Hacker News Reader

Comment by ujjwalg | Hacker News Reader

I think the concept is very intriguing and if what you say is what is on your site and you have actually made it by collecting all the information in the last 3 years from all the major news websites, I think you will end up being bought pretty soon. Rather, what you should do is patent your process asap, if you haven't done it already and then license it. Amazing and good luck.

View on HN

The patent application is in the works. As far as the the data set - we definitely have it. In fact we were thinking of releasing it under some type of open license (i.e. creative commons) after the site gets some traction. In terms of search engines - we ARE in fact a search engine. Just type something in the box at the top and you can also use the available 4 score to filter rank search results. So the search works just like google but you get to sort the results by Controversial, Popular, Engaging, and Deceptive. Some pretty interesting combinations can be created using the scores. Like for instance, you can search for "paris hilton" but you are interested in her scandalous "achievements" rather than her community work. Well, then you search for "paris hilton" and sort by Controversial. Google can't give you that - i.e. sort results based on what impact are they likely to have on people. Thanks for suggestions!

I tried it for a couple of keywords with different ways of sorting and it seems to be working great.

My personal feedback for this would be: to make it sticky you need to have something similar to what google news webpage looks like and have 3 sections (controversial, popular, engaging) in every section. And then you should have similar features (keyword news and number of news article in every section) and I will make it my homepage, no kidding. Currently, you are not utilizing the complete web space very efficiently.

Yes, this is what we were thinking of having eventually - separate sections based on score. The current design is something we threw together very quickly to get it out of the door. Btw, the article tags are clickable and run a search for that tag in the background. Also, you can filter news by domain. As we accumulate more data, we can start ranking domains based on the articles they have produced so far. Kinda like PageRank but not based on links. Finally, the algorithm can guess authoriship and cluster articles based on author as well. So again, we can rank authors/people after some time. This seems to be a much needed feature as this links discusses: http://threeminds.organic.com/2009/06/docs_are_old-school_we...

I think the above article came up on HN today.

Reply on news.ycombinator.com