Gartner's schedule is like a haute couture show for those in tech. By looking at it, you can find out in advance which words are the most hyped this season and what you will hear at all the upcoming conferences.
We've deciphered what's behind the beautiful words in this graph so you can speak the language too.
To begin with, just a few words, what kind of chart is this. Every year in August, the Gartner consulting agency releases a report - Gartner Hype Curve. In Russian, this is the “hype curve”, or, more simply, hype. 30 years ago, rappers from the group Public Enemy sang: "Don't believe the hype." Believe it or not, it's up to you, but it's worth at least knowing these keywords if you're in tech and want to know the world's trends.
This is a graph of public expectations from a particular technology. According to Gartner, in the ideal case, technology goes through 5 successive stages: technology launch, high expectations peak, disappointment valley, enlightenment slope, productivity plateau. But it also happens that it sinks into the “valley of disappointment” - you can remember examples yourself very easily, take the same bitcoins: initially hitting the peak as “money of the future”, they quickly rolled down when the shortcomings of the technology became obvious, first of all restrictions on the number of transactions and the crazy amount of electricity required to generate bitcoins (which already entails environmental problems). And of course, we must not forget that the Gartner chart is just a forecast: here, for example, you can read a detailed , where the most striking unfulfilled predictions are sorted out.
So, let's go over the new Gartner chart. Technologies are divided into 5 large thematic groups:
- Advanced AI and Analytics (Advanced AI and Analytics)
- Postclassical Compute and Comms
- Sensory and mobility (Sensing and Mobility)
- Augmented Human
- Digital Ecosystems
1. Advanced AI and Analytics (Advanced AI and Analytics)
Over the past 10 years, we have seen the high point of deep learning (Deep Learning). These networks are truly effective for their range of tasks. In 2018, Jan LeCun, Geoffrey Hinton and Joshua Bengio received the Turing Award for discoveries in them - the most prestigious award, analogous to the Nobel Prize in computer science. So, the main trends in this area, which are plotted on the chart:
1.1. Transfer Learning
You do not train a neural network from scratch, but take an already trained one and assign it a different goal. Sometimes this requires retraining part of the network, but not the entire network, which is much faster. For example, taking a ready-made ResNet50 neural network trained on the ImageNet1000 dataset, you will get an algorithm that can classify a lot of different objects from an image at a very deep level (1000 classes based on features generated by 50 layers of the neural network). But you don't have to train the entire network, which would take months.
В Samsung "Neural networks and computer vision", for example, in the final with the classification of plates into clean and dirty, an approach is demonstrated that in 5 minutes gives you a deep neural network that can distinguish dirty from clean plates, built according to the above architecture. The original network did not know what plates were at all, it only learned to distinguish birds from dogs (see ImageNet).
Source: Samsung "Neural Networks and Computer Vision"
For Transfer Learning, you need to know what approaches work, what are the ready-made basic architectures. In general, this greatly accelerates the emergence of practical applications of machine learning.
1.2. Generative Adversarial Networks (GAN)
This is for those cases when it is very difficult for us to formulate the goal of learning. The closer the task is to real life, the clearer it is for us (“bring a bedside table”), but the more difficult it is to formulate it as a technical task. GAN is just an attempt to save us from this problem.
Two networks work here: one is a generator (Generative), the other is a discriminator (Adversarial). One network learns to do useful work (classify pictures, recognize sounds, draw cartoons). And the other network is learning to learn that network: it has real examples, and it learns to find a previously unknown complex formula for comparing the generations of the generative part of the network with real-world objects (training set) according to really important deep features: the number of eyes, proximity to the Miyazaki style, correct pronunciation of English.
An example of the result of a network for generating anime characters.
But there, of course, it is difficult to build architecture. It's not enough just to throw neurons, they need to be cooked. And it takes weeks to learn. The topic of GAN is being studied by my colleagues at the Samsung AI Center, and it is one of their key research questions. For example, like this : Using generative networks to synthesize realistic photographs of people with a changeable posture - for example, to create a virtual fitting room, or for facial synthesis, which can reduce the amount of information that needs to be stored or transmitted to ensure high-quality video communications, broadcasting or personal data protection.
1.3. Explainable AI
In some rare tasks, advances in deep architectures have suddenly brought the capabilities of deep neural networks closer to those of humans. Now the battle is on to increase the range of such tasks. For example, a robotic vacuum cleaner could easily tell a cat from a dog in a head-on encounter. But in most life situations, he will be unable to find a cat sleeping among linen or furniture (however, like us, in most cases ...).
