Hello HN, we are proud to introduce Aero, the first bot in our family.
Aero is a semantic search engine in the spiritual lineage of "Knowledge Graph".
We are in beta, so it still is for advanced users, but advanced users will have some fun tinkering with it.
The machine learned natural language processing is not yet plugged in, we are currently collecting more data for it, we will keep you updated when we turn it on.
The technology can scale, but we are not scaling it up yet. In particular, latency and throughput will be improved once we switch to our distributed backend.
Developped under firefox, Not working on IPad, and it has some css quirks on non-Edge IE.
Hi, it seems interesting! But I wasn't able to understand how to use it. I believe it's because the "top interface" is not that self explanatory. Is the training only done manually by users?
I look forward to try a more advance version!
The top interface is for advanced users, it is not self explanatory but it is very close to the standardized language Sparql (which would get you started). We also worked on the auto-completion to help. You can also learn by example, by using "Grab Spaql" button.
We plan to make some tutorial videos to explain how to use it.
We are using various ways for gathering data including users and mechanical turks. But we also have some semi-supervised algorithms. And we are also working on a fully unsupervised algorithm, but we are not there yet, and we won't disclose more for now.
3 comments
[ 3.2 ms ] story [ 19.3 ms ] threadAero is a semantic search engine in the spiritual lineage of "Knowledge Graph".
We are in beta, so it still is for advanced users, but advanced users will have some fun tinkering with it.
The machine learned natural language processing is not yet plugged in, we are currently collecting more data for it, we will keep you updated when we turn it on.
The technology can scale, but we are not scaling it up yet. In particular, latency and throughput will be improved once we switch to our distributed backend.
Developped under firefox, Not working on IPad, and it has some css quirks on non-Edge IE.
Have fun exploring.
Please give us some feedback.
Thank you
The top interface is for advanced users, it is not self explanatory but it is very close to the standardized language Sparql (which would get you started). We also worked on the auto-completion to help. You can also learn by example, by using "Grab Spaql" button.
We plan to make some tutorial videos to explain how to use it.
We are using various ways for gathering data including users and mechanical turks. But we also have some semi-supervised algorithms. And we are also working on a fully unsupervised algorithm, but we are not there yet, and we won't disclose more for now.
Thank you again for your interest.