Re:Show HN: Findsimilar.info - my weekend project
by chunkyslink on Monday November 08, 2010 @07:03AM
(#1882187)
I’m often interested in buying some book, film or piece of music based on someone I already know I like. The logic goes something like – ‘if I like this I will like something similar’. Now it is already fairly easy to do this online by using one of many websites providing this information but it is a distraction to do so an requires navigating away from the work I am doing engage in some website for a couple of minutes.
I wanted something quick and easy that would allow me to send off a request in less than 10 seconds then get back to my work. The information is now waiting for me when I’m ready to disengage from what I’m doing.
Its also command line-esque which I prefer than clunky interfaces.
I use Tweetie for Twitter which is always open and in less than 10 seconds I can have it open with a quick command line style query written and being processed while I get back to work.
I used a cron each minute to scrape twitter and search for the tweets to the @findsimilar account. These are then stored in a MySql table. Another script then runs over them and looks for the hashtag.
At this point we search a different part of Amazon depending on what is being searched for. The results are stored again in MySql tables (one for the page, one for the Artists / authors and one for the products).
I then used wildcard DNS entry for *.findsimilar.info and the same for apache so that we can use the format xxx.findsimilar.info which will be good for seo and also easy to guess.
In essence that is it. The biggest problem I can see is when someone is an author, artist and actor. We have a namespacing problem. Perhaps I need to add xxx.findsimilar.info/actor /author /artist
Re:Show HN: Findsimilar.info - my weekend project
by daystar on Monday November 08, 2010 @10:27AM
(#1882700)
would be great to have a web interface.. for 1. convenience and 2. privacy as i wouldn't want people that check my twitter timeline to know that i'm trying to find something similar to fill in the gap
Its also command line-esque which I prefer than clunky interfaces.
I use Tweetie for Twitter which is always open and in less than 10 seconds I can have it open with a quick command line style query written and being processed while I get back to work.
To give it a go have a look at http://www.findsimilar.info to see how it works and what you can search for.
I used a cron each minute to scrape twitter and search for the tweets to the @findsimilar account. These are then stored in a MySql table. Another script then runs over them and looks for the hashtag.
At this point we search a different part of Amazon depending on what is being searched for. The results are stored again in MySql tables (one for the page, one for the Artists / authors and one for the products).
I then used wildcard DNS entry for *.findsimilar.info and the same for apache so that we can use the format xxx.findsimilar.info which will be good for seo and also easy to guess.
In essence that is it. The biggest problem I can see is when someone is an author, artist and actor. We have a namespacing problem. Perhaps I need to add xxx.findsimilar.info/actor /author /artist