Show HN: alltech.io, My Tech Discovery Project

2 points by cpayne624 ↗ HN
I tried to build a discovery-focused tech aggregator, where incoming articles are auto-evaluated for mentioned topics and then those topics are used to expose the user to other relevant people & concepts.

It'll be in perpetual enhancement but would love to hear what you think, especially any suggestions for improvement. Thx -c

http://alltech.io

4 comments

[ 2.9 ms ] story [ 21.6 ms ] thread
definitely an interesting site. looks a little like rebel mouse but for tech in general.

What are your plans with it?

More just a labor of love, at least @ this point. Thanks -c
Are incoming articles just pulled via RSS and then a sorting algorithm decides what is displayed?
Here's the basic (scheduled, async) process...

* background threads continuously check source RSS/ATOM for new entries, respecting 304s (via TTL/ETAG/LASTMOD headers if provided); there's also a configurable throttle to limit the number of times any particular source will be checked, irrespective of anything else

* when new RSS/ATOM entries are found, the _full_ article from the host site is retrieved & parsed and evaluated for keywords to figure out what main topics the article's about; lots of feeds are partial, so being able to grab the full article content (w/out the header/comment/nav/footer cruft) is necessary for it to work

* then the full article is checked for a "suitable" (size/dimensions) image; if one's found, it's pulled down, resized/optimized, and stored locally for display alongside the entry when browsing the site

* then the (full article) text is added to the search index; again, w/partial feeds the full text is necessary to make search useful

* the "research" function leverages the YouTube, Wikipedia, Bing, Reddit, Slideshare (although I may be yanking this out soon; hella-slow responses sometimes), and Amazon APIs.

For display, the "grid" lists all entries in reverse chrono that had a suitable image. The "index" lists every entry.