Using Prodigy For Rumors Classification


Can I use Prodigy for Rumors Classification I have Tweeter dataset and want to classify rumors from it anyone please help me regarding this I scrap 1200 rumors data from fact check websites and want to train my model using prodigy and apply this to my dataset which contain around million tweets is it possible in prodigy.

waiting for reply

Hi! How you set up the annotation task and which machine learning approach you choose really depends on the specifics of your problem and the end goal. For instance, how you define a "rumor" and what you want to use this information for later on. For instance, are you only looking to identify tweets discussing rumours, and then compile stats about that? Are you looking to categorise them or link the back to factchecking sites? Is there enough context to even predict whether something is a rumour (since many people who are spreading misinformation will present the content as facts, not rumours)? These are all decisions you'll have to make, based on the end goal and the data.

In general, starting by categorising the tweets into relevant / not relevant could be a good first step. In the next step, you could then assign more specific topical categories and train a model using only the examples classified as relevant. The documentation on text classification should help you get started: