@FootieWriter And points many other flaws in establishing if tweets are accurate or not and that’s where the confidence level in results falls majorly and his success is identifying those failures and suggesting further paths to correct in future
@FootieWriter It does seem we are discussing 2 different things 😂 I think the sets of accounts chosen is odd but that was based off having ITK or football journo in the bio. The evaluation of rumours is did that transfer happen (author admits that posts could have been true when posted)
@FootieWriter It does seem we are discussing 2 different things 😂 I think the sets of accounts chosen is odd but that was based off having ITK or football journo in the bio. The evaluation of rumours is did that transfer happen (author admits that posts could have been true when posted)
@FootieWriter While you are concerned about your own score , he’s proving that spreading false rumours creates a positive gain for any accounts that do. While you may disagree that wasn’t your intention, the logic behind accounts spreading the story to begin with is strong
@FootieWriter The language model is a necessity to identify tweets as transfers rumours, using keywords. He got a 97% confidence in that, that’s really the main part of the dissertation. He proves false rumours get more attention and attention is what all journos want on X.
@FootieWriter I agree with you and so does the author, I wouldn’t be knocking the dissertation though because he basically built a language model similar to what AI is now and that’s what the computer science dissertation is more about than if early results aren’t perfect
@FootieWriter Statistically significant to worth including on their own? It comes down to scraping a mass of data and the limitations of that. A PhD or researcher I would expect to go the extra mile but with a masters, they proved they can get results, measure their accuracy and show flaws.
@FootieWriter Dissertations aren’t expected to be perfect, highlighting flaws is a part of it, the confidence level in results matters. The data handling is the difficulty, how many other accounts do you think have a mix of exclusives and aggregating? Would your sample size of exclusives be…
@FootieWriter Either would increase the confidence level of the results but wouldn’t change the accuracy of people who only post exclusives, which we both agree is what is more important e.g. excluding you wouldn’t make Ornstein more or less accurate. A mountain of work to gain very little
@FootieWriter how would you treat aggregator accounts in your study? Bearing in mind, it’s a statistical formulae that is creating the results and not the student going through each tweet individually. They’ve created a base that another student can work from to be more accurate
@themagic_tophat The socialist in me would have much bigger issues with the wages in the PL, agent fees and in their youth academies and the consequence of that on the football pyramid. Recognised as a cultural feature of the UK, this is a way of sharing the benefit of football throughout the UK
@Andy_Leonard01@KieranMaguire What percentage of worldwide “fans” who have zero affiliation to Manchester and started supporting City in the last 15 years do you expect to stay supporting City if the appeal fails? It should zero but some people have no morals in fairness
@Michael__Lawlor@KieranMaguire Do you honestly believe American fans have any genuine loyalty to Man City? These guys have swapped more crests than Peter Crouch and are much less likable
@TheTimothySteve If you apply that same logic domestically the only available argument is that football as a whole has gotten a lot better since 2019. LFC were quite literally the best team in Europe and the world, and still somehow not the best team in the PL. I’m not sure you can get any better