About the Cookbook
The Daylife Cookbook is a collection of recipes to help you bootstrap applications that use the Daylife platform (DayPI).
These recipes come from Daylifers and the cookbook community, with entries ranging from sample code to tutorials, half baked ideas, and spotlights of the latest deployments of the DayPI.
Feel free to share, steal, criticize, applaud, and feed back!
Login to the Cookbook
Resources
Testimonials
Daylife is a perfect fit for our strategy of presenting the best and most tightly focused content, whether it is produced by us at USA TODAY or anywhere else on the Web. It helps us provide our readers with a full 360-degree view of a given topic, and adds depth and richness to niche areas important to our readers and advertisers.- Jeff Webber, SVP, USAToday.com
You guys have what seems like an amazingly powerful and easy to use API, we're pretty excited about playing with it and possibly integrating it with Hubdub. - Tom Griffith, Hubdub.com
It's amazing how simple this is to integrate into django. When a view is creating a page, it just calls a thin wrapper function that I put around the daypi module. - Bracket Boy
I found the Daylife API for PHP really easy to use, and coded something up quickly for my site, Brooklyn and Beyond - Brooklyn & Beyond
This tool makes it really easy to get a quick look or dive in deep thanks to our partner Daylife, who built a comprehensive service that yields up-to-the-minute results - The Washington Post
I was trying to make it work by using Google News RSS feeds but turned out impossible since they don't organize their data in the feeds. I now know its possible with the Daylife API - Akash Xavier
OpenPlaces travel news
We present a way to build feeds of travel-related news (e.g., local cultural/sport event, traveler blog, hotel strike, etc.) for specific destinations (city, state, country). We've used the Openplaces' travel ontology and traveliness guesser and mixed them up with the DayLife API to create a feed of travel-related news.
OpenPlaces travel news is demoed at: http://www.openplaces.com/news
The RecipeThere are six steps to this recipe:
- Choose a place (e.g., Montreal).
- Match the search term to a DayLife topic of type
place - Call
search_getMatchingTopicswith the search term. - Loop through topics and find topic with type equals to
place. - If DayLife topic and search term does not match, try these two backup plans, leveraging ontological data:
- Check for alternate spelling ("also known as") of the search term.
- Check for parent places.
- Get articles related to the topic using
topic_getRelatedArticles(Use thetopic_idfound in step 2 as parameter). - Get articles related to the topic from known travel sources.
- Use
source_getArticlesfor known travel sources (e.g., Travel Channel, Smarter Travel) - keep articles with matching topic using
article_getRelatedTopics. - Fetch web pages and rank them by "traveliness".
- Use a web page classifier, i.e., a trained classifier that knows the vocabulary of url, title, content, anchors, etc., and can distinguish between a Web page related to travel and one not related to travel.
- Select final news to the travel feed.
- Promote all articles with very high score in document classification.
- For all articles with medium score, get related articles using
article_getRelatedArticles. - If the mean score of related articles is also medium/high then it is safe to promote the article in the feed.
Relying solely on document classification tends to produces accurate but shallow result sets (i.e., high precision, but low recall). Through simple observation, we have noticed that direct classification will ignore some highly relevant results that have a medium traveliness score. Because these "medium scoring" items are lost in a sea of irrelevant news that has similar traveliness scores, lowering the acceptance threshold is not a viable option. The DayLife API helps us by providing a cluster of related news (called "stories"). By analyzing the mean traveliness score of articles related to the medium scoring articles, we are able to further differentiate them from the background noise, thus increasing the overall recall.
ConclusionWe've shown how the DayLife API can be mashed-up with ontological data and web classification technology to create a feed of travel-related news. We present the Traveliness Guesser in more detail in this White Paper.
Enjoy your next trip!
About Openplaces:Openplaces is a semantic search engine for travel. We are a startup based in Montreal, Canada. Visit www.openplaces.com to read more, and register for our upcoming launch.