MEGOGO launched an interactive channel “Look How You Hear” with sign language translation for people with hearing impairments. The IBM Watson supercomputer won the TV quiz show Jeopardy (Russian equivalent - Its Own Game)
The way they work modern technologies, is often difficult to explain, and when it comes to cognitive computing and IBM's Watson system, the topic seems beyond the understanding of the average person. But this is not at all true: professionals can explain everything, including the most complex matters, in fairly accessible words. Today IBM specialists talk about the work of the IBM Watson cognitive system and other innovative solutions of the company. These are Vladimir Alekseev, business consultant for industrial solutions at IBM in Russia and the CIS, Alexander Dmitriev, leading system architect at IBM in Russia and the CIS, and Yulia Pakina, business development manager at IBM in Russia and the CIS.
IBM has been working in the field of cognitive technologies for many years. Please tell us what projects the company is currently involved in in this direction?
Alexander Dmitriev
To answer this question, we need to talk a little about the technologies themselves. Cognitive technologies are a direction in the development of systems artificial intelligence, whose main task is to help a person make decisions in a difficult environment. There are a number of industries and processes that require management with a huge number of changing parameters, complex dependencies, and difficult to predict results. In this case, decisions must be made in near real time. The simplest example would be electronic trading on a stock exchange or online purchases. For example, popular shopping sites often run promotions where cheap items are put up for sale from a certain point in time. The person doesn’t even have time to press the button, and the product is already sold: they trigger automatic systems procurement In the same way, the purchase and sale of shares on world exchanges is supported by systems that collect a huge amount of information from various sources and “automatically” make decisions on the purchase or sale of certain blocks of shares. In fact, the decision is made by a person when he creates and trains this system.
From all this follows an understanding of the range of clients and possible projects for which cognitive technologies are suitable: these are primarily large companies from those business sectors where it is necessary to quickly and efficiently process large volumes of data, identify complex connections and dependencies, and using algorithms close to human thinking, to develop solution options so that management specialists can quickly make a choice of the necessary actions. These are primarily large production facilities, where large resources and volumes of data are involved - both from technical systems, and from the actions of personnel (oil industry, banks, construction, heavy engineering, etc.). These are also areas where mastering and understanding a large amount of information about new developments, taking into account accumulated experience and high risks when making decisions is required. This applies to the field of medicine, social management. The cost in medical decisions is human life and health. This is where cognitive technologies are especially valuable. They provide specialists with abstract information on new advances in the field of a particular medical area and help select the best treatment option for each patient, taking into account the history and specifics of his disease.
In addition, one of the most important areas is the creation of knowledge databases in a particular field of science, technology, type of activity, history, etc.
Now about specific projects. Today, a number of large companies are already creating complex decision support systems, not only abroad, but also in Russia. If we talk about foreign examples, the largest Australian oil company Woodside has created a knowledge base for its industry, taking into account its own experience. This made it possible to solve a number of complex problems - first of all, to increase the efficiency of personnel, significantly speed up the training cycle, and made it possible to use and replicate experience already completed projects. This is a huge money saving for a large company.
A Russian example would be the implementation of a system for working with large volumes texts at VINITI RAS, where technologies based on Watson Explorer are being successfully implemented. This system will help process various structured and unstructured information to identify correlations between indicators characterizing thematic areas scientific research in Russia.
There was also a project carried out jointly with the All-Russian Center for the Study public opinion on the topic of international relations. More than 55 thousand texts from open sources were analyzed using Watson cognitive technologies in order to highlight important points in cultural and social connections between Russia and South Korea.
A number of pilot projects in the field of medicine based on the Watson Health product have also been completed, and the results show ample opportunities to improve the quality of patient care.
It must be said that the field of cognitive technologies is still very new, and therefore almost every day new customers from a variety of fields come to us, and we select the necessary tools for them from the entire range of solutions available to IBM.
Please tell me what IBM Watson was originally and why they decided to use this system in areas such as medicine, business, and insurance?
Yulia Pakina
Watson's very first appearance was in 2011 on Jeopardy! Watson was a big complex back then different programs, which was compiled specifically for this game, and the capabilities inherent in it at that time were aimed at understanding natural language. To do this, it was necessary, firstly, to translate the voice into text and, secondly, to correctly interpret the resulting text material. Thus, the system was initially able to interpret the text, parse questions and “understand” their meaning. Of course, Watson's level and productivity are much higher now than in 2011.
