1 Query Language Baeza-Yates and Navarro Modern Information Retrieval, 1999 Chapter 4.

34
1 Query Language Baeza-Yates and Navarro Modern Information Retrieval, 1999 Chapter 4
  • date post

    21-Dec-2015
  • Category

    Documents

  • view

    216
  • download

    1

Transcript of 1 Query Language Baeza-Yates and Navarro Modern Information Retrieval, 1999 Chapter 4.

1

Query Language

Baeza-Yates and Navarro

Modern Information Retrieval, 1999

Chapter 4

2

Query Language

Keyword-based Querying» Single-word Queries» Context Queries

– Phrase

– Proximity

» Boolean Queries» Natural Language

3

Query Language (Cont.)

Pattern Matching» Words» Prefixes» Suffixes» Substring» Ranges» Allowing errors» Regular expressions

4

Query Language (Cont.)

Structural Queries» Form-like fixed structures» Hypertext structure » hierarchical structure

5

Structural Queries

(a) form-like fixed structure, (b) hypertext structure, and (c) hierarchical structure

(a) (b) (c)

6

Hierarchical Structure

An example of a hierarchical structure: the page of a book,

its schematic view, and

a parsed query to retrieve the figure

Chapter 4

4.1 Introduction

W e conver in th is chapter thedifferent k inds of .......

4.4 Structural Queries....

chapter

section section

title title figure

In troduction W e cover ......... S tructural... ......

in

figure

section

title

with

with

"structural"

7

The Types of Queries

Boolean queries

fuzzy Boolean

Natural language

structural queries basic queries

proximity

phrases pattern matching

errors

words substrings regular expressions

keywords and context prefixes extended patterns

suffixes

8

Query Operations

Baeza-Yates, 1999

Modern Information Retrieval

Chapter 5

9

Query Modification

Improving initial query formulation» Relevance feedback

– approaches based on feedback information from users

» Local analysis – approaches based on information derived from the set of

documents initially retrieved (called the local set of documents)

» Global analysis– approaches based on global information derived from the

document collection

10

Relevance Feedback

Relevance feedback process» it shields the user from the details of the query reformulation pr

ocess» it breaks down the whole searching task into a sequence of sm

all steps which are easier to grasp» it provides a controlled process designed to emphasize some t

erms and de-emphasize others

Two basic techniques» Query expansion

– addition of new terms from relevant documents

» Term reweighting– modification of term weights based on the user relevance judgem

ent

11

Vector Space Model

Definitionwi,j: the ith term in the vector for document dj

wi,k: the ith term in the vector for query qk

t: the number of unique terms in the data set

t

i

kijikj wwqdsimilarity1

,,),(),,,( ,,2,1 jtjjj wwwd ),,,( ,,2,1 ktkkk wwwq

t

k ktf

tf

itf

tf

ji

idf

idfw

jkk

jk

jkk

ji

1

22}{max

}{max

,

)5.05.0(

)5.05.0(

,

,

,

,

12

Query Expansion and and Term Reweighting for the Vector Model

Ideal situation» CR: set of relevant documents among all documents in the collection

Rocchio (1965, 1971)» R: set of relevant documents, as identified by the user among th

e retrieved documents» S: set of non-relevant documents among the retrieved documen

ts

RjRj Cdj

RCd

j

Ropt d

CNd

Cq

||

1

||

1

Sdj

Rdjm jj

dS

dR

qq||||

13

Rocchio’s Algorithm

Ide_Regular (1971)

Ide_Dec_Hi

Parameters = = =1 >

}|{ SddMaxdqq jjRd

jm j

Sdj

Rdjm jj

ddqq

14

Probabilistic Model

Definition» pi: the probability of observing term ti in the set of relevant do

cuments

» qi: the probability of observing term ti in the set of nonrelevant documents

Initial search assumption» pi is constant for all terms ti (typically 0.5)