What is the reason for the success of deep neural networks? They develop a task representation based not on “visible to the naked eye” information (photo pixels, sound volume jumps…), but on features obtained after preprocessing this information by several hundred layers of a neural network. Unfortunately, these relationships can also be meaningless, inconsistent, or bear traces of imperfections in the original data set. For example, there is a small computer game about what the thoughtless use of AI in recruiting can lead to. .
The system for labeling images called the person who cooks a woman, although the picture is actually a man () it at the Virginia Institute.
To analyze complex and deep relationships that we often cannot formulate ourselves, Explainable AI methods are needed. They organize the features of deep neural networks so that after training, we can analyze the internal representation learned by the network, and not just rely on its solution.
1.4. Edge Analytics / AI (Edge Analytics / AI)
Everything where there is the word Edge means literally the following: transferring part of the algorithms from the cloud/server to the level of the end device/gateway. Such an algorithm will work faster and will not need to be connected to a central server in order to work. If you are familiar with the “thin client” abstraction, then here we thicken this client a bit.
This may be important for the Internet of Things. For example, if the machine is overheated and needs to be cooled down, it makes sense to signal this immediately, at the plant level, without waiting for the data to get into the cloud and from there to the shift foreman. Or another example: self-driving cars can deal with traffic conditions on their own, without contacting a central server.
Or another example of why this is important from a security point of view: when you type texts on your phone, it remembers typical words for you, so that the phone keyboard will conveniently suggest them to you later - this is called predictive text input. Sending everything you type on the keyboard to somewhere in the data center would be a violation of your privacy and simply unsafe. Therefore, keyboard learning occurs only within the framework of your device itself.
1.5. AI platform as a service (AI PaaS)
PaaS - Platform-as-a-Service is a business model where we get access to an integrated platform, including its cloud storage and out-of-the-box procedures. Thus, we can free ourselves from infrastructural tasks, and fully concentrate on the production of something useful. An example of PaaS platforms for AI tasks: IBM Cloud, Microsoft Azure, Amazon Machine Learning, Google AI Platform.
1.6. Adaptive Machine Learning (Adaptive ML)
What if we let the artificial intelligence adapt… How, you ask?.. Doesn’t it already adapt to the task? The problem is this: we painstakingly design each such task before building an artificial intelligence algorithm to solve it. They will answer you - it turns out that this chain can be simplified.
Conventional machine learning works on the principle of an open system (open-loop): you prepare data, invent a neural network (or whatever), train, then look at several indicators, and if you like everything, you can send the neural network to smartphones to solve user problems . But in applications where there is a lot of data and their nature is gradually changing, other methods are needed. Such systems that adapt and educate themselves are organized into closed, self-learning circuits (closed-loop), and they must work without interruption.
Applications - this can be streaming analytics (Stream Analytics), on the basis of which many businessmen make decisions, or adaptive production control. On the scale of modern applications, and given the better understood risks to humans, the methods that constitute the solution to this problem, all these methods are collected under the general name of Adaptive AI.
Looking at this picture, it’s hard to get rid of the feeling that don’t feed futurists with bread – let them teach the robot to breathe…
Postclassical Compute and Comms
2.1. Mobile communications of the fifth generation (5G)
This is such an interesting topic that we immediately refer to our . Well, here is a short summary. 5G, by increasing the frequency of data transfer, will make the speed of the Internet unrealistically fast. It is more difficult for short waves to pass through obstacles, so the design of networks will be completely different: base stations need 500 times more.
Along with speed, we will get new phenomena: real-time games with augmented reality, performing complex tasks (such as surgery) through telepresence, preventing accidents and difficult situations on the roads through communication between cars. From a more prosaic one: mobile Internet will finally stop falling during mass events, such as a match at a stadium.
Image source: Reuters, Niantic
2.2. Next-Generation Memory
Here we are talking about the fifth generation of RAM - DDR5. Samsung has announced that by the end of 2019 there will be products based on DDR5. It is expected that the new memory will be twice as fast and twice as capacious while maintaining the form factor, that is, we will be able to get memory sticks with a capacity of up to 32GB for our computer. In the future, this will be especially true for smartphones (the new memory will be in a low-power version) and laptops (where the number of DIMM slots is limited). And machine learning requires large amounts of RAM.