What else could Watson do back then? Search for an answer to asked question in the depths of the information contained in it. The distinctive feature of that system was that it was not connected to external sources- neither to the Internet, nor to anything else. What was put into her memory, she used. The most interesting thing is that, thanks to the built-in logic, the system ranked the possible answer options and gave the one that, thanks to its logic, it was sure was the most correct. In 2011, the Watson system won by a large margin over the rest of the human participants; the gap was serious. And after this game the question arose: “What’s next?”
The game is great! But what could be the further application of the system? After this, experts began to think about where Watson technologies could be commercially applied, in what area of business, in what markets. IBM decided that Watson should be used where there is a large flow of textual unstructured information in natural language, and where a person needs an intelligent assistant. And we started working in the healthcare sector. The thing is that Watson, as we already know, can process a huge amount of information. And, for example, in oncology, 500 thousand new scientific articles appear annually in various fields and research areas. And it is clear that a living person cannot cope with such a deluge of information. But new materials in this area cannot be ignored; they must be processed intellectually, weighing, comparing, turning to previous experience.
At the same time, it is imperative to take into account that everything that Watson can now do was put into it by talented programmers, linguists, and subject matter experts.
What is your vision for the future of IBM Watson? For example, what problems will the system be able to solve in 5-10 years? What can she not do now, but the company’s specialists plan to teach her to do soon?
Alexander Dmitriev
Lead system architect at IBM in Russia and the CIS
It is quite difficult to predict the future; it is not for nothing that most experts, when talking about modern solutions, use the phrase “in today’s fast-paced world.” The situation is indeed changing very quickly and often unpredictably, but the main trends in the field of cognitive technologies are still quite clear.
First of all, this is the creation of large knowledge bases at the private and public level. Now all the leading countries of the world are concerned about this. They are very actively striving to collect, process and put on stream a system for extracting valuable information from a wide variety of sources. We see that this process occurs at two main levels. Firstly, at the level of large companies of international importance with hundreds of thousands of employees, branches in different countries and complex production. The main driver here is getting competitive advantage. It is clear that projects at this level require serious investments, but they begin to pay off almost immediately, dramatically increasing operational efficiency. The emphasis is on predictive analysis, which is provided by Watson technologies: management occurs not after the completion of certain events, but taking into account the entire experience of the company in forecasting mode. A higher level is the state level, when systems for accumulating and processing knowledge are created on a national scale and information from other countries. These are areas related to the development of science, technology, national health, and social management.
Watson has already mastered the professions of a cook, doctor, financier and translator. What other professions is she going to master in the near future?
Alexander Dmitriev
Lead system architect at IBM in Russia and the CIS
As for Watson's "professions", there are two aspects. The first is expanding the range of opportunities within already mastered professions. Let's say in the field of medicine, Watson is used in the treatment of a number of oncological diseases. But the specificity of medicine is that not only are there a huge number of types of diseases, but the patients themselves differ in personal characteristics and histories of their diseases. Therefore, development is due to both an increase in the range of treatable diseases and the possibility of developing an increasingly detailed personalized course of treatment for a specific patient.
The second is “mastering” other professions. Watson has already “mastered” the specialization of the oil industry: a number of foreign companies have implemented decision support systems for their oil specialists. Another promising direction is working with social groups and the population. These are also areas where it is necessary to process information and develop services and offers for large groups of clients (hundreds of thousands and millions of people). Thus, the immediate prospects for development are professions from the banking industry, telecommunications, where the volumes of data are incredibly large, and decisions must be made in real mode time.
Generally speaking, I believe that Watson will soon come in the form of a service to every person - it will be possible to ask a question on almost any area of knowledge of interest and receive a qualified answer.
Yulia Pakina
IBM Business Development Manager in Russia and the CIS
Recent areas where IBM Watson has been used include mining. Alexander has already mentioned good example success stories - the Australian company Woodside Energy, which the cognitive system helped to work much more efficiently, optimizing the work process. Before Watson, the decision to drill wells was made by Woodside Energy specialists based on long and painstaking work to collect all possible documentation in the field, including the geological structure of the area, the presence of wells nearby, the type of deposit, the possibility of using the equipment that needs to be used for this project.
Moreover, previously this preparatory period took up to 80% of the company’s time. Accordingly, only 20% of the time was left for developing the well itself. Now, together with Woodside Energy, we have achieved that only 20% of the time is devoted to research and preparation for drilling, and the rest of the time is allocated to drilling and developing new wells.
