Query Languages and Query

Operation
Chapter seven

1 04/29/21
 Common types of query
A. Keyword-based querying
Queries are combinations of words.
The document collection is searched for documents that
contain these words.
 Word queries are intuitive, easy to express and provide fast
ranking.
 Here, the concept of word must be defined.
 A word is a sequence of letters terminated by a separator (period,
comma, space, etc).
 Definition of letter and separator is flexible; e.g., hyphen could be
defined as a letter or as a separator.
 Usually, common words (such as “a”, “the”, “of”, …) are ignored.

2 04/29/21
 B. Single-word queries
A query is a single word: Usually used for searching in
document images
Simplest form of query.
All documents that include this word are retrieved.
 Documents may be ranked by the frequency of this
word in the document.

3 04/29/21
 C. Phrase queries
 In this case a query is a sequence of words treated as a
single unit.
Also called “literal string” or “exact phrase” query.
 Phrase is usually surrounded by quotation marks.
 All documents that include this phrase are retrieved.
 Usually, separators (commas, colons, etc.) and common
words (e.g., “a”, “the”, “of”, “for”…) in the phrase are
ignored.
In effect, this query is for a set of words that must appear
in sequence.
Allows users to specify a context and thus gain more precision.
Example: “Information Processing for Document
Retrieval”.
4 04/29/21
 D. Multiple-word queries
In this case a query is a set of words (or phrases).
Two options: A document is retrieved if it includes:
 Any of the query words, or
 Each of the query words.
Documents are ranked by the number of query words
they contain:
 A document containing n query words is ranked higher than a
document containing m < n query words.
 Documents are ranked in decreasing order:
Those containing all the query words are ranked at the top,
only one query word at bottom.
 Frequency counts may be used to break ties among documents
that contain the same query words.
5 Example: what is the result for the query “Red Bird” ? 04/29/21
 E. Boolean queries
Based on concepts from logic: AND, OR, NOT
It describes the information needed by relating multiple
words with Boolean operators.
Operators: AND, OR, NOT
Semantics: For each query word w a corresponding set Dw
is constructed that includes the documents that contain w.
The Boolean expression is then interpreted as an
expression on the corresponding document sets with
corresponding set operators:
AND: Finds only documents containing all of the
specified words or phrases.
OR: Finds documents containing at least one of the
specified words or phrases.
NOT: Excludes documents containing the specified word
or phrase.
6 04/29/21
 Boolean Queries
Precedence: Order of operations
NOT, AND, OR
use parentheses to override precedence
process left-to-right among operators with the same
precedence.
Truth Table

P Q NOT P P AND Q P OR Q
0 0 TRUE FALSE FALSE
0 1 TRUE FALSE TRUE
1 0 FALSE FALSE TRUE
7
1 1 FALSE TRUE TRUE 04/29/21
 Examples: Boolean queries
1.computer OR server
Finds documents containing either computer, server or both

2. (computer OR server) NOT mainframe

8 04/29/21
 Penalizing documents
When interpreting queries, some models demote
or reduced documents that include keywords that
were not requested. For example:
Example: Assume the vector model with the
cosine measure and the simple case that both
documents and queries use binary values.
Consider these two documents and a query:
 d1 = (0,1,0,1,0), d2= (0,1,1,1,0), q= (0,1,0,1,0)
 sim(q, d1) = 1.0, sim(q, d2) = 0.82
 d2 is demoted because it includes an extra keyword
not requested by q.
In contrast, the Boolean model does not
“penalize” documents with extra (non-requested)
keywords
9 04/29/21
 Query Operations
Relevance Feedback &
Query Expansion

10 04/29/21
 Problems with Keywords
May not retrieve relevant documents that include
synonymous terms.
◦ “restaurant” vs. “café”
◦ “PRC” vs. “China”
May retrieve irrelevant documents that include
ambiguous terms.
◦ “bat” (baseball vs. mammal)
◦ “Apple” (company vs. fruit)
◦ “bit” (unit of data vs. act of eating)

11 04/29/21
 Query operations
No detailed knowledge of collection and retrieval
environment
 difficult to formulate queries well designed for retrieval
 Need many formulations of queries for effective retrieval
 First formulation: often naïve attempt to retrieve
relevant information
 Documents initially retrieved:
 Can be examined for relevance information by user,
automatically by the system
 Improve query formulations for retrieving additional
relevant documents

12 04/29/21
 Query reformulation
Two basic techniques to revise query to account for
feedback:
Query expansion: Expanding original query with new terms
from relevant documents.
This is done by adding new terms to query from
relevant documents.
Term reweighting in expanded query: Modify term weights
based on user relevance judgements.
Increase weight of terms in relevant documents
decrease weight of terms in irrelevant documents

13 04/29/21
Approaches for Relevance Feedback
 Approaches based on Users relevance feedback
Relevance feedback with user input
Description of cluster built interactively with user assistance
 Approaches based on pseudo relevance feedback
Use relevance feedback methods without explicit user
involvement.
Obtain cluster description automatically
Identify terms related to query terms
e.g. synonyms, stemming variations, terms close to query terms in text

14 04/29/21
 User Relevance Feedback
Most popular query reformulation strategy
Cycle:
 User presented with list of retrieved documents
After initial retrieval results are presented, allow the user to provide
feedback on the relevance of one or more of the retrieved documents.
 User marks those which are relevant
In practice: top 10-20 ranked documents are examined
 Use this feedback information to reformulate the query.
Select important terms from documents assessed relevant by users
 Enhance importance of these terms in a new query
Produce new results based on reformulated query.
 Allows more interactive, multi-pass process.
Expected:
 New query moves towards relevant documents and away from non-
relevant documents

15 04/29/21
 User Relevance Feedback
Architecture
Query Document
String corpus

Revised Rankings
IR ReRanked
Query System Documents
1. Doc2
2. Doc4
Query 3. Doc5
Ranked 1. Doc1
Reformulation 2. Doc2 .
Documents 3. Doc3 .
1. Doc1  .
2. Doc2  .
3. Doc3 
Feedback .
16 . 04/29/21
 Pseudo Relevance Feedback
Just assume the top m retrieved documents are relevant,
and use them to reformulate the query.
Allows for query expansion that includes terms that are
correlated with the query terms.
 Two strategies:
Local strategies: Approaches based on information
derived from set of initially retrieved documents (local set
of documents)
Global strategies: Approaches based on global
information derived from document collection.

17 04/29/21
 Pseudo Feedback Architecture

Query Document
String corpus

Revised Rankings
IR ReRanked
Query System Documents
1. Doc2
2. Doc4
Query 3. Doc5
Ranked 1. Doc1
Reformulation 2. Doc2 .
Documents 3. Doc3 .
.
1. Doc1 
.
Pseudo 2. Doc2 
3. Doc3 
Feedbac .
18 04/29/21
k .
Query Expansion Conclusions
Expansion of queries with related terms
can improve performance, particularly
recall.
However, must select similar terms very
carefully to avoid problems, such as loss
of precision.

19 04/29/21
 Thank You

20 04/29/21