2.3. Low-Earth-Orbit Satellite Systems
The idea of replacing heavy, expensive, powerful satellites with a swarm of small and cheap ones is far from new and appeared back in the 90s. About what now only the lazy did not hear. Here, the most famous company is Iridium, which went bankrupt in the late 90s, but was saved at the expense of the US Department of Defense (not to be confused with iRidium, the Russian smart home system). Elon Musk's (Starlink) project is far from the only one - Richard Branson (OneWeb - 1440 estimated satellites), Boeing (3000 satellites), Samsung (4600 satellites), and others are participating in the satellite race.
How things are in this area, how the economy looks there - read in . And we are waiting for the first tests of these systems by the first users, which should take place next year.
2.4. 3D printing at nanoscale (Nanoscale 3D Printing)
3D printing, although it has not entered the life of every person (in the form promised by an individual home plastic factory), nevertheless, has long since left the niche of technologies for geeks. You can judge by the fact that any schoolchild knows about the existence of at least 3D sculptural pens, and many dream of buying a box with skids and an extruder for ... "just like that" (or have already purchased it).
Stereolithography (3D laser printers) allows printing with individual photons: new polymers are being investigated, for which two photons are enough to solidify. This will allow in non-laboratory conditions to create completely new filters, mounts, springs, capillaries, lenses and ... your options in the comments! And here it is not far from photopolymerization - only this technology allows you to "print" processors and computing circuits. In addition, not the first year there is , but without radical development.
3. Sensory and mobility (Sensing and Mobility)
3.1. Autonomous Driving Level 4 & 5
In order not to get confused in the terminology, it is worth understanding what levels of autonomy are distinguished (taken from the detailed to which we refer all who are interested):
Level 1: Cruise control: assisting the driver in very limited situations (for example, keeping the car at a set speed after the driver has taken his foot off the pedal)
Level 2: Limited assistance with steering and braking. The driver must be ready to take control almost instantly. His hands are on the steering wheel, his eyes are on the road. This is what Tesla and General Motors already have.
Level 3: The driver no longer has to constantly keep an eye on the road. But he must remain alert and ready to take control. This is something that commercially available cars do not yet have. All currently existing are at level 1-2.
Level 4: True autopilot, but with limitations: only trips in a known area that is carefully mapped and generally known to the system, and under certain conditions: no snow, for example. Waymo and General Motors have such prototypes, and they plan to launch them in several cities and test them in a real environment. Yandex has pilotless taxi test zones in Skolkovo and Innopolis: the trip takes place under the supervision of an engineer sitting in the passenger seat; by the end of the year, the company plans to expand its fleet to 100 unmanned vehicles.
Level 5: Full automatic driving, full replacement of a live driver. Such systems do not exist, and they are unlikely to appear in the coming years.
How realistic is it to see all this in the foreseeable future? Here I would like to redirect the reader to the article . This is partly due to the lack of 5G connectivity: the available 4G speeds are not enough. Partly with the very high cost of autonomous machines: they are still unprofitable, the business model is incomprehensible. In a word, “everything is complicated” here, and it is no coincidence that Gartner writes that the forecast for the mass implementation of Levels 4 and 5 is not earlier than in 10 years.
3.2. Cameras with 3D vision (3D Sensing Cameras)
Eight years ago, the Microsoft Kinect game controller made a splash by offering an affordable and relatively inexpensive solution for 3D vision. Since then, physical education and dance games with Kinect have experienced their short rise and fall, but 3D cameras have been used in industrial robots, unmanned vehicles, and mobile phones for face identification. Technology has become cheaper, smaller and more accessible.
The Samsung S10 has a Time-of-Flight camera that measures the distance to an object to help you focus.
If you are interested in this topic, then we redirect to a very good detailed overview of depth cameras: , .
3.3. Drones for delivery of small cargoes (Light Cargo Delivery Drones)
Amazon made a splash this year when it unveiled a new flying drone at the show that can carry small payloads up to 2kg. For the city, with its traffic jams, this seems like the perfect solution. Let's see how these drones will prove themselves in the very near future. Perhaps a cautious skepticism should be included here: there are many problems, from the possibility of easy theft of a drone, to legal restrictions on UAVs. Amazon Prime Air has been around for six years but is still in its testing phase.
Amazon's new drone revealed this spring. There is something from Star Wars in it.
