4 ms·

Show HN: Filter web content according to any criteria

5 points by romzmr 5y ago

romzmr 5y ago
Hi HN! We are developing a new search engine powered by machine learning to filter web content. The beta of our platform has just been released. Anyone can contribute by training our search engine, own tokens and use them to add content to the platform. The search engine still makes a lot of mistakes but it is getting better everyday thanks to contributors. Don't hesitate to add your own projects! For more information: Twitter: https://twitter.com/Sievable https://twitter.com/Sievable Discord: https://discord.gg/x9rmRNvdnS https://discord.gg/x9rmRNvdnS Linkedin: https://www.linkedin.com/company/sievable/about/ https://www.linkedin.com/company/sievable/about/ Hoping to get some feedback!
rezuz 5y ago
Hi romzmr It looks like a interesting concept, and website looking good ! I played a bit with it, and I often have no results it that normal?
- romzmr 5y ago
  Hi rezuz, Thanks for your comment. Yes unfortunately we just got started and we don't have a lot of content for now but we hope to make the amount of content grow quickly to make the search engine really useful. For now it's more like a demo.
  - rezuz 5y ago
    What about the wikipedia content or dbpedia ? >90 millions items should able to make you start right away. Or is it just too big to train your system with ? Or at least avoid the big mistakes it is doing right now If I type "website" I get a movie and an apple haha
    - romzmr 5y ago
      Yes indeed that would be nice source of content but we are just not ready for that yet. The search engine can in theory handle that amount of data but we don't have the infrastructure. Also, getting content automatically doesn't solve the problem entirely as we also have to train our model to filter it. The search engine makes a lot of mistakes but it gets better everyday
az_3 5y ago
Hi romzmr, The concept looks interesting but it just doesn't seem to work. I typed shoes but I got animals, movies and Companies. I think it's broken.
- romzmr 5y ago
  Hi az_3, Yes in fact we rely on our community to train our model (they get tokens in return) and we don't have a lot of data yet so it's still doing a lot of mistakes. But as it's written on our website, the search engine keeps improving everyday!

5 points by romzmr 5y ago

Hi HN! We are developing a new search engine powered by machine learning to filter web content. The beta of our platform has just been released. Anyone can contribute by training our search engine, own tokens and use them to add content to the platform. The search engine still makes a lot of mistakes but it is getting better everyday thanks to contributors. Don't hesitate to add your own projects! For more information: Twitter: https://twitter.com/Sievable https://twitter.com/Sievable Discord: https://discord.gg/x9rmRNvdnS https://discord.gg/x9rmRNvdnS Linkedin: https://www.linkedin.com/company/sievable/about/ https://www.linkedin.com/company/sievable/about/ Hoping to get some feedback!

rezuz 5y ago

Hi romzmr It looks like a interesting concept, and website looking good ! I played a bit with it, and I often have no results it that normal?

romzmr 5y ago
Hi rezuz, Thanks for your comment. Yes unfortunately we just got started and we don't have a lot of content for now but we hope to make the amount of content grow quickly to make the search engine really useful. For now it's more like a demo.
- rezuz 5y ago
  What about the wikipedia content or dbpedia ? >90 millions items should able to make you start right away. Or is it just too big to train your system with ? Or at least avoid the big mistakes it is doing right now If I type "website" I get a movie and an apple haha
  - romzmr 5y ago
    Yes indeed that would be nice source of content but we are just not ready for that yet. The search engine can in theory handle that amount of data but we don't have the infrastructure. Also, getting content automatically doesn't solve the problem entirely as we also have to train our model to filter it. The search engine makes a lot of mistakes but it gets better everyday

az_3 5y ago

Hi romzmr, The concept looks interesting but it just doesn't seem to work. I typed shoes but I got animals, movies and Companies. I think it's broken.

romzmr 5y ago
Hi az_3, Yes in fact we rely on our community to train our model (they get tokens in return) and we don't have a lot of data yet so it's still doing a lot of mistakes. But as it's written on our website, the search engine keeps improving everyday!