Thoughts on Rails

Riding the Train of Thought

My Face When… #geek

leave a comment »

I've been having fun with using ; it is a brilliant concept and executed in a very simplistic manner. It works and it is fun.

However, it could be better. 

Here's a proposal to build an app that solves the problem of "expressing yourself better with photographs of verbs" in an intuitive way.

First, add some constraints:
  • all images must be scaled up/down to certain sizes (thumbnail 120×120, small 300×300, large 500×500)
  • prefer square images, provide a cropping tool
  • keywords come from dictionary only. allow crowd-sourcing keywords/tags for each image
  • autofill/autosuggest keywords (use NLTK/wordnet to help pick forms of verbs / synonyms like "frown/ frowning"
This causes one problem, that character names cannot be added. This can be addressed by letting people add character-names by clicking "add character name" instead of "add tag" – which is then moderated. This moderation list can have a whitelist and blacklist associated, with moderators having three buttons "add to whitelist" "not applicable / ignore" "add to blacklist" – which further automates this process and frees up moderators as the lists are populated.

These constraints make sure the database of images is clean and semantic.

Now for the interfaces. 

1. Adding / curating
  • people can add images, crop and center and tag them
  • people can add/flag tags 
  • people can browse the images based on various criteria (missing tags, few tags, popularity)
  • people can search based on tags
  • if possible, image-matching algorithm can avoid duplicates while adding
  • allow people to flag duplicates – in which case, the URL can be redirected to original
2. Using
  • browse entire catalog
  • search by keyword
  • sort by popularity
  • sort by newest
  • search "similar" (based on keywords, later, if possible based on image analysis)
  • click on a button to copy url to clipboard
  • click a button to send to twitter / fb etc.
  • When somebody searches for keywords and is shown an image, allow them to vote down a keyword for that image (say, somebody tagged an image as "angry" but the expression really is "sly"… voting down "angry" will either flag/remove the keyword or reduce its priority.
And finally URL scheme

1.  If a number is at the end of the url, the proper image is pulled out and displayed.
2/3. If text follows the url, it is parsed for keywords and the closest match is served – if more than one images match the keyword, one is picked up randomly. Searching for keywords should form this url too. Therefore:

Your feedback is welcome!

Written by hiway

March 12, 2012 at 5:27 pm

Posted in Uncategorized

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: