HackU IIT Kgp 2013 BOSS + CA

BOSS around the web

Saurabh Sahni YDN Developer, Hacker, Evangelist

Souri DattaStructured Data Extraction Team

http://www.flickr.com/photos/sumrow/1267682594/sizes/l/

BOSS is Build your own search service

http://developer.yahoo.com/search/boss/

Provides APIs

To our Searchdatabase

TO BUILD your ownpowerful

Search applications

BOSS allows you to search over

Web, images, news & Blogs

What can be done on top of BOSS?

• Blend and re-rank search results

• Your own look and feel• Mix it with other APIs

BOSS Pricing

Free for building your hacks!!

BOSS uses OAuth for securityCode : https://github.com/sourind/hacku/

Get a FREE consumer key and

secret

http://hackyourworld.org/hacku/

http://developer.yahoo.com/yql/console/

3. Copy This url

1. Select yql query

2. Select output format

Finding images of “The Dark Knight Rises”

select * from boss.search where q="The Dark Knight Rises" and service="images"

and ck="..." and secret="..."

Finding “The Dark Knight Rises” in IMDB, movies.yahoo.com

select * from boss.search where q="The Dark Knight Rises" and

sites="imdb.com,movies.yahoo.com" and ck="..." and secret="..."

Spell Check and Correction

select * from boss.search where q="The Dirk Knight Rises" and service="spelling" and

ck="..." and secret="..."

Finding news on “The Dark Knight Rises”

select * from boss.search where q="The Dark Knight Rises" and service="news" and

ck="..." and secret="..."

Finding interesting objects:Content Analysis

select * from contentanalysis.analyze where text="Sachin Tendulkar is batting very well"

Content Analysis from a URL

select * from contentanalysis.analyze where url="http://www.cnn.com/"

Lets See it in Action!

Query Cheatsheet• Find images of “The Dark Knight Rises”• select * from boss.search where q="The Dark

Knight Rises" and service="images" and ck="..." and secret="..."

• Find reviews of “The Dark Knight Rises”• select * from boss.search where q="reviews

intitle:The Dark Knight Rises" and service="web" and ck="..." and secret="…"

• Search for Avatar but not the movie: • select * from boss.search where q="Avatar -

movie" and ck="..." and secret="... "

• Search pdfs of “The Dark Knight Rises”• select * from boss.search where q="The Dark

Knight Rises" and type="pdf" and ck="..." and secret="..."

Query Cheatsheet• Find all the news of “The Dark Knight Rises”• select * from boss.search where q="The Dark

Knight Rises" and service="news" and ck="..." and secret="..."

• Get long abstracts in the results• select * from boss.search where q="The Dark

Knight Rises" and abstract="long" and ck="..." and secret="…"

• Retrieve 51-100 results of the query• select * from boss.search where q="The Dark

Knight Rises" and start=51 and ck="..." and secret="... "

EXAMPLES

duckduckgo.com

Data Extraction

Why extraction is difficult?• Internet has lot of information• Not all can be processed by machines

– Unstructured data– E.g. DiscountedPrice and RedcudedPrice of a

product (both mean the same)

• Ultimate aim is to publish data in structured format

• Most simple way- xml,json

Web Scraping• Demo Dapper

More Resources• Yahoo! BOSS:

http://developer.yahoo.com/boss • BOSS Technical Documentation: http://

developer.yahoo.com/search/boss/boss_api_guide/

• Content Analysis : http://developer.yahoo.com/contentanalysis/

• Oauth sample code : https://github.com/sourind/hacku/

Questions??http://www.flickr.com/photos/reem_unique/4119729692/

• http://slideshare.net/souridatta

• https://github.com/sourind/

Thanks!!