I am building up a database of recommendations for interesting and off the beaten path activities people can do in various cities around the world...Europe, US, Asia, etc.<p>To speed up the process, as well as get content from places I have no friends in, I am thinking of posting a job on Amazon Mechanical Turk but I don't know about the quality I will get.<p>Do you guys think it will work? Or maybe there are some other better alternatives?
It depends...<p>If you are just looking for top 10 lists, it should be fine. If you want detailed reports like WikiTravel (<a href="http://wikitravel.org/en/Main_Page" rel="nofollow">http://wikitravel.org/en/Main_Page</a>), you'll have a harder time finding people who can write well for a low price.<p>Mechanical Turk is like any other system of inputs and outputs: what you get out of it depends on what you put in. From my experience, if you give very detailed instructions and examples of what you want, and you pay a decent reward, you'll get good results. If you want people to write 500 word articles for $0.05, you'll get crap.
You could start off with some automation i.e. make a crawler. Start with seeds from top rated places from Yelp, Gowalla, etc. Then, blast your queries across some travel site APIs and freebase.com to see what kinds of interesting places you get. Then, you can perhaps do model based collaborative filtering and do the recommendations automatically.<p>Or for inspiration... <a href="http://www.tripbase.com" rel="nofollow">http://www.tripbase.com</a>