Answer the question
In order to leave comments, you need to log in
How to parse vk comments by keyword?
I conduct research on the image of the city in social networks. It is necessary to collect a corpus of comments from urban communities in different cities. Comments are collected by keyword in the interval 10/01/2019-10/01/2020.
1) Used classic parsers (targethunter, barkov.net, pepper.ninja). None completed the request correctly. Pepper.ninja does the best job - it collects comments from 10/01/2019 - 05/01/2020 and from 09/23/2020-11/17/2020. Between May and September there is some kind of black hole.
2) Used VK API. With a stretch, the only method is suitable - wall.search. The method produces an unacceptably small number of comments. For example, in the city "Overheard" with the word "Perm" has collected 2000 comments. The last comment is also in September 2020. There is some kind of bug, but those. support says everything is fine.
Perhaps someone knows what the problem is? What methods would you use?
Answer the question
In order to leave comments, you need to log in
I would not use the API at all, but stupidly directly go through all the comments with the help of a JavaScript bot.
For example, you can parse all posts (more precisely, their id and date) from a group using the usual scraping tools, after processing the date, go through these posts using the wall.getComments method, simultaneously executing the necessary logic for comments.
Didn't find what you were looking for?
Ask your questionAsk a Question
731 491 924 answers to any question