Newsletter #014

My 3 Data For Good Finds

Hello and welcome back!

I’m legitimately really excited to bring this newsletter to ya’ll. Accessibility and inclusivity might sound like boring topics, but as part of ignosis’ mission of helping bring data to all, it’s extremely important!

Also, the items that I found I think are really cool examples of what we can do with the combination of data and technology.

As you go through the finds below, think about the following;

  • How can these finds be useful to you personally and professionally?

  • Is the transparency in these finds something that can help build your businesses legitimacy?

  • How can you pull creativity from these finds and implement creative solutions for your users, co-workers, etc.


1 - Fingerspelling with Machine Learning

This one blew my mind away.

Fingerspelling will use your webcam and hand recognition technology, matched with machine learning to help teach you how to sign letters.

The hand recognition technology is able to differentiate your fingers, joints, and palm, and the machine learning aspect is used to identify how different sized hands can make signs that humans would recognize.

You have to try it to see how incredible it is. I believe learning to sign by doing is THE way to learn how to sign, and this program does a great job of doing just that.

And, in case you’re wondering. The processing happens on your local browser, so your information or video feed isn’t going anywhere.

I hope this program expands to teach how to sign different words and sentences.


2 - Common Voice

Remember when Siri first came out, and you’d ask a question like,

“Hey Siri, show me pictures of the Eiffel Tower.”

and Siri goes,

“Ok, here are pictures of waffle towers.”

This happens, and still happens, partly because these assistants can not tell the differences between our dialects and pronunciations.

You say to-may-to and I say to-mah-to, but we’re both saying we want food.

Introducing Common Voice.

Common Voice is an open source, publicly available dataset of different voices reading a large set of sentences.

The complete dataset is offered for free so that folks building and training machine learning models can have a diverse dataset to build voice applications around.

You can also contribute to this free data set by either reading sentences in your own voice, or listening to sentences and confirming or denying the accuracy of the sentence you heard. (This one is fun because sometimes you hear people speaking in exaggerated, but clear accents).

They so far have 27,000 hours of voices recorded, and 17,000 hours of validated recordings.

Hopefully the dataset continues growing and Google Assistant can finally understand when I tell it “Order a shawarma Halal Guys from Doordash.”


3 - Sleuth

Sleuth was actually co-founded by a friend who I used to work with in the past. I like to shamelessly plug all my friends in case they become millionaires one day.

The singularly greatest thing about the internet is the collective knowledge being pooled into one resource. While misinformation is an issue, we still have access to a lot of factual knowledge through sites such as YouTube, Reddit, and heck ignosis.co?!

Sleuth was developed to crowd source information from parents about children symptoms and experiences. Using that information to help other parents whose children might be going through something similar.

While crowd sourcing knowledge is nothing new, what I particularly respect about Sleuth is their level of transparency about the data that they have.

If you look into their About Slueth’s Data section, they break down what demographic the contributing population makes up. This would usually be information that is locked up as it could create a bias in the mind of the user.

“I’m not represented in this dataset.”

Could be a real thought for users of this data, but Sleuth wants to do right by it’s users so that they can make the most informed decision when using the data they are providing.

What other companies do you wish were as transparent with their data? Should this be used as a playbook for any company that deals with data knowledge transfer?


Would love to hear more about YOU my audience! Is there anything in the data world that could use more clarification? What information can I provide to help you utilize data to it’s fullest?

Respond to this email and I’ll curate my newsletters to fit your needs.

Do you or your business need help with your data? Let’s TALK!

Missed past newsletters? Catch up on those great resources HERE!

Thank you and be good,
Irfan - Founder

Previous
Previous

Newsletter #015

Next
Next

Newsletter #013