We are big fans of Hacker News here at MonkeyLearn. We read HN on a daily basis, but not all of us have the same interests; some are more interested news related to programming, and others are more interested in news related to startup and business.
So, what if we can read Hacker News but only see relevant news related to our interests?
This is why we created My Hacker News, a demo that classifies Hacker News posts with MonkeyLearn, you can check it out here: http://hackernews.demos.monkeylearn.com
By using My Hacker News demo, users can see HN top news classified with categories like ‘programming’, ‘business’, ‘design’, ‘entertainment’ and more. The interesting thing is that you can filter HN news based on the category you are interested. For example, if you are only interested in programming, you just click on ‘programming’ tab and voilà; you see only HN posts related to javascript, python, docker, elasticsearch and other programming related content.
How we built My Hacker News?
We used MonkeyLearn to train a machine learning categorizer with categories like ‘programming’, ‘business’, ‘science’, and downloaded samples from relevant sub-reddits to do the training.
The plot at the top shows you the distribution of categories for each hour, and you can filter the news by category.
The actual classifier is public so anyone can use it with a free MonkeyLearn account: https://app.monkeylearn.com/accounts/login/?next=/categorizer/projects/cl_GLSChuJQ/
The source code can be found here: https://github.com/monkeylearn/hackernews
Every five minutes the app polls the HN API to categorize the latest submissions.
Although it is just a demo, we would love to hear your thoughts on it and how we can make it better. Any comment and suggestion is welcome!
Absolutely love the concept but it still needs some work! Article title “Virgin Media stores user passwords in plaintext” has a .814 probability of being Entertainment. This is clearly a Security article.
Thanks Adam! Glad you like the concept! Although we think that most of HN posts are classified corretly, we totally agree that there is definitely room for improvement. The good thing is that with MonkeyLearn, we can keep adding training samples to make this HN classifier even more accurate. We plan to keep working on this!
Great idea! Could you add an email coding category?
Thanks Lori! You can check out the source code of My Hacker News here: https://github.com/monkeylearn/hackernews . You can build your own classifier with MonkeyLearn so you can use the categories you want! Cheers