CoXML: A Cooperative XML Query Answering System

1
CoXML A Cooperative XML Query Answering System

• Shaorong Liu and Wesley W. Chu
• APWeb/WAIM 2007

2
Motivation

• XML has become the standard format for
  information representation and data exchange
• XML schema is usually very complex
• E.g., the XML schema for the IEEE Computer
  Society publications contains about 170 distinct
  tags and more than 1000 distinct paths
• It is often unrealistic for users to fully
  understand a schema before asking queries
• Exact query answering is inadequate and
  approximate query answering is more desirable!

3
Our Contribution CoXML
4
Roadmap

• Introduction
• Background
• CoXML
• Related Work
• Conclusion

5
XML Query Relaxation Types

• Value relaxation enlarging a value conditions
  search scope
• Node relabel changing the label a node to a
  similar or a more general label by domain
  knowledge

1 Tree Pattern Relaxation (S. Amer-Yahia, et
al., 2000)
6
XML Query Relaxation Types

• Edge generalization relaxing a / edge to a
  // edge
• Node deletion dropping a node from a query tree

7
XML Relaxation Properties

• Definition
• Relaxation operation an application of a
  relaxation type to a specific query node or edge
• Lemma
• Given a query tree with n applicable relaxation
  operations, there are potentially up to 2n
  relaxed trees
• Possible combinations

8
XML Query Relaxation Challenges

• Query relaxation is often user-specific
• Different users may have different approximate
  matching specifications for a given query tree
• How to provide user-specific approximate query
  answering?
• A query with n relaxation operations has
  potentially up to 2n relaxed queries
• How to systematically relax a query?
• Query relaxation generates a set of approximate
  answers
• How to effectively rank the returned approximate
  answers?

9
CoXML System Overview
relaxation language
ranked results
query
Relaxation Engine
Ranking Module
results
relaxed query
Relaxation Index Builder
exact answers
query
CoXML
XML Database Engine
10
Roadmap

• Introduction
• Background
• CoXML
• Relaxation Language
• Relaxation Index Structure
• Ranking of Approximate Answers
• Experimental Studies
• Related Work
• Conclusion

11
Relaxation Language

• A relaxation-enabled query is a tuple T, R, C,
  S
• T tree-pattern query
• R relaxation constructs
• E.g., delete/re-label a node, generalize an edge
• C relaxation controls
• E.g., prefer/reject certain relaxation
  operations, use certain relaxation types, control
  relaxation orders, etc
• S stop condition
• E.g., the minimum of approximate answers to be
  returned

12
Relaxation Language Example
ltinex_topic topic_id"267" gt ltcastitlegt
//article//fm//atlabout(., "digital
libraries") lt/castitlegt ltdescriptiongt
Articles containing "digital libraries" in their
title. lt/descriptiongt ltnarrativegt I'm
interested in articles discussing Digital
Libraries as their main subject. Therefore I
require that the title of any relevant article
mentions "digital library" explicitly. Documents
that mention digital libraries only under the
bibliography are not relevant, as well as
documents that do not have the phrase "digital
library" in their title. lt/narrativegt lt/inex_to
picgt
13
How to Relax Queries?

• Naïve approach
• Generate all possible relaxed queries
  iteratively select the best relaxed query to
  derive approximate answers
• Exhaustive, but not scalable
• Observation
• Many queries share the same (or similar) tree
  structures
• Our approach relaxation index structure
• Consider the structure of a query tree T as a
  template
• Build indexes on the relaxed trees of T
• Use the index to guide the relaxations of any
  query with the same (or similar) tree structure
  as that of T

14
Relaxation Index Structure - XTAH

• XTAH
• A hierarchical multi-level labeled cluster of
  relaxed trees for a given query tree
• Building an XTAH
• Given a query structure template T, generate all
  possible relaxed trees
• Each relaxed trees uses an unique set of
  relaxation operations
• Cluster relaxed trees into groups based on
  relaxation operations and distances -- similar to
  suffix-tree clustering

15
XTAH Example for Template Structure T
gen(eu, v) relaxing the edge between u and
v del(u) deleting the node u
16
XTAH Properties

• Each group consists of a set of relaxed trees
  derived from similar relaxation operations
• The relaxed trees can be located efficiently
  based on the type of relaxation operation
• The higher level group in the XTAH yields lesser
  relaxation than the lower group
• Query can be relaxed to different level of
  granularities by traversing up and down the XTAH

17
Ranking of XML Approximate Answers

• Content similarity cont_sim(A, Q)
• An extended vector space model 2
• Structure similarity struct_dist(A, Q)
• Use tree editing distance for measuring structure
  similarity
• Propose a cost model that assigns operation cost
  based on relaxation semantics
• Overall relevancy sim(A, Q)
• A ranking model combing both content similarity
  and structure distance

? is a small constant between 0 and 1
2 Configurable Indexing and Ranking for XML
Information Retrieval (S. Liu, et al., 2004)
18
Experimental Studies

• Experiment Setup
• INEX (INitiative for the Evaluation of Xml) 05
  test collection
• Document collection
• Query set
• Gold standard
• Evaluation Metrics
• nxCG (normalized extended cumulative gain)
• the official evaluation metric used in INEX 05
• Given a number i (i?1), nxCG_at_i, similar to
  precision_at_i, measures the relative gain users
  accumulated up to the rank i

19
Retrieval performance improvements with semantic
cost model

• Query set all content-and-structure queries in
  INEX 05

nxCG_at_10 (?, cost model)

Assigning relaxation operation with different
cost based on the similarities of the nodes being
operated improves retrieval performance! nxCG_at_25
and nxCG_at_50 yield similar results
20
Evaluation of Relaxation Control

• Query topic 267
• Result

Relaxation control enables the system to provide
answers with greater relevancy!
21
Related Work

• Relaxation based on schema conversions (LC01,
  LMC01, LMC03)
• Without structure relaxation
• Native XML relaxation
• Proposed structure relaxation types e.g., KS01,
  ACS02
• Used the relaxation types ACS02 in our work
• Investigate efficient algorithms for deriving
  top-K answers based on relaxation types e.g,
  Sch02, ACS02, ALP04, AKM05
• Without relaxation control

22
Conclusion

• Cooperative XML (CoXML) query answering
• Relaxation-enabled query language allows users to
  effectively express the relaxed query conditions
  as well as controlling the relaxation process
• XTAH provides systematic query relaxation
  guidance
• Used both content and structure similarity
  metrics for evaluating the relevancy of
  approximate answers
• Evaluation studies with the INEX test collections
  validate the effectiveness of our methodology