Now many companies are talking about their developments in the field of artificial intelligence. IBM talks about a cognitive platform. Please tell me what is special about IBM cognitive services and can they be called a certain type of artificial intelligence?
Alexander Dmitriev
Lead system architect at IBM in Russia and the CIS
When it comes to artificial intelligence, I wouldn't put too much emphasis on the terminology. While science as a whole does not fully understand the methods of human thinking (and there are still many blind spots in this area), it is inappropriate to argue what is artificial intelligence and what is not. We can say that the Watson system in 2011 “passed” a slightly modified formal Turing test for the right to be called artificial intelligence. The general idea of the test is simple: if a person, communicating with some system and asking it a series of questions in free form, cannot distinguish whether he is communicating with a person or with machine system, then such a system can claim the title of “artificial intelligence”.
Having won the game Jeopardy, where it was necessary to answer questions from a variety of areas of knowledge, Watson beat the live participants and passed this test. But that's not the point. Whatever we call cognitive technologies, it is important that they fulfill their main task, becoming an “amplifier” of the mind when accepting difficult decisions, both operational and strategic. Human memory is not limitless; training competent specialists in any industry is an expensive and time-consuming undertaking. Cognitive systems seem to create virtual specialist consultants whose services anyone can turn to. This is the essence of artificial intelligence. It is important that the final decision on any issues remains with the person.
Yulia Pakina
IBM Business Development Manager in Russia and the CIS
Yes, in general, the cognitive system was created precisely in order to relieve a person of routine and give more time for creativity, solving complex problems and creating new systems. That is why we are talking about a solution not of artificial intelligence, but of enhanced intelligence, added intelligence.
Please tell us more about using the capabilities of cognitive technologies in business.
Alexander Dmitriev
Lead system architect at IBM in Russia and the CIS
The use of cognitive technologies in business is aimed at solving a number of problems related not just to large volumes of rapidly changing data, but to the need to quickly extract from this data necessary information and use it for business, taking into account the industry and the company’s own experience. Thus, cognitive systems connect to various sources of information (the company’s own databases, the Internet, streaming video, information from technical sensors various systems, data on events in a particular area). Based on this data, cognitive systems use special algorithms to find the necessary solutions and offer them to managers and specialists.
It is important that with the accumulation of work experience and successful activity in a particular area, cognitive systems can be trained, configured, and also set a self-learning mode. Therefore, cognitive systems for business have one important quality, which no other system has: the longer they work, the higher their efficiency. They themselves become more valuable to the company during operation. And the important thing is that this accumulated experience is available to the company’s employees and is thus constantly used - repeatedly, repeatedly, whenever necessary. The usual situation is that a specialist leaves, and his personal knowledge and experience are lost for the company. With the implemented cognitive system, all experience remains in the company and can be easily transferred to other specialists.
How can blockchain be useful for business? Now they say that this technology can change the usual world of entrepreneurship. Is this true, and if so, what are these changes?
Vladimir Alekseev
Early ideas about how technology could change the world of entrepreneurship usually centered around the creation of peer-to-peer networks, that is, an environment where every company could interact directly with every other without any intermediaries. Admittedly, this is an overly simplified description, and over time the idea has evolved and been supplemented. Now we can say that blockchain, firstly, allows us to ensure distributed responsibility, which is extremely important if we have several companies that do not trust each other very much and are not connected in any way. Secondly, the transparency of all transactions and the impossibility of making changes to already completed transactions. A transaction means not only bank transaction, but more the fact of transferring an asset from one company to another. Thirdly, this is the possibility of using smart contracts for business logic, namely ensuring the entire operation process. Otherwise, the blockchain could only be used as a storage system, and all the logic of operations could be done outside its framework, which would not ensure either transparency or reliability of operations.
Are there already positive examples of blockchain use by commercial companies?
Vladimir Alekseev
Business consultant for IBM industrial solutions in Russia and the CIS
Over the past year, IBM has piloted more than 400 pilots around the world with customers across a wide range of industries. This, of course, includes the financial sector, retail, and energy. In particular, pilot projects with ABN Amro in the field of financial restructuring and property management. A project was completed with Bank of Tokio-Mitsubishi to use blockchain technology to automate outsourcing contracts in IT.
