How to learn in natural language processing?

The field of natural language processing is too big. A very costly suggestion is to buy a few books, research the catalog, and flip through the content. Find your favorite segment, and then find a variety of courseware, code online, learn a piece first. On which day, when you read the book and feel that most of it is understood, I personally feel that I can live by natural language processing.

If you like using GitHub and can understand basic English materials, it is highly recommended to check out this set of GitHub-based open source natural language processing courses launched by Russian technology giant Yandex. It is completely free. In the past month, this course has been particularly popular in the machine learning section of Reddit, which has already harvested more than 2,000 stars in less than two weeks. Although this is a course launched by the Fighting Nation, most of the materials in it are in English, and only a small part of the materials and videos are in Russian. Why is this a GitHub-based course? Because most of the learning materials and resources used in it come from GitHub, especially the code and project examples, all have a warehouse on GitHub, after reading the theory, just click on the code and it is very convenient. Let's talk briefly about this course.

• Week 1: Learning Embedding 

• Week 2: Learning Text Classification 

• Week 3: Learning Language Models 

• Week 4: Learn Seq2seq and Attention Mechanism

• Week 5: Learn Structured Learning 

• Week 6: Maximizing Learning Expectations 

• Week 7: Learning Machine Translation 

• Week 8: Learning Transfer Learning and Multitasking 

• Week 9: Learning Area Adaptation 

• Week 10: Learning the Dialogue System 

• Week 11: Learning Generative Models 

It can be seen that the entire learning process is gradual, and what is learned every week is refined, such as the "embedded" learning in the first week, which will specifically learn word embedding, semantic distribution, LSA, Word2Vec, and GloVe.

Then look at the learning materials. Weekly learning resources include slides, Yandex self-made courses, lectures & seminar videos, video courses from Stanford University and other famous schools. There are also practice questions attached to the course, just open the corresponding notebooks. 

In addition, papers and articles related to what is learned each week are provided. In short, if you can complete this course of Yandex, I believe that you can basically master the knowledge of natural language processing. Even if you feel that English is not proficient, it is difficult to study English courses. It is also possible to search for Chinese learning materials at the corresponding stage according to the learning ideas of this course.

Course Portal: https://github.com/yandexdataschool/nlp_course