Browsing and Keyword Search in Relational Databases

1
Browsing and Keyword Searchin Relational
Databases

• B. Aditya
• Soumen Chakrabarti
• Rushi Desai
• Arvind Hulgeri

• Hrishikesh Karambelkar
• Rupesh Nasre
• Parag
• S. Sudarshan

http//www.cse.iitb.ac.in/banks/ Dept. of
Computer Science and Engineering Indian Institute
of Technology, Bombay
2
Motivation

• Why keyword search on databases
• Keyword queries very popular for Web search
• Query languages (SQL/QBE/..) not appropriate for
  casual users
• Form interfaces cumbersome, give limited views
• Key difference from IR/Web search
• Normalization splits related data across multiple
  tuples
• BANKS provides keyword search on databases,
  coupled with extensive browsing features

3
Basic Model

• Database modeled as a graph
• Nodes tuples
• Directed edges for foreign key, inclusion
  dependencies, etc
• Graph keyword querying is great for information
  integration
• can model relational, XML, HTML, .., data in a
  single graph

MultiQuery Optimization
paper
BANKS Keyword search
writes
S. Sudarshan
Prasan Roy
Charuta
author
4
Answer Model

• Answer Rooted, directed tree connecting keyword
  nodes
• Multiple answers possible
• Answer relevance computed from inverse edge
  weight combined with node weight (prestige)

paper
MultiQuery Optimization
writes
writes
author
author
S. Sudarshan
Prasan Roy
Eg. Sudarshan Roy
5
Answer Relevance

• Earlier model Tree score E 1/ (S edge weights)
• May need to follow edges backward to form answer
  tree
• Weight of backward edge u?v ? indegree of u
• New Tree-weight model based on spreading
  activation
• Path weight ? probability of following each
  edge
• Edge probability Forward ? 1/out-degree
  Backward ? 1/in-degree
• Tree score E (harmonic) mean of path weights
• Node weight (prestige) based on indegree
• More incoming edges gt higher prestige
• New Google PageRank style transfer of prestige
• Node weight computer using biased random walk
  model
• Node score N S log(leaf and root node weight)
• Overall score EN1/l

6
Other New Features in this Release

• Node selections
• e.g. optimization (yeargt1990)
• Node weight by proximity
• E.g. author (near recovery)
• Node prestige gt if close to multiple nodes
  matching near keywords
• Answer pattern selection node disambiguation
• Answer pretty formatting
• E.g. display relationship names (e.g.
  writes/written-by) instead of displaying
  linking-tuples in answer tree