Talk about practical results implementations (quantitative business indicators) are now quite early: blockchain is still new technology, which also takes time to test. Blockchain cannot exist in isolation, so integration with existing systems, service development is required, competencies are required. 2016 was dedicated to piloting; 2017 should be marked by the integration of blockchain technology into the existing IT infrastructure of organizations.
According to a study by the IBM Institute for Business Value, more than 50% of finance executives surveyed plan to move to the commercialization phase of the technology in 2018-2020.
Blockchain, as far as one can understand, provides great opportunities for many areas of business. What can you say about securities trading exchanges? Could this technology be useful there?
Vladimir Alekseev
Business consultant for IBM industrial solutions in Russia and the CIS
It is worth recognizing that exchanges were one of the first organizations that became interested in the technology and actively participated in its development. For example, the German Exchange (Deutsche Boerse) is a prime participant in the HyperLedger blockchain project along with IBM, and the Moscow Exchange is also a member. From the practical experience of using technology by exchanges, I would like to note the following: last year, the Japanese Exchange, with the help of IBM, conducted a study of the possibilities of using distributed registries in its operations. In its report, the exchange emphasized the promise of the technology, noting among the key advantages the ability to create new innovative financial services and cost reduction. According to experts from the Japan Exchange, blockchain will help automate approval processes trading procedures and increase the fault tolerance of the system as a whole by introducing the principle of distribution.
Please tell us what IBM plans to do in the next 5-10 years? How does the company see the business world of the future?
Vladimir Alekseev
Business consultant for IBM industrial solutions in Russia and the CIS
Earlier this year, IBM unveiled its vision of how technological innovation will change the world in the future. The report was produced by IBM Research and reflects the company's views on how the world will change in five years in five ways. First, the company pays close attention to how we all speak and write, and believes that these factors will be used as indicators of psychological state and physical health. Next, people will be able to gain “super vision” thanks to tiny and powerful cameras, which will make it possible to explore almost 100% of the electromagnetic spectrum versus less than 1% currently. The technology can be built into mobile devices and help analyze the composition of foods or medications. On the other hand, in five years we will be able to understand the complexity of the Earth in stunning detail. This will become possible due to the development of the Internet of Things (IoT) and machine learning algorithms, on the basis of which conclusions can be drawn from the analysis of the measured parameters. IBM scientists refer to this collectively as a "macroscope." It will help predict phenomena such as changes in climate, water levels, pollution threats or the impact external factors to our planet.
The next direction of IBM technology development is the creation of medical laboratories “on chips” to track diseases at the nano-level, which will help predict diseases at earlier stages. In IBM laboratories, specialists are working on creating 20 nm chips that can be connected to both artificial intelligence systems and other sensors in real time. And finally, the fifth area is the creation and distribution of “smart” sensors for earlier detection of pollution levels environment. Such sensors can also be extremely useful for gas pipelines, as well as near natural sources of emissions, for example, methane, to alert about increasing concentrations of various substances.
It is worth noting that in all areas the technologies are already in development, so the forecast does not look too futuristic. On the other hand, it takes time and effort to finalize existing products and bring them to mass use.
From a long-term perspective (10 years or more), an example can be given of quantum computer technology. The operating algorithm of quantum computers contains completely different principles than those on which they operate. modern computers. Therefore, their use can completely change existing processes, such as cryptography, and give absolutely new level computing power. IBM is one of the leaders in this area, already providing free trial access to real quantum computer via the IBM Quantum Experience cloud infrastructure.
Ask an IBM expert Elena Sinka, IBM solutions sales representative Ask
Thank you!
Your question has been sent
Modern supercomputers are several server computers connected to a network. Their computing speed is measured in petaflops.
- 1 petaflops = 10 15 operations per second
The average performance of the human brain is 20 petaflops. Only a few supercomputers in the world have great performance, but none of them can replace the human brain.
There are currently several hundred supercomputers in the world. The most powerful ones are included in the annual TOP-500 ranking. In 2016, this rating was topped by the Chinese Sunway TaihuLight. Prior to this, he also held the leadership for three years Chinese computer Tianhe-2. IBM has two supercomputers in this ranking: Mira and Sequoia. The latter was the leader in 2012, and now ranks fourth.
Andrey Filatov ( CEO IBM in Russia and CIS countries) about cognitive technologies
Dr. Watson is the most famous supercomputer
The main advantage of Watson is that it understands questions in natural language and answers them by analyzing data. In 2011, Watson beat people on a game show Jeopardy!(Russian equivalent - “Own Game”).
