Is filtering tweets proved to be useful?

There are a lot of research going on tweets like fake news, event detection, sarcasm detection etc., So depending upon situation, people may leverage a common observable pattern to filter relevant tweets from the tweet stream.

Have anyone come across any such implementation or research?
How such application can be proved to be useful from theoretical/implementation/research view point, apart from the fact they will be time efficient?

Tags: analytics, filtering, research, text analytics, tweet, twitter

0 members like this

▶ Reply to This

Replies to This Discussion

Permalink Reply by Vincent Granville on March 27, 2018 at 5:42am

We have posted a few articles on this topic, see here. One way to handle it is to

Automatically crawl the web to see if the same article is also posted in trustworthy outlets
Has the media source in question been flagged in the past for posting news that might be fake?
Does the source domain contain words such as "viral", e.g. viralnews.com?
Is the source domain a new website, or has it been around for a while?

Eventually, you want to attach a trustworthiness score to each post, and make those with a low score less visible.

▶ Reply

Permalink Reply by surya prakash Sahu on March 28, 2018 at 6:35am

I am mainly focusing on researches/fields that are being done by filtering tweets or things that have been achieved by filtering tweets. So here, I want to know about filtering of tweets and it's various uses and implications.But your suggestion is towards different aspects and procedures of "fake news detection".

Vincent Granville said:

We have posted a few articles on this topic, see here. One way to handle it is to

Automatically crawl the web to see if the same article is also posted in trustworthy outlets

Has the media source in question been flagged in the past for posting news that might be fake?

Does the source domain contain words such as "viral", e.g. viralnews.com?

Is the source domain a new website, or has it been around for a while?

Eventually, you want to attach a trustworthiness score to each post, and make those with a low score less visible.