bijection

Neet idea!

What representation are you using for each site (ie a sparse vector, full text, etc)? How do you compare newly visited sites to old ones?

1. I tried it before and I users feedback was: I do not understand when actually the best time to surf. Probably my UI was bad.

Haha well whatever works I guess!

2. Is "Use my current location" should be a button after click it should open pop-up and ask for a permission to use location?

Exactly.

4. You mean after the search or after getting first success load home page with current location?

Either one. It just doesn't make sense to show it if your app didn't find anything

(Sorry for the late reply!)

You're right! My scheme is rusty I guess.

I should have said:

  (x . (y . (z . NIL))) <=> ['x', 'y', 'z']

Although I'm also skeptical of coding in JSON, s-expressions are pretty close to a JSON subset (with the caveat that everything needs to be quoted).

  (x . (y . (z . NIL))) <=> ['x', ['y' , ['z', null]]]

Congrats on making your first app!

I don't surf, but it presents the information nicely :)

Here are a few tips:

1. Consider putting the 'details' on the results page (instead of giving them their own page). There is definitely enough space, and it might help your users understand how you made your recommendation.

2. When sites ask for location permissions on the first page load, a lot of users automatically click 'Deny'. One way you might fix this is to wait for the user to come to you, by using a link under the search bar, like this:

                     ___________________
  I want to surf in | San Francisco     |
                     -------------------
                     Use my current location

3. In the short term (before you do bigger changes like reorganizing the search bar, if you choose to), "Where you want to surf" should read "Where do you want to surf"

4. Consider hiding the subscription box until after your user has successfully gotten a recommendation.

Good luck!

Plenty of scientists, entrepreneurs and creative thinkers have backgrounds that would allow them to subsist with minimal work, and choose to do what they do anyway.

Re: "It feels like there's a subtext to a lot of it which sort of suggests that ideas can only be heard and understood clearly by those who own the appropriate labels", linguistic prescription[1] seems relevant. From wikipedia:

"Prescription has a tendency to favour the language of one particular region or social class over others ... frequently, a standard dialect is associated with the upper class"

Interesting stuff.

[1]https://en.wikipedia.org/wiki/Linguistic_prescription

Thanks, I didn't see that.

What part of your process is the bottle neck? Are you rendering the posters server side?

It looks like your page is down, possibly due to HN load.

What about them was terrible?

In fact antimatter15, who made ocrad.js, is a coauthor of tesseract.js! Javascript OCR is a small world.

This would be really cool! We would welcome a pull request and / or link to any downstream project working on this. In a similar vein, we've been considering adding the stroke width transform [0] as an optional preprocessing step.

[0] https://www.microsoft.com/en-us/research/publication/stroke-...

Yep, this is completely client side :)

You can even host the external language files yourself as described in the readme: https://github.com/naptha/tesseract.js#local-installation

Tesseract.js uses webworkers in the browser.

If you want to use a real Promise, you can wrap the call to recognize in Promise.resolve:

  Promise.resolve(Tesseract.recognize(myImage)).then(result => console.log(result))

That's interesting! Given that Tesseract.js wraps an Emscripten'd copy of Tesseract, I would have expected close to identical performance. This might have to do with the way we threshold images, with the age of the tesseract version we're using, or both. I'll look into it!

Edit: In addition to those differences, I think your font size is still a bit too small. On an unedited screenshot from a macbook (https://i.imgur.com/iv4ZdSt.png) I get

  Although Google's API is certainly better, Tesseract.js should work similarly if you increase the font size. Screenshots taken on 'retina' devices are around the smallest text it can handle well.

  Edit:

  A screenshot of the same text at a higher resolution: https:[[imgur.com[a[W7IGu

  Tesseract.js output: https:[[imgur.com[a[niIfM

  “I am also a top professional on thumbtack which is a site for people looking for professional services like on gig salad. Please see my reviews from my clients there as well"

Upscaling worked ok for me:

Upscaled image: https://imgur.com/a/4IQA7

Result on demo page: http://imgur.com/a/A0v5C

The hammerman Tikes flsosushsath: Greetings. My name is Tikes Leafsilk.

You: Rh. hello. I'm Stasbo Murderknower the Craterous Trance of Fins. Don't travel alone at night. or the bogeyman will get you.

You: Tell me about this hall.

Tikes: This is The flccidental Ualley. In 123, Stasho Steamdances ruled from The flccidental Ualley of The Council of Cobras in Ueilapes.

...

Although Google's API is certainly better, Tesseract.js should work similarly if you increase the font size. Screenshots taken on 'retina' devices are around the smallest text it can handle well.

Edit:

A screenshot of the same text at a higher resolution: https://imgur.com/a/W7IGu

Tesseract.js output: https://imgur.com/a/niIfM

"I am also a top professional on thumbtack which is a site for people looking for professional services like on gig salad. Please see my reviews from my clients there as well"

Did you try increasing the font size a bit? On a retina macbook (so effectively ~2x bigger font) I get:

Excited about this... but the OCR quality seems to be very bad. Maybe it's not optimized for recognizing black text on a white background. For example, I took a screenshot of this comment and ran it through the demo and got this:

As another commenter mentioned, Tesseract.js won't perform very well on 'natural' images (e.g. the very light text you tried).

It should work better if you feed it a screenshot of the black text at the top of the demo page though (Tesseract.js is a pure Javascript port etc...).

It's a wrapper around an Emscripten port of that library. See https://github.com/naptha/tesseract.js-core

Thanks! Fixed. The actual link is https://github.com/naptha/tesseract.js/blob/master/docs/tess...

What browser / OS are you on?

Edit: this affected every browser because it was a typo. Fixed!

What search engine are you using? I see figma at #1 at https://www.google.com/search?q=figma .

In https://github.com/bijection/g9/blob/master/src/Data2Rendera... each shape gets a unique id.

Dragging a shape calls the 'desire' function in https://github.com/bijection/g9/blob/master/src/index.js , which runs gradient descent with something like the shape's distance from the mouse as the cost.

MIT. Added to the repo.

First, you make up some names and give each name a number. Then you draw shapes based on the names, not the numbers.

For example, you might make a number called my_cool_number, and set it to 10, and then say "draw a circle my_cool_number pixels from the top, and 300 pixels from the left.".

Then your screen shows a circle 10 pixels from the top and 300 pixels from the left.

If I go and drag the circle, g9 will figure out how to change your numbers. For example, if I drag it down, g9 might say "now my_cool_number is set to 100". If my_cool_number is 100, then "draw a circle my_cool_number pixels from the top, and 300 pixels from the left" will draw a circle that's lower down.

That way it's like I'm actually moving the circle.

The cool part is that if you had other parts of your drawing that also depended on my_cool_number, they'll also move when it changes.

Thanks! Fixed.

Sorry about that!

I can't reproduce this on the most recent ff on Mac OSX, what operating system are you using?