[NetEase Smart News, June 3rd] Microsoft Research held a natural language processing media sharing meeting in Beijing. Dr. Zhou Ming, vice president of Microsoft Asia Research Institute and chairman of the International Computer Language Association (ACL), exchanged and discussed about NLP technology. .
At the meeting, Dr. Zhou Ming spoke on three aspects of natural language processing, latest developments in technology, and future prospects. Zhou Ming stated that natural language processing is a branch of artificial intelligence for analyzing, understanding, and generating natural language to facilitate people Communicate with computer equipment and communicate with people.
Zhou Ming stated that NLP technology has a very important significance for Microsoft, including reshaping productivity and business processes, building a smart cloud platform, and creating more personalized computing. At present, Microsoft is deeply involved in NLP technology projects including Bing, Cortana, Microsoft Ice, recommendation system, smart input, assisted writing, machine translation, smart customer service, Bot Framework, cognitive computing, knowledge mapping, business intelligence, and more. At present, Microsoft Asia Research Institute has published more than one hundred ACL papers. Dr. Zhou Ming was elected as the ACL-appointed chairman last year.
Microsoft Research Asia is also committed to developing and researching NLP technology in combination with Chinese culture. Last month, Li Di, deputy dean of the Microsoft (Asia) Institute of Internet Engineering, announced that Microsoft’s Xiao Bing, who will soon be three years old, has gained more status as a poet. Since February 2017, Xiao Bing has been at the End of the World, watercress, paste it, The simplified book platform published works with 27 pseudonyms. The current open names include: Luo Meng, the fingertips of the wind, a lotus, and a smiling white. A netizen commented that: Xiao Bing, who has learned how to read and write poetry, looks smarter and lovely. Many of these technologies are based on the research accumulated by Microsoft Asia Research Institute over NLP technology for Chinese culture over the years.
In addition, Dr. Zhou Ming cited many examples to demonstrate the current research progress of Microsoft NLP. In terms of simultaneous interpretation meetings in different languages, the author believes that this is worth looking forward to, and that the future maturity of the project will solve the pain point of this work.
However, Dr. Zhou Ming believes that there are still some problems and difficulties that need to be solved in NLP research, such as machine reading comprehension. He compares machine reading comprehension to the crown jewel in the field of natural language computing.
In the above picture, we can easily give the answer: Danube, but even the best system model R-NET gives unsatisfactory output. His answer is: Cologne.
It can be seen that it is very difficult for the computer to truly understand the text content and reasoning about the text like human being. When answering the question, the computer needs to further process the algorithm and model in addition to the “it†in the text. We reasoned about the expressions "larger than" and "after", and learned that Danube is the correct answer. In addition, because the text does not explicitly mention that Danube is "river," this increases the difficulty of reasoning in the system. .
However, the development of big data also allows researchers to see the dawn, coupled with deep learning algorithms and massive cloud computing resources can be used for long-term text to learn from point to point, that is, to model sentences, phrases, contexts, which are hidden With a certain amount of reasoning ability, based on this, the Research Group for Natural Language Computing at Microsoft Asia Research Institute is conducting further research and exploration. They are adopting an end-to-end deep learning model solution.
Finally, Zhou Ming stated that personalized service through user portraits, domain adaptation through transfer learning, untagged data through unsupervised learning, etc. will become the main directions for the next study. In addition, he predicts, In the next 5-10 years, NLP technology can mature, Spoken machine translation is fully popularized, Natural language conversations (chat, question and answer, dialogue) reach pragmatic degree, Automatic writing poems, etc. NLP will also be involved in other artificial intelligence technologies. Vertical fields such as law, education, and medicine are widely used.
燑br>
Pharmaceuticals,2-Methyl- Propanoic Acid Monohydrate Price,2-Methyl- Propanoic Acid Monohydrate Free Sample,Pure 2-Methyl- Propanoic Acid Monohydrate
Zhejiang Wild Wind Pharmaceutical Co., Ltd. , https://www.wild-windchem.com