WebArena: A realistic web environment for building autonomous agents

Interesting work with the representation of the Content through the URL, to allow the agent/actor to discover the information through different path.

↓

[...]

- CSS(--variable=value)

- DOM(attribute=value)

- FORM(input[name].value)

- URL(path?param#resource)

- HTTP(?param=value{body})

- SCRIPT(--attribute=value)

- DB(model?param=value)

- FS(folder/filter/{value})

[...]

↑

- https://www.w3.org/OWL/ maybe to harmonize the representation/attribute/value

- https://opensearch.org/ for sharing patterns

- informal work on https://internet4000.github.io/find

- thread https://matrix.to/#/!EHWwJaqeEyREvfcUIK:matrix.org/$SObS1VSS...

FTA: > "WebArena is a standalone, self-hostable web environment for building autonomous agents. WebArena creates websites from four popular categories with functionality and data mimicking their real-world equivalents. To emulate human problem-solving, WebArena also embeds tools and knowledge resources as independent websites. WebArena introduces a benchmark on interpreting high-level realistic natural language command to concrete web-based interactions. We provide annotated programs designed to programmatically validate the functional correctness of each task."

The tasks are things like (an agent for github) "create a new repo" or (agent for ecommerce site) "add the cheapest item matching foo to shopping cart". Having a standard, isolated and realistic sandbox to train bots in is probably valuable to some.

Interesting to see this side of the "prove you're human" arms race that's been escalating in the news lately.

WebArena looks interesting, though the main topic of autonomous agents is not my expertise at all.

However, I am wondering.. this submission is now 10 hours old, still middle of the front page on HN, and only collected 39 points and 4 comments in that time. I wonder a) what keeps this so long on the front page? and b) why so few comments for what seems like a very innovative approach?

Is this from Google or can anyone use the typeface Google Sans?

Looks like an interesting project, however I wish they would have more text on their website instead of an array of uncaptioned videos.

That's kind of an unfortunate name, there's a major VPS/web hosting in that brand...

What is this for, bots?