Show HN: Hacker News with categories and popularity (newshack.io)
Did you want to know topics of the news or popularity of posts?
"Hacker News Hack" helps you to find news based on some categories and importance or popularity.
It tagged news with 9 categories with a natural language processing technology and supervised learning for categorizing news.
It also calculates a speed of number of comments in each posts as popularity based on term, that is 1 hour, 1 day, 1 week, 1 month, or 1 year.
67 comments
[ 4.3 ms ] story [ 135 ms ] threadOur implementation and algorithm detail is followings.
Its categorizing process is written in Python.
Using nltk, it makes corpus with TFIDF model from HN topics and comments. And it generates classifiers from this corpus with SVM algorithm using scipy and numpy.
FYI, its web interface is written in Clojure and ClojureScript.
I trained classifiers with hand annotations (about 1000 contents or so)
But, humour is almost universally subjective. There are some aspects of Obama-Care that I thought were highly amusing, especially the "500 million lines of code" gag.
Here is a very serious article about the C programming language.
http://uncyclopedia.wikia.com/wiki/C_programming_language
Having a giggle does not have to be a frivolous activity.
FYI: I just noticed your time calculations aren't working. Search for "item.time_str" on your indexes.
It crawls HN topics every one hour. So news in our application would delay up to one hour.
That may help with analyzing comment content at least. Not so much with news rankings though.
You're right. I have limited frequency of crawling due to HN's access thresholds.
hnstream will help me. I will try it.
I have been forcing myself off HN lately. One thing that I miss the most due to this is the SHOW:HNs of really cool products and hacks that hit the front page. This gives me a great month-to-month access to those great stories at one place.
Bookmarked!
You might want to checkout my newsletter which includes the best "show hn" links each week - http://hackernewsletter.com. It also helps with the HN addiction.
http://blog.gcdc2013-hackerio.appspot.com/what-is-hackerio
Another similar thing between us is that when we showed HN for our projects, we both got 0 comment from the HN community.
So the question is why they are not more welcomed than OP's NLP solution, even if we provide more tools to save and share?
Do you have any idea? If you are interested in further discussion, I can be reached at danmark.clara _at_ yahoo.com.
http://www.wired.com/insights/2014/02/search-today-beyond-op...
The direction is right because we all need to improve the web towards Semantic Web which is more meaningful. However, the first round of Semantic Web effort is dead: http://bit.ly/JP3KQO
Now it's the second round. Still the machine intelligence is not comparable with human intelligence. So what I believe is the way we are applying: organize the web with human interaction and automated it with some facilities.
I don't think we suck. It takes time for others to dive into this world. So I'll continue to make this effort and push forward. The problems for yours and mine are also pretty common: UI is not appealing, lacking of data.
I know you are running it for fun. But if there is any chance for us to work together, I think it's the best, because you understand this area and solution so well. Check out my failed Kickstarter project here http://kck.st/JNqv8z and give a try of my existing website: http://bingobo.com. I'll appreciate your feedback.
Also, I believe you haven't escaped the time field properly server-side, I see "item.time_str" where I imagine the item time should be dynamically displayed.
I'd noticed myself finding less and less to click on on the front page over the last year, but this has really driven home why.
If you haven't seen it yet, try http://hackermonthly.com/. I'm not a subscriber, but the free issues I've seen had more business stories in them. You can download the current issue from http://hackermonthly.com/issue-45.html
DEA redacts tactic that's more secret than parallel construction
Jamaican Bobsledders Ride Dogecoin Into Olympics
Swiss police: Screen in Tesla cars is too large
Ask HN: What will you pay for?
============
I guess maybe that last one could be somehow related but IDK. Seems that important things are missing from the categories (like entrepreneurship or running a business, which are not the same thing). In another comment on this thread, it was mentioned that these are auto-tagged - maybe they can train it a bit better. Those are, to me, off-topic for 2008 HN but right in the wheelhouse for 2014 HN.
The guidelines haven't actually changed any since then.
Other than that - good job. Thank you.
You're right. That's nice idea! I will add that feature shortly.
1. fix some bugs (e.g. "item.time_str")
2. support responsive design (currently optimized with iPhone)
* fix some stylesheet issues (support Chrome on Android)