Using AI for keyword detection can be a tedious and slow process, it is also relatively computationally expensive. This too offers an alternative solution to training and modelling an AI model for such a task. Using a few algorithmic techniques, it is able to separate thousands of domain names per second into distinct words that appear in the domain as such:
"registeryourcar.com"=>register your car"cutecatvideos.net"=>cute cat videos
Compared to AI models that perform word tokenization to achieve a similar result, but only process 20-30 words per second. This tool provides a 3500% speed increase. With a tested accuracy of approximately 98%.
Through more optimization techniques, an extra 40% gain in speed was obtained by leveraging recursion.
Speed: the time it takes to separate joint words of the entire list of given domains and return the outputWord Accuracy: The ratio of the number of words guessed correctly out of the total number of words in all the given domainsDomain Accuracy(More Strict): The ratio of the number of domains whose words were guessed with perfect accuracy
| Domains No. | Speed | Word Accuracy | Domain Accuracy |
|---|---|---|---|
| 1000 | 1.6s | 98.63% | 96.7% |
| 10,000 | 14.1s | 98.61% | 97.07% |
| 100,000 | 2.1m | 98.63% | 97.12% |
The web app UI for small scale tasks (only a couple of thousand domains) is a nice addition to easily and quickly convert a relatively small list of domains into separated words and download an Excel file with the results!
domain-name-detector-web-app.webm
- Download the code, extract and open the folder.
- Open a terminal and go to the folder.
- Run this once when you first install:
npm install- Now to start the website whenever you need to, run:
node dist/index.js- Open the url:
http://localhost:8080- Done!