How does Kinopoisk determine keywords for a movie?

Question

Algorithms

I wonder if this is done by a content manager who watches a movie and then describes the keywords, or is it done somehow in a machine way?

Answer 1

It looks like it's all about describing the content of the movie, sorted according to certain rules:
1. full descriptive action,
2. descriptive character type,
3. all emotional characteristics or actions,
4. proper names, place names, etc.
5. all word combinations of objects/actions containing any word form: "woman, man, name, motive, villain, hero, actor, origin" (etc.).
Word forms (item 5) is most likely a GENERAL table of nouns based on the descriptions of ALL films that are most often found in the descriptions of all these films combined. (i.e. auto-generated tags)
Also, I think that the manual addition of any initial word forms has been done to maximize the construction of the necessary thematic groups (tags / nodes) that are interesting for the reader.

Answer 2

Here is a screenshot from imdb.
Well, then, a script that passes words through its prism. Voila!

Answer 3

Kinopoisk itself does nothing. He has been parasitizing with IMDB for a very long time. Their know-how is that they once wrote a parser with imdb and then stupidly run it through the translation. The description is corrected by the editor, but the actors and the entire film crew, words, genres, countries - he will drive away by comparing the word-translation.
Then they also suggested that ordinary users do this through the "add" form - the user can send a keyword and covers and frames there - but they got tired of it - some materials are not added for months. Because They don’t have money for the development of the CP - they hire people - the toad is strangling, well, their idea failed with the cinema. So they bought a site once written on their knees - now they don’t know what to do with it. The audience forbids them to change the redesign - now they are slowly strangling it.

Answer 4

it makes a person.

Answer 5

I wonder when someone will write a neuron for automatically generating tags from descriptions and reviews?