graph search

14
Graph Search

Upload: chaitali-patel

Post on 21-Aug-2015

35 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Graph Search

Graph Search

Page 2: Graph Search

OutlineIntroduction to Graph SearchTraditional Search EngineGraph Search ConceptNatural Language SearchArchitecture of Natural Language InterfaceRanking Search Results

Page 3: Graph Search

What is Graph Search ?

• Graph Search is a semantic search engine introduced by Facebook in March 2013.

• It is designed to give answers to user natural language queries rather than a list of links.

• Can be used for,– Find more of the people, places and things you’re looking

for.– Discover new connections based on what others have

shared with you on Facebook.

Page 4: Graph Search

• The Graph Search engine is built upon highly structured data in the form of a graph, representing hundreds of types of nodes and thousands of types of edges.

• Facebook's technology for finding people and things that are socially connected.

What is Graph Search ?

Page 5: Graph Search

Traditional Search Engine [PPS]

• Search using Keywords.• Adding Typeahead features.• Nearby Tagging within Posts.• Location Tagging of photos and Post• But, keyword-based system has not been the best

choice.• For example, “friends Facebook” can mean “friends

on Facebook,” “friends who work at Facebook Inc.,” or “friends who like Facebook the Page.”.

Page 6: Graph Search

Graph Search Concept• The Facebook graph is the collection of entities and

their relationships on Facebook.• Entities are the nodes and the relationships are

the edges.• Every user, page, place, photo, post, etc. are nodes

in the graph.• Edges between nodes represent friendships, check-

ins, tags, relationships, ownership, attributes, etc. 

Page 7: Graph Search

Nodes and Edges

Page 8: Graph Search

Designing a Graph Search System

PPS and Typeahead

• Search entities based on their metadata.• Using entity name (title).

Graph Search • Extended capabilities to search based on the

relationship between entities.• Search over the edges between the

corresponding nodes, Using natural language as input for queries.

Page 9: Graph Search

Natural Language Search

Entity recognition and resolution

Lexical analysis

Semantic parsing

Page 10: Graph Search

Architecture of Natural Language Interface

Page 11: Graph Search

Ranking Search Results

• Graph Search uses a variation of the N-shortest path algorithm, an extension of Dijkstra’s algorithm, to solve the problem of finding the top K best parse trees.

• “EdgeRank” is the name commonly given to the algorithm that Facebook uses to determine what articles should be displayed in a user's News Feed.

Page 12: Graph Search

Message From Graph Search DevelopersGraph Search is one important pillar of

Facebook in our mission to make the world more open and connected. We are only 1% complete,

and we are looking forward to the rest of the

journey!

Page 13: Graph Search

References• http://whatisgraphsearch.com/• https://www.facebook.com/notes/facebook-

engineering/• http://searchengineland.com/• https://www.facebook.com/about/

graphsearch/privacy• http://www.socialmediaexaminer.com/

facebook-graph-search-review/

Page 14: Graph Search

Presented by :Chaitali Patel

Thank You..!!