In addition to Amazon, there are other players in this market (there is a detailed ), but not a single finished product: everything is at the stage of testing and marketing campaigns. Separately, it is worth noting quite interesting highly specialized medical in Africa: blood donation in Ghana (14 deliveries, Zipline) and Rwanda (Matternet).
3.4. Flying Autonomous Vehicles
It's hard to say anything definite here. According to Gartner, this will appear no earlier than in 10 years. In general, here are all the same problems as in unmanned vehicles, only they acquire a new dimension - vertical. Porsche, Boeing, Uber declare their ambitions to build a flying taxi.
3.5. Augmented Reality Cloud (AR Cloud)
A permanent digital copy of the real world, allowing you to create a new layer of reality that is common to all users. In more technical terms, it's about making an open cloud platform into which developers can integrate their AR applications. The monetization model is understandable, it is a kind of analogue of Steam. The idea is so ingrained that now some people think that AR without the cloud is simply useless.
How it might look in the future is shown in a short video. Looks like another episode of Black Mirror:
More can be read in
4. Augmented Human
4.1. Emotion AI
How to measure, simulate and respond to human emotions? Some of the customers here are companies that make voice assistants like Amazon Alexa. They will be able to really get used to the house if they learn to recognize the mood: understand the reason for the user’s dissatisfaction, try to correct the situation. In general, there is much more information in the context than in the message itself. And the context is the facial expression, and intonation, and non-verbal behavior.
From other practical applications: analysis of emotions during a job interview (based on video interviews), evaluation of reactions to commercials or other video content (smiles, laughter), learning assistance (for example, for independent practices in the art of public speaking).
It is difficult to speak on this topic better than the author of a 6-minute short film . Cleverly and stylishly crafted, the video shows how we can measure our emotions for marketing purposes, and find out from the momentary reactions of your face whether you like pizza, dogs, Kanye West, and even what is your income level and approximate IQ. By going to the film's website at the link above, you become a participant in an interactive video using the built-in camera of your laptop. The film has already been screened at several film festivals.
There is even such an interesting study: how to recognize sarcasm in a text. They took tweets with the #sarcasm hashtag and made a training sample of 25 sarcasm tweets and 000 regular tweets about everything in the world. We applied the TensorFlow library, trained the system, here is the result:
So now, if you are not sure about your colleague or friend - he said something to you seriously or with sarcasm - you can already use !
4.2. Augmented Intelligence
Automation of intellectual work using machine learning methods. It would seem that nothing new? But the wording itself is important here, especially since it coincides in abbreviation with Artificial Intelligence. This brings us back to the “strong” versus “weak” AI controversy.
Strong AI is the same artificial intelligence from science fiction films that is completely equivalent to the human mind and is aware of itself as a person. This does not yet exist and it is not clear whether it will exist at all.
Weak AI is not an independent person, but an assistant-assistant of a person. He does not pretend to human-like thinking, but simply knows how to solve information problems, for example, determine what is shown in the picture or translate text.
In this sense, Augmented Intelligence is in its purest form “weak AI”, and the wording seems to be successful, because it does not introduce confusion and temptation to see here the very “strong AI” that everyone dreams of (or fears, if we recall the numerous arguments about the “uprising machines"). Using the expression Augmented Intelligence, we immediately seem to become the heroes of another movie: from science fiction (like Asimov’s “I, Robot”) we find ourselves in cyberpunk (“augmentations” in this genre are called all kinds of implants that expand human capabilities).
Как Eric Brynjolfsson and Andrew McAfee: “Over the next 10 years, this is what will happen. AI will not replace managers, but those managers who use AI will replace those who have not had time.”
examples:
- Medicine: Stanford University developed who is on average as good at recognizing abnormalities on a chest x-ray as most physicians
- Education: assistance to the student and teacher, analysis of student response to materials, building an individual learning path.
- Business analytics: data preprocessing, according to statistics, takes 80% of the researcher's time, and only 20% - the experiment itself
4.3. Biochips
This is the favorite theme of all cyberpunk movies and books. In general, chipping pets is not a new practice. But now these chips are also being implanted into people.
In this case, the hype is most likely related to the sensational case in the American company Three Square Market. There, the employer began offering to implant chips under the skin in exchange for a fee. The chip allows you to open doors, log in to computers, buy snacks in a vending machine - that is, such a universal employee card. At the same time, such a chip serves exactly as an identification card; it does not have a GPS module, so it is impossible to track anyone using it. And if a person wants to remove a chip from his hand, it takes 5 minutes with the help of a doctor.
