say how many times a person turns to Google.
Are there any differences between male and female Google queries?
If you look closely, especially at several requests in a row, then you can probably guess whose requests were. If they ask how to cook a cake, then, most likely, it’s still a woman, and if they ask how to cook sausages, then, most likely, a man. But this, of course, is a very superficial assessment, in fact, we do not know who is behind this or that request. We are very careful about the user's personal data. For example, neither I, nor any of the engineers can find out what kind of person I was looking for - it is absolutely impossible.
Recently there was a series of articles that Google knows about you that you do not know about yourself. Honestly, I have the same impression sometimes. When you analyze my search queries, and then on the basis of this you give me contextual advertising, I think: damn, Big Brother is watching me.
When users provide us with data, for example, by registering a Google account, we use it to improve the services and make them even more convenient. With this data, we show you more relevant search results and advertising and provide the ability to easily share information. At the same time, we always make sure that users understand how data is used, and how to protect their privacy. In particular, each user can manage certain types of information in a Google account using the “My Account”, customize the appearance of their Google profile using an editor, control other people's access to data, and so on.
Advertising is a completely separate system that works without human intervention. That is, all the information that it processes, it does not show anyone. Therefore, as search engineers, we have no access to this data. Everything that happens in Vegas stays in Vegas. Accordingly, everything that happens in advertising remains solely in advertising. There is no way that would allow at least some information to be extracted from there. Advertising can guess that you are a woman. But no one can use this information in any way. This is a fundamental position of the company, because it is bad for us and bad for users. In addition, the user can change the settings of their advertising preferences, including interest categories, as well as refuse certain Google advertising services.
As I understand it, you are directly engaged in voice search. I have never seen anyone use a voice search, never at all. Is that when people are driving and they urgently need to know the address.
The fact is that when someone looks at you, it is less natural to talk to the phone. In public places, people prefer to keep their privacy and no longer communicate with the phone through a voice. But there are situations where it is absolutely necessary. As an example - just which you brought when driving a car. I do not drive and I don’t know how we do with it, but in the USA, for example, this is a violation of the law if a person activates his phone while driving. If a person needs something while driving a car, then the only way is to use voice search, there is simply no other way. And this is right, because there are situations when you just can not hold the phone in your hands. Plus, there are situations when you need to get something quickly, because typing is more difficult and slower than asking with a voice. There are people with disabilities, there are those who just find it difficult to get into these small keys.
Still, voice search is a relatively new feature that appeared not so long ago, users continue to open it, they continue to explore how it works. They continue to find situations in which this would be more convenient. Because only 15 years, users are looking for something on the Internet, filling text in search windows, and over the 15 years they have developed a habit. In situations where it would be more convenient to ask the phone right this way, they do not, because the power of the habit outweighs. It takes some time to adapt, to understand how to fit new scenarios into the usual ones. But, of course, there are more voice calls, people learn to use it in a variety of situations.
If you are asked how to cook a cake, then, most likely, it is a woman. And if they ask how to cook sausages, then, most likely, a man
You study user interaction with a search engine. It seems understandable, but if in more detail - are you learning what the user wants to see in a search engine?
Yes. In fact, I study what he wants. Even not necessarily what he wants to see, but what kind of reaction the user expects from the search engine. Because the farther, the more natural the language of interaction becomes. For example, one thing that, when you go on a journey, ask how many degrees are in Milan, and then ask "what time is it", and the search engine will understand that you are interested in time in Milan - this is reminiscent of how one person talking to another person. This is a more natural speech, which makes us less distracted by the construction of the phrase.
That is, the more natural speech, the more natural answers in natural language we are able to provide, the more we make this process of human interaction with the knowledge that is contained on the Internet smooth and transparent. A person does not need to think which words to choose, which one to pick up the request. That is, we make it so that a person communicates with a search engine as he communicates with a friend or childhood friend. Accordingly, I am engaged in understanding what people are asking and responding to them so that it does not sound clumsy.
Do you collaborate with linguists to study how language changes?
Undoubtedly, linguists are needed. But the fact is that any person who speaks his native language understands when he is spoken with in the correct language. Even if someone makes a mistake and builds the phrase incorrectly, we always understand what is literate Russian. Therefore, of course, we have our own linguists, we have people with computer linguistic experience. Of course, we have a pile of books that can tell you how to build a phrase in the most difficult cases when we, like educated people, find ourselves at a dead end.
From the point of view of linguistics, since the arrival of Google in Russia, how have human needs changed, what is the current trend?
The language has become more understandable, more verbal, so to speak. Requests have become conversational. When in 2004-2005 all operating systems were in their childhood period, the user tried to formulate queries to search engines in the bird's language, in a search language, not in the natural language. They are accustomed to using words in the nominative case, not to harmonize phrases, that is, in fact, simply to name a set of words that they would like to see in the documents. If earlier you would say "Moscow weather forecast", now you can say "Moscow weather forecast", "what is the weather in Moscow?", And this sounds like a normal question. Secondly, the search for this time has gained additional services. There were videos, maps, there was a search for pictures, people are beginning to look for addresses. In the end, now a person may ask, "Why does the water not freeze under the ice?" and get an answer.
Getty Images launched the project Lean In - this is such a library of pictures depicting women of various professions. his goal - destroy stereotypes about women. Does Google have similar projects?
In fact, Google is doing a lot for girls to come to IT and build a successful career in this area. For example, Google founded a scholarship named Anita Borg, which is awarded annually to the most powerful girls - students in the field of computer science. And in March of this year, Google launched the Women Techmakers series of events — more than 100 meetings in 52 countries around the world, during which girls could attend career planning workshops and share experiences with leading female experts in the technology industry. And in our company there are quite a lot of female engineers who hold high positions.
How has your life changed personally while working at Google?
First of all, I got used to a comfortable work environment. It includes the people I work with, the projects we do and how this work is organized; availability of a huge amount of information and computing power, and just living conditions in the office. In some ways, the local atmosphere is similar to the atmosphere of a student campus with its minimum hierarchy, with many plans and ideas, and with a wide variety of knowledge and interests.