You basically have an API that scans the input for specific Words, problem is as an example if you write "I like Poptarts" it is marked as an offensive Message because it contains "tart"
You can't just make it so easy for yourself.
As an example "ho" is detected as an bad word. So "How are you doing" and everything else with how or hope in it is detected as insult.
If I would need something like this I wouldn't buy this product, because of 3 things:
1. The API is detecting friendly messages, the mechanism behind it is generally extremely simple which doesn't mean its good. Also "kys" isnt detected.
2. Privacy. Also I would be commiting a crime if I redirect all private messages to your API without clearly telling the client that I redirect all messages to you (and intelligent users will using that service because they dont want their private messages being broadcasted to an weird API)
3. The price is way too high. I think you are just trying to make a few fast bucks, but the price is really just to big, also considering that the API does nothing else than checking a string for other insults out of an insult list, and even that isn't done good.
As an example:
Whatsapp has 1.3 Billion monthly users, and 55 Billion Messages per day, so every user sends approximately 42 messages a day.
So lets say I have a website with a chat function and I use your API to filter their messages
Each user sends 1260 messages a month, your cheapest plan costs 0.1599 cents per message, so every single active user would cost me 20 cents per month.
If I have a medium size application with 5k users, I would need to pay you 1000 Dollar for your service, that shouldn't be too much of a problem because I would probably lose most of my customers because they can't communicate properly without triggering this API and they don't want to share all their private messages with some weird API
Brilliant idea and great execution! However, the use of an illustration that looks like President Trump is just ironical. There's no need to politicize a very serious and genuine issue that's plaguing the internet.
@jittarao Absolutely right 😃 That was purely an incentive to click on our part. To be 100% honest, the image sort of represented what the API is solving but you are 200% right about that statement. We'll update the graphics to match the core values better :) Thank you kind sir
While the idea is impeccable and aims to solve a very real problem - the execution at it's best is less than mediocre.
For starters, this project should've first been launch in an alpha stage with a handful of testers before the pricing was outed.
Second, this is just negative word detection and does not inculcate sentiment analysis at all.
Most chat softwares and forum/message board softwares (wherein such an API would prove useful) include an option to include negative keywords that are banned rendering the use-case for such a third party service obsolete.
Secondly, as @altay_akkus pointed out - this does not even get close to identifying the word. It wildcards a string and then regexed the whole of it without taking into consideration word, spaces or carriage returns. Ultimately leading to terms such as "Who" and "How" being marked as insults too.
I'll not get into the Privacy and Pricing bit of it. As @altay_akkus covered it.
But if you're running a text analytics API against bullying, you should have some sort of ML in the background running at least very basic sentimental analysis. Something very basic that understands expressions or at least makes a basic attempt to understand connotations and innuendo.
// Example
Friend - Lost my dog today man, feeling too low!
Me - That sucks bro. I am sorry for your loss! It'll get better
//
Your API marks my reply as invalid, wherein I am just trying to tell a friend that I agree with his situation being down, but it'll get better.
TL;DR - The product is not ready to ship. A whole lot of issues going on.
I have seen your work and admire that you're trying to ship as many products as possible in 2018 - but unfortunately this one needs a lot of work. Right now, it's just a negative word detection API that too incomplete.
Take a look at projects like polarity or the Neutrino Bad Word Filter API or something like Textgain.
The Idea is good, but the implementation isnt.
You basically have an API that scans the input for specific Words, problem is as an example if you write "I like Poptarts" it is marked as an offensive Message because it contains "tart"
You can't just make it so easy for yourself.
As an example "ho" is detected as an bad word. So "How are you doing" and everything else with how or hope in it is detected as insult.
If I would need something like this I wouldn't buy this product, because of 3 things:
1. The API is detecting friendly messages, the mechanism behind it is generally extremely simple which doesn't mean its good. Also "kys" isnt detected.
2. Privacy. Also I would be commiting a crime if I redirect all private messages to your API without clearly telling the client that I redirect all messages to you (and intelligent users will using that service because they dont want their private messages being broadcasted to an weird API)
3. The price is way too high. I think you are just trying to make a few fast bucks, but the price is really just to big, also considering that the API does nothing else than checking a string for other insults out of an insult list, and even that isn't done good.
As an example:
Whatsapp has 1.3 Billion monthly users, and 55 Billion Messages per day, so every user sends approximately 42 messages a day.
So lets say I have a website with a chat function and I use your API to filter their messages
Each user sends 1260 messages a month, your cheapest plan costs 0.1599 cents per message, so every single active user would cost me 20 cents per month.
If I have a medium size application with 5k users, I would need to pay you 1000 Dollar for your service, that shouldn't be too much of a problem because I would probably lose most of my customers because they can't communicate properly without triggering this API and they don't want to share all their private messages with some weird API
Pros:-----
Cons:see the full text
SameTime.co
Toolhouse
Gist
VibeAtWork
Divjoy