Chips are usually implanted between the thumb and forefinger.
Read detailed on the state of affairs with chipping in the world.
4.4. Immersive Workspace
“Immersive” is another new word that simply has nowhere to go. It is everywhere. Immersive theatre, exhibition, cinema. What is meant? Immersiveness is the creation of an immersive effect, when the boundary between the author and the viewer, the virtual and the real world is lost. In relation to the workplace, it must be assumed that this means erasing the boundary between the performer and the initiator and encouraging employees to take a more active position through reformatting their environment.
Since we now have Agile everywhere, flexibility, close interaction, then workplaces should be as easily configurable as possible, should encourage group work. The economy dictates its own conditions: there are more temporary employees, the cost of renting office space is growing, and in a competitive labor market, IT companies are trying to increase employee satisfaction from work by creating recreational areas and other benefits. And all this is reflected in the design of workplaces.
Of knoll
4.5. Personification
Everyone knows what personalization in advertising is. This is when you are discussing with a colleague today that the air in the room is somewhat dry, and you should buy a humidifier for the office, and the next day you see an advertisement on your social network - “buy a humidifier” (a real case that happened to me).
Personalization, according to Gartner, is a response to the growing concern of users about the use of their personal data for advertising purposes. The goal is to develop an approach in which we will be shown ads that are relevant to the context in which we are, and not to us personally. For example, our location, type of device, time of day, weather conditions are things that do not violate our personal data, and we do not feel the unpleasant sensation of being “surveilled”.
For the difference between these two concepts, read Andrew Frank on a blog on the Gartner site. There is such a subtle difference and such similar words that you, not knowing the difference, risk arguing with your interlocutor for a long time, not suspecting that, in general, both are right (and this is also a real case that happened to the author).
4.6. Biotech - Artificial tissue (Biotech - Cultured or Artificial Tissue)
This is, first of all, the idea of growing artificial meat. At the same time, several teams around the world are busy developing laboratory "Meat 2.0" - it is expected that it will become cheaper than usual, and fast foods will switch to it, and then supermarkets. Investors in this technology include Bill Gates, Sergey Brin, Richard Branson and others.
Source
The reasons why everyone is so interested in artificial meat:
- Global warming: Methane emissions from farms. This is 18% of the global volume of gases that affect the climate.
- Population growth. The demand for meat is growing, and it will not be possible to feed everyone with natural meat - it is simply expensive.
- Lack of space. 70% of the Amazon's forests have already been cleared for pasture.
- ethical considerations. There are those for whom this is important. The animal rights organization PETA has already offered a $1 million prize to a scientist who brings artificial chicken meat to the market.
Swapping real meat for soy is a partial solution, because people can feel the difference in taste and texture, and are unlikely to give up steak in favor of soy. So what is needed is real, namely organically grown meat. Now, unfortunately, artificial meat is too expensive: from $ 12 per kilogram. This is due to the complex technical process of growing such meat. Read about it all .
If we talk about other cases of growing tissues - already in medicine - then the topic of artificial organs is interesting: for example, a “patch” for the heart muscle, special 3D printer. known like an artificially grown mouse heart, but in general, everything is still within the scope of clinical trials. So we are unlikely to see Frankenstein in the coming years.
Here Gartner is very cautious in his estimates, apparently keeping in mind his failed 2015 prediction that in 2019 10% of the population in developed countries will have a 3D printed medical implant device. Therefore, it denotes the time to reach a productivity plateau - at least 10 years.
5. Digital Ecosystems
5.1. Decentralized Web
This concept is closely associated with the name of the inventor of the web, Turing Award winner, Sir Tim Burners-Lee. Ethical issues in computer science have always been important to him and the collective essence of the Internet is important: laying the foundations of hypertext, he was convinced that the network should work like a web, and not like a hierarchy. So it was at an early stage of network development. However, with the growth of the Internet, its structure has become centralized for a variety of reasons. It turned out that network access for an entire country can be easily blocked with just a few providers. And user data has become a source of power and income for Internet companies.
“The Internet is already decentralized,” Burners-Lee says. “The problem is that one search engine dominates, one big social network, one microblogging platform. We don’t have technological problems, but we have social ones.”