Watson is a set of application technologies called " cloud services" Watson is most actively used in medicine, helping to diagnose and treat cancer. Its memory contains more than 600,000 medical reports. It is also used in finance, law, hospitality and many other industries. Moreover, he is even able to carry on a conversation with celebrities.
An error occurred while loading.
Watson talks to American tennis player Serena WilliamsApplications for IBM Watson
Education. Schools across the US are testing Teacher Advisor with Watson, a cognitive tool that offers advice on improving curricula and personalizing learning programs.
The science. Johnson & Johnson uses Watson to analyze scientific literature. From a colossal amount of materials, he selects those necessary for research, and research can be carried out much faster and more efficiently.
Safety. Rolled steel manufacturer North Star BlueScope Steel is looking to use Watson Internet of Things to create solutions to protect workers in extreme situations. Workers will also wear devices to collect and process data. If hazardous conditions occur, information will be immediately sent to North Star management.
Cybersecurity. Cybercriminals hack Information Systems enterprises, and then sell access to them on the “black” Internet. If in one part globe If there is a failure or fraud, the Watson system will alert other users of the system.
Medicine. University North Carolina and 12 other cancer research centers use Watson to analyze patients' DNA to then develop personalized treatments.
No doctor is able to analyze such a huge sample of information, only a computer
A computer can do a lot, at least when it comes to processing information. However, teaching him natural language is an extremely non-trivial task. This challenge formed the basis of the IBM DeepQA project, which resulted in the birth of a cognitive technology called IBM Watson, in honor of Thomas Watson, who stood at the origins of IBM.
It is not difficult to explain what Watson is - it is a cognitive system capable of communicating with a person in natural language. That is, understand written speech and respond in the same way. And if IBM had limited themselves to this, Watson would have remained nothing more than an experimental setup. But a job was quickly found for him, and for many companies he became a truly indispensable employee.
It turned out that this know-how can be applied wherever large volumes of unstructured data need to be processed. For high-quality and fast analytics of such data, they need to be processed using all the available tools of modern computer technology: machine learning, computational linguistics, ontological constructions and high-performance computing. That's what IBM Watson is designed to do.
IBM Watson's core competencies can be summed up in four points:
- Natural language understanding.
- Building hypotheses based on processed data.
- Learning on the job.
- Making a recommendation, accompanied by the facts on which the conclusion is based.
A person is not capable of analyzing a truly large amount of data in a reasonable amount of time, and in any case he will have to discard most of the information, highlighting, in his opinion, the main thing. Errors are inevitable here, in addition, discarded data also matters and should influence the result. And in this aspect, Watson is many times superior to humans: it takes into account everything, none known fact will not go unappreciated.
The first public test of the system was participation in the American game Jeopardy! (Russian equivalent - “Own Game”). Without an Internet connection, using open sources information such as the text of the entire Wikipedia, general encyclopedias and dictionaries, Watson was able to defeat the two record holders of this game.
We invite you to the IBM client center for a seminar Watson Analytics and Hi-tech in the field of analytics!
IBM Watson is one of the first cognitive systems in the world. This system can do a lot, thanks to which Watson's capabilities are used in many areas - from cooking to accident prediction in populated areas. In general, most of Watson's capabilities are not unique, but taken together, all these capabilities represent a very powerful tool for solving a variety of issues.
For example - natural language recognition, dynamic system learning, construction and evaluation of hypotheses. All this allowed IBM Watson to learn to give direct, correct answers (with high degree reliability) to the operator’s questions. At the same time, the cognitive system is able to use large arrays of global unstructured data, Big Data. What are the basic principles of how IBM Watson works with language? More on this in the sequel.
Main challenges of natural language recognition
For humans, language is a means of expressing thoughts. We use language to convey our opinions, data and information. We can make predictions and form theories. It is language that is the cornerstone of our consciousness. At the same time, here is a paradox: human language is very inaccurate.Many terms are illogical, and it can be very difficult for computer systems to understand us. For example, how can a voice be thin? How can you burn out of shame? For a machine this is a problem, but for a person it is a completely ordinary thing. The fact is that in order to correctly answer a question, in many cases it is necessary to take into account the existing context. Without sufficient factual information, it is difficult to answer a question correctly, even if you can literally find the exact answer to the elements of the question.
