What are the options for automatic tagging?

Question

metadata

There are a large number of articles. There was a need to put down tags for them. What options are there to automate the process?
I thought about putting frequently used words in the article into tags, but these can be garbage words, and besides, keywords can occur only once. Moreover, tags can contain phrases of two or three words. In this case, you will have to put all the words and possible phrases into the tags, and this is several hundred - not suitable.
A variant with stop words and a pre-compiled list of possible tags is being considered. In the course of the analysis of articles, these arrays will be replenished. For each article there will be manual moderation, I just want to speed up the work by partially automating this process.
Perhaps you have come across ready-made algorithms or services with api (even if paid) or without it, which can form a semantic core?

Answer 1

You can make separate lists of tags that are more suitable for articles. Complete the list manually. Then use this list, keep track of what is most often found in the text. If it does not find it, then offer to enter it manually and write it to the list. So this list will grow and less and less can be entered.
However, what they wrote. In my opinion the most optimal.
You can also read an article about the algorithm for finding words similar in meaning: habrahabr.ru/post/110078 .

Answer 2

Oh, it's probably better to see how all sorts of search engines do it, when they give a list of files containing this word by keyword