» qi can be approximated by the distribution of ti in the whole collection

t

i ii

iiqijij pq

qpwwqdsim

1,, )1(

)1(log),(

iii

i

ii

iii idf

df

N

df

dfN

pq

qpwt

log)(

log)1(

)1(log

15

Term Reweighting for the Probabilistic Model

Robertson and Sparck Jones (1976) With relevance feedback from user

N: the number of documents in the collection

R: the number of relevant documents for query q

ni: the number of documents having term ti

ri: the number of relevant documents having term ti

Document Relevance

DocumentIndexing

+

-

+

ri

R-ri

R

N-ni-R+ri

-

ni-ri

N-R

ni

N-ni

N

16

Initial search assumption» pi is constant for all terms ti (typically 0.5)

» qi can be approximated by the distribution of ti in the whole collection

With relevance feedback from users» pi and qi can be approximated by

» hence the term weight is updated by

)(R

rp i

i )(RN

rnq ii

i

t

i i

iqijij n

nNwwqdsim

1,, log),(

t

i iii

iiiqijij rnrR

rRnNrwwqdsim

1,, ))((

)(log),(

Term Reweighting for the Probabilistic Model (cont.)

17

However, the last formula poses problems for certain small values of R and ri (R=1, ri=0)

Instead of 0.5, alternative adjustments have been propsed

)1

5.0(

R

rp i

i )1

5.0(

RN

rnq ii

i

)1

(

R

rp N

ni

i

i

)1

(

RN

rnq N

nii

i

i

Term Reweighting for the Probabilistic Model (Cont.)

18

Characteristics» Advantage

– the term reweighting is optimal under the asumptions of term independence binary document indexing (wi,q {0,1} and wi,j {0,1})

» Disadvantage– no query expansion is used

– weights of terms in the previous query formulations are also disregarded

– document term weights are not taken into account during the feedback loop

Term Reweighting for the Probabilistic Model (Cont.)

19

Evaluation of relevance feedback

Standard evaluation method is not suitable» (i.e., recall-precision) because the relevant documents used to rew

eight the query terms are moved to higher ranks.

The residual collection method» the set of all documents minus the set of feedback documents pro

vided by the user» because highly ranked documents are removed from the collection

, the recall-precision figures for tend to be lower than the figures for the original query

» as a basic rule of thumb, any experimentation involving relevance feedback strategies should always evaluate recall-precision figures relative to the residual collection

mq

q

20

Automatic Local Analysis

Definition» local document set Dl : the set of documents retrieved by a q

uery

» local vocabulary Vl : the set of all distinct words in Dl

» stemed vocabulary Sl : the set of all distinct stems derived from Vl

Building local clusters» association clusters» metric clusters» scalar clusters

21

Association Clusters

Idea» co-occurrence of stems (or terms) inside documents

– fu,j: the frequency of a stem ku in a document dj

» local association cluster for a stem ku

– the set of k largest values c(ku, kv)

» given a query q, find clusters for the |q| query terms» normalized form

||

1,,),(

D

jjvjuvu ffkkc

),(),(),(

),(),(

vuvvuu

vuvu kkckkckkc

kkckks

22

Metric Clusters

Idea» consider the distance between two terms in the same cluster

Definition» V(ku): the set of keywords which have the same stem form as ku

» distance r(ki, kj)=the number of words between term ku and kv

» normalized form

)( )( ),(

1),(

u vkVi kVj jivu kkr

kkc

|)(||)(|

),(),(

vu

vuvu kVkV

kkckks

23

Scalar Clusters

Idea» two stems with similar neighborhoods have some synonymity

relationships

Definition» cu,v=c(ku, kv)

» vectors of correlation values for stem ku and kv

» scalar association matrix

» scalar clusters– the set of k largest values of scalar association

),,,( ,2,1, tuuuu cccs ),,,( ,2,1, tvvvv cccs

||||,

vu

vuvu

ss

ssS

24

Automatic Global Analysis

A thesaurus-like structure Short history

» Until the beginning of the 1990s, global analysis was considered to be a technique which failed to yield consistent improvements in retrieval performance with general collections