In its On the occasion of the 30th anniversary of the World Wide Web, the creator of the Web outlined three main problems of the Internet:
- Targeted harm, such as government-sponsored hacking, crime, and online harassment
- The very device of the system, which, to the detriment of the user, creates the basis for such mechanisms as: financial incentives for clickbait and the viral spread of false information
- Unintended consequences of system design that lead to conflicts and reduced quality of online discussion
And Tim Berners-Lee already has an answer on what principles could be based on the “Healthy Internet”, devoid of problem number 2: “For many users, advertising revenue remains the only model for interacting with the web. Even if people are terrified of what is happening to their data, they are willing to make a deal with the marketing machine to get content for free. Imagine a world where paying for goods and services is easy and enjoyable for both parties.” One way this could be done is for musicians to sell their recordings out of the box in the form of iTunes, and for news sites to use micropayments to read a single article instead of making money from ads.
As an experimental prototype for such a new Internet, Tim Berners-Lee launched the SOLID project, the essence of which is that you store your data in a "pod" - information storage, and can provide this data to third-party applications. But in principle, you yourself are the masters of your data. All this is closely related to the concept of peer-to-peer networks, that is, your computer not only requests services, but also provides them, so as not to rely on one server as the only channel.
5.2. Decentralized Autonomous Organizations
It is an organization that is governed by rules written in the form of a computer program. Its financial activities are based on the blockchain. The purpose of creating such organizations is to eliminate the state from the role of an intermediary and create a common trusted environment for contractors, which is not owned by anyone alone, but is owned by everyone together. That is, in theory, if the idea takes root, it should abolish notaries and other usual verification institutions.
The most famous example of such an organization was the venture-focused The DAO, which raised $2016 million in 150, of which $50 was instantly stolen through a legal loophole in the rules. Immediately, a difficult dilemma arose: either roll back and return the money, or admit that the withdrawal of money was legal, because it in no way violated the rules of the platform. As a result, in order to return the money to investors, the creators had to destroy The DAO, rewriting the blockchain and violating its basic principle - immutability.
Comic about Ethereum (left) and The DAO (right).
This whole story has ruined the reputation of the DAO idea itself. That project was made on the basis of the Ethereum cryptocurrency, next year the Ether 2.0 version is expected - perhaps the authors (including the well-known Vitalik Buterin) will take into account the errors and show something new. This is probably why Gartner put DAO on the upstream.
5.3. Synthetic data (Synthetics Data)
Neural networks require large amounts of data to train. Labeling data manually is a huge job that can only be done by a human. Therefore, artificial datasets can be created. For example, the same collections of human faces on the site . They are created using GAN - algorithms, which have already been mentioned above.
These faces do not belong to people.
A big plus of such data is that there are no legal difficulties in their use: there is no one to give consent to the processing of personal data.
5.4 Digital Ops
The “Ops” suffix has become incredibly trendy since DevOps has taken root in our speech. Now about what DigitalOps is - it's just a generalization of DevOps, DesignOps, MarketingOps ... Are you bored yet? In short, it's about moving the DevOps approach from the software realm to every other part of the business—marketing, design, and so on.
The idea of DevOps was to remove the barriers between Development itself (development) and Operations (business processes), through the creation of common teams, where there are programmers, testers, security officers, and administrators; implementation of certain practices: continuous integration, infrastructure as code, reduction and strengthening of feedback chains. The goal was to speed up the launch of the product to the market. If you thought this was Agile, you are right. Now mentally transfer this approach from the field of software development to development in general - and you understand what DigitalOps is.
5.5. Knowledge Graphs
A software way to model a knowledge area, including using machine learning algorithms. The knowledge graph is built on top of existing databases to link together all the information: both structured (a list of events or persons) and unstructured (the text of an article).
The simplest example is the card you see in the Google search results. If you are looking for a person or institution, you will see a card on the right:
Please note that "Upcoming Events" is not a copy of information from Google Maps, but an integration of the schedule with Yandex.Afisha: you can easily see this if you click on the events. That is, it is the combination of several data sources together.
If you ask for a list - for example, "famous directors" - you will be shown a "carousel":
Bonus for those who read to the end
And now, when we have clarified for ourselves the meaning of each of the points, we can look at the same picture, but in Russian:
Feel free to share it on social media!
Tatyana Volkova — Author of the IoT track curriculum at the Samsung IT Academy, Specialist in Corporate Social Responsibility Programs at the Samsung Research Center
Source: habr.com