• PhilipTheBucket
    link
    fedilink
    English
    312 days ago

    So anyone can look it up, but in order to look it up, you’d have to build some tools and it’s a whole project?

    • @TropicalDingdong@lemmy.world
      link
      fedilink
      0
      edit-2
      12 days ago

      I mean any one could go find some examples from memory that they experienced. I could dig far enough into my comment history to find them. In-fact I was digging through some banned community members and found some examples just the other day.

      What I’m doing is far, far larger in scope. I’m not trying to find one instance, I’m trying to find all of them. I’m also interested in correlating that to “shifts in the overall narrative” to the sub. And I’m trying to do this across several prominent subs. And I’m not doing this in an adhoc way. When I have results they’ll be publishable.

      I’ve built some of the more important tools already which allow me to pull the entire comment history of a user and perform significant sentiment analysis, key phrase extraction, etc… but some aspects aren’t reliable enough yet to be completely useful.

      This is some example output using flyingsqids data: https://tmpweb.net/jS19ePfgNdz0/

      (scroll to the bottom, then scroll up instead of starting at the top)

      The first analysis is a “trolling/ not trolling” analysis. Then its a frequency analysis. I used squid because of their preposterous number of comments. Some weeks they were commenting almost ever 3 minutes for hours on end.

      If life we’re simpler I’d be further along on this project, but alas, the bills. They do not pay themselves. And its a hobby thing I’m not getting paid for, so its the last to get access to my time.

      • PhilipTheBucket
        link
        fedilink
        English
        112 days ago

        I mean any one could go find some examples from memory that they experienced.

        Could you find some examples from memory that you experienced, for me?

        I feel like we keep having the same conversation here.

      • PhilipTheBucket
        link
        fedilink
        English
        -112 days ago

        Well, I wasn’t the one who said it was easy (and IDK that digging through the modlog is the easiest way even for someone who is sure that it happened to find out when it did), but sure. Here are all the posts from Dec 2023 and Jan 2024 that were removed that had “poll” in the title:

        7554770 | 2023-12-29T13:56:49.802793Z | Sarah Huckabee Sanders lowest approval rating for governor in last 20 years, Arkansas Poll says | https://www.thv11.com/article/news/politics/sarah-huckabee-sanders-lowest-approval-rating-governor-20-years/91-c76da35b-4704-46de-abc0-0a42ee19ea95
        2806047 | 2023-12-29T13:18:36.770457Z | Trump Fan Who Threatened Poll Workers And Officials Sent To Prison | https://crooksandliars.com/2023/08/trump-fan-who-threatened-poll-workers-and
        2461059 | 2023-12-29T13:18:19.629020Z | Donald Trump Has an Absurd Amount of Support From Republicans Who Believe He Committed “Serious Federal Crimes”: Poll | https://www.vanityfair.com/news/2023/07/donald-trump-ron-desantis-2024-poll
        3653177 | 2023-12-29T13:16:15.792290Z | Democrat Adam Frisch leads against Rep. Lauren Boebert in poll for 2024 race | https://www.denverpost.com/2023/08/22/adam-frisch-lauren-boebert-poll-2024-race/
        10024810 | 2023-12-29T13:08:37.582079Z | Trump Shares Poll Result Predicting 'Revenge' And 'Dictatorship' As Top Priorities | https://www.huffpost.com/entry/donald-trump-poll-dictatorship-revenge_n_658beb48e4b0cd3cf0e41a98
        

        I was assured there would be some that showed Biden behind in the polls, that the mods were trying to cover up…

        It’s a silly thing to get hung up on, but it helps to demonstrate that the person I’m talking with is talking about some situation that didn’t happen in reality.