» This perception has changed with the appearance of modern procedures for global analysis

25

Query Expansion based on a Similarity Thesaurus

Idea by Qiu and Frei [1993]» Similarity thesaurus is based on term to term relationships rathe

r than on a matrix of co-occurrence» Terms for expansion are selected based on their similarity to the

whole query rather than on their similarities to individual query terms

Definition» N: total number of documents in the collection» t: total number of terms in the collection

» tfi,j: occurrence frequency of term ki in the document dj

» tj: the number of distinct index terms in the document dj

» itfj : the inverse term frequency for document dj

jj t

titf log

26

Similarity Thesaurus

Each term is associated with a vector

» where wi,j is a weight associated to the index-document pair

The relationship between two terms ku and kv is

» Note that this is a variation of the correlation measure used for computing scalar association matrices

),,,( ,2,1, Niii wwwki

N

k ktf

tf

jtf

tf

ji

itf

itfw

kik

ki

kik

ji

1

22}{max

}{max

,

)5.05.0(

)5.05.0(

,

,

,

,

N

jjvjuvuvu wwkkc

1,,,

27

Term weighting vs. Term concept space

tfij

Term ki

Doc dj tfijTerm ki

Doc dj

t

k ktf

tf

itf

tf

ji

idf

idfw

jkk

jk

jkk

ji

1

22}{max

}{max

,

)5.05.0(

)5.05.0(

,

,

,

,

N

k ktf

tf

jtf

tf

ji

itf

itfw

kik

ki

kik

ji

1

22}{max

}{max

,

)5.05.0(

)5.05.0(

,

,

,

,

28

Query Expansion Procedure with Similarity Thesaurus

1. Represent the query in the concept space by using the representation of the index terms

2. Compute the similarity sim(q,kv) between each term kv and the whole query

3. Expand the query with the top r ranked terms according to sim(q,kv)

uqk

kwqu

qu

,

vuQk

quvqk

uquvv cwkkwkqkqsimuu

,,,),(

qk qu

vqv

uw

kqsimw

,',

),(

29

Example of Similarity Thesaurus

The distance of a given term kv to the query centroid QC might be quite distinct from the distances of kv to the individual query terms

ka kb

ki

kj

kv

QC

QC={ka ,kb}

30

Query Expansion based on a Similarity Thesaurus

» A document dj is represented term-concept space by

» If the original query q is expanded to include all the t index terms, then the similarity sim(q, dj) between the document dj and the query q can be computed as

– which is similar to the generalized vector space model

jv u

jvu

dkvu

qkqujvj

dkvjv

qkuquj

cwwdqsim

kwkwdqsim

,,,

,,

),(

),(

jv dk

vjvj kwd ,

31

Query Expansion based on a Statistical Thesaurus

Idea by Crouch and Yang (1992)» Use complete link algorithm to produce small and tight

clusters» Use term discrimination value to select terms for entry into a

particular thesaurus class

Term discrimination value» A measure of the change in space separation which occurs

when a given term is assigned to the document collection

32

Term Discrimination Value

Terms» good discriminators: (terms with positive discrimination values)

– index terms

» indifferent discriminators: (near-zero discrimination values)

– thesaurus class

» poor discriminators: (negative discrimination values)

– term phrases

Document frequency dfk

» dfk >n/10: high frequency term (poor discriminators)

» dfk <n/100: low frequency term (indifferent discriminators)

» n/100 dfk n/10: good discriminator

33

Statistical Thesaurus

Term discrimination value theory» the terms which make up a thesaurus class must be

indifferent discriminators

The proposed approach» cluster the document collection into small, tight clusters» A thesaurus class is defined as the intersection of all the low

frequency terms in that cluster» documents are indexed by the thesaurus classes» the thesaurus classes are weighted by

||

||

1 ,

C

wwt

C

i CiC

34

Discussion

Query expansion » useful» little explored technique

Trends and research issues» The combination of local analysis, global analysis, visual

displays, and interactive interfaces is also a current and important research problem