Natural Language Processing - Getting Started
Many computer systems are able to analyze language, but at the same time a superficial analysis is carried out. This may make sense, for example, in order to make a statistically valid assessment of trends in changes in emotions on large amounts of information. Here, the accuracy of information transfer is not very important, since even if we assume that the number of false positive results is approximately equal to the number of false negative results, they cancel each other out.But if all cases matter, then systems that work with superficial language analysis can no longer do their job properly. A striking example of this could be the task for voice assistant any of mobile devices. If you say “find me a pizza,” the assistant will display a list of pizzerias. If you say “don’t look for pizza for me in Madrid,” for example, the system will still search. Such systems work by identifying certain keywords and using a specific set of rules. The result may be accurate in a given system of rules, but incorrect.
Deep Natural Language Processing
In order to teach the system to analyze complex semantic structures, taking into account emotions and other factors, experts used deep natural language processing. Namely, a question-answering content analytics system (Deep Question*Answering, DeepQA). If greater accuracy is required, then you have to use additional methods natural language processing.IBM Watson is a deep natural language processing system. When analyzing a specific question, the system tries to evaluate as broad a context as possible in order to give the correct answer. This uses not only the question information, but also knowledge base data.
The creation of a system capable of deep processing of natural language made it possible to solve another problem - analysis huge amount information that is generated daily. This is unstructured information, such as tweets, messages social networks, reports, articles and more. IBM Watson has learned to use all this to solve human problems.
IBM Watson Cognitive System
Watson is a different level of computing capabilities. The system can separate certain statements in natural language and find connections between these statements. At the same time, Watson copes with the task, in many cases, even better than man, while data processing is much faster, work is carried out with much larger volumes - a person is simply incapable of this.
Basic characteristics of the cognitive system
The system works in this order:1. When Watson receives a question, it executes it. parsing to highlight the main features of the issue.
2. The system generates a series of hypotheses by scanning the corpus in search of phrases that, with some degree of probability, may contain the required answer. In order to lead efficient search in streams of unstructured information, completely different computing capabilities are needed * they are called cognitive systems. (I don’t really understand the last sentence and the role of the asterisk)
3. The system performs a deep comparison of the question language and the language of each of the possible answer options, using various algorithms logical inference.
This is a difficult stage. There are hundreds of inference algorithms, and they all perform different comparisons. For example, some search for matching terms and synonyms, others look at temporal and spatial features, while others analyze suitable sources contextual information.
4. Each inference algorithm provides one or more scores indicating the extent to which a possible answer follows from a question in the domain considered by the algorithm.
5. Each score obtained is then assigned a weight by a statistical model that records how well the algorithm did in identifying logical connections between two similar phrases in that domain during Watson’s “training period.” This statistical model can then be used to determine the overall level of confidence Watson systems is that a possible answer follows from the question.
6. Watson repeats the process for everyone possible option answer until he finds answers that have a better chance of being correct than others.
As mentioned above, to correctly answer a question, the system needs to access additional data sources. These can be textbooks, manuals, FAQs, news and everything else. Watson processes huge amounts of information in seconds to get the right answer. At the same time, the found content is also checked, outdated and useless data is eliminated.
Elements of the cognitive system
Watson deduces the general meaning of the text from the information received, from additional base. This uses the title of the document, part of the text of the document, or all of the text.
Cognitive systems, their methods of collecting, remembering and retrieving information are similar to how humans analyze information. In this case, cognitive systems can transmit information and act. Here are examples of behavioral constructs that are used in this case:
Ability to create and test hypotheses;
- the ability to break down into components and build logical conclusions about the language;
- ability to extract and evaluate useful information(such as dates, locations and characteristics).
Without these abilities, neither a computer nor a human will be able to determine the correct relationship between questions and answers.
Cognitive processes are more high order can achieve a high level of understanding by focusing on basic behaviors. In order to understand something, we must be able to divide information into smaller elements that are fairly well organized at the level in question. Physical processes in humans proceed completely differently from processes on a cosmic scale or at the level of elementary particles. Likewise, cognitive systems are designed to operate at the human level, even though they represent a huge variety of people.
In this regard, understanding language begins with understanding more simple rules language - not only the formal grammar, but also the informal conventions that are observed in everyday use.
What is this all for?
Now, thanks to years of training and improvement, the IBM Watson cognitive system can perform work in a variety of areas. Here we have medicine, cooking, linguistics, and solving business problems with scientific problems.
Initially, specialists had a choice - to make the system universal or specialized. Each option has its own advantages and disadvantages, but the choice was made in the direction of versatility.
The company has already been convinced many times that its choice was correct - before