3 ms·

Ooooo oooo pick me! We're doing something in this space. Very much early days but we DO have seed funding :) Https://sempo.ai if you're interested

by enjeyw 9y ago

thebiglebrewski 9y ago
Why do you need "AI" to solve this problem?
- enjeyw 9y ago
  Yeah good question - we're particularly interested in filtering, categorising and verifying the social media data that published during a disaster - the volume of this data is absolutely huge, so if you can use some sort classification algorithm (even basic stuff like SVM), then you can potentially massively reduce the workload of responders, who just don't have the time to that sort of manual work themselves - they're too busy doing the actual response.
  - thebiglebrewski 9y ago
    Is that really an AI though or just a bunch of if-elsif-else loops? Is a true artificial intelligence really required?!
    - enjeyw 9y ago
      If-elif loops definitely get you part of the way there; Lets say we're trying to find the tweets that actually come from people affect by a disaster, rather than just supplying commentary about them. You could just say "if tweet is not within x distance of disaster, then tweet is likely to be commentary rather than from someone actually affected". However this could always miss a tweet like "My elderly mother just called me, and she's in trouble but doesn't know who to contact" So really this is a case of "Given tweet is not within x distance, likelihood of being from affected person is lower", but of course we don't know what the affect on likelihood is (I couldn't even ballpark it for you). If you had some classified data though, you could start to get and idea of what all these Bayes Factors are. And just like that, you're doing a naive bayesian classifier. As I said, not fancy "Deep learning with CNNs", but ML nonetheless.
      - thebiglebrewski 9y ago
        I guess I see what you're saying. ...couldn't you classify the data with if/elsif/else statements though? And store likelihoods of past records within the database? Do you need to use a, "Bayesian Classifier"? Is this all that Artificial Intelligence means now? I thought it was like, "a machine that can pass the Turing test"? Sorry if I'm just naive.

Ooooo oooo pick me! We're doing something in this space. Very much early days but we DO have seed funding :) Https://sempo.ai if you're interested

by enjeyw 9y ago

Ooooo oooo pick me! We're doing something in this space. Very much early days but we DO have seed funding :) Https://sempo.ai if you're interested

thebiglebrewski 9y ago

Why do you need "AI" to solve this problem?

enjeyw 9y ago
Yeah good question - we're particularly interested in filtering, categorising and verifying the social media data that published during a disaster - the volume of this data is absolutely huge, so if you can use some sort classification algorithm (even basic stuff like SVM), then you can potentially massively reduce the workload of responders, who just don't have the time to that sort of manual work themselves - they're too busy doing the actual response.
- thebiglebrewski 9y ago
  Is that really an AI though or just a bunch of if-elsif-else loops? Is a true artificial intelligence really required?!
  - enjeyw 9y ago
    If-elif loops definitely get you part of the way there; Lets say we're trying to find the tweets that actually come from people affect by a disaster, rather than just supplying commentary about them. You could just say "if tweet is not within x distance of disaster, then tweet is likely to be commentary rather than from someone actually affected". However this could always miss a tweet like "My elderly mother just called me, and she's in trouble but doesn't know who to contact" So really this is a case of "Given tweet is not within x distance, likelihood of being from affected person is lower", but of course we don't know what the affect on likelihood is (I couldn't even ballpark it for you). If you had some classified data though, you could start to get and idea of what all these Bayes Factors are. And just like that, you're doing a naive bayesian classifier. As I said, not fancy "Deep learning with CNNs", but ML nonetheless.
    - thebiglebrewski 9y ago
      I guess I see what you're saying. ...couldn't you classify the data with if/elsif/else statements though? And store likelihoods of past records within the database? Do you need to use a, "Bayesian Classifier"? Is this all that Artificial Intelligence means now? I thought it was like, "a machine that can pass the Turing test"? Sorry if I'm just naive.