INFO624 - Week 4 Query Languages and Query Operations

1
INFO624 - Week 4Query Languages and Query
Operations

• Dr. Xia Lin
• Associate Professor
• College of Information Science and Technology
• Drexel University

2
Query

• Query is a representation of the users
  information needs
• It may not represent the information needs
  exactly because
• Information needs are difficult to describe --
  semantic difficulty
• Query must be in a format acceptable to the
  retrieval system -- syntactic difficulty

3
Content-based queries
Pattern Matching
word matching
Words Phrases Proximity
Prefix/suffix Wildcard search Error
handling Extended patterns
Boolean
Vector
Natural Language
4
Boolean Queries

• Request
• What are the likely problems when someone gets
  hurt on his knees when playing basketball?
• Write your best Boolean query for this request
• If the query returns zero hits, how do you modify
  the query?
• If the query returns too many hits, how do you
  modify the query?

5

• How does AskJeeves translate the request?
• What are the likely problems when someone gets
  hurt on his knees when playing basketball?

6

• Construct your best Boolean query for this
  request
• I am doing a research on personal space
  boundaries. I want to know if there are any sex
  or race differences in personal space boundaries.

7
Interaction with Queries

• Starts with a SEED query
• The System responds with a list of related terms
• Adds selected terms from the list to the query
• The system updates the list of related terms
• Repeat as needed

8
Example MedLine Search Assistant
9
Association-based Queries

• Find documents similar to this document.
• Find documents that links to this document
• Explicitly
• Implicitly

10
Field-based Queries
11

• Field-based queries will likely improve search
  precision.
• Field-based queries require that the data source
  has a fixed structure and are indexed by the
  structure.

12
Citation-based Queries

• Retrieve all documents that document A cites.
• Find all documents that cite document A.
• Find all documents that cite this author
• Find all document that cite both document A and
  document B
• Find documents that cites both author A and
  author B

13
Co-Citation

• The college has more than 20 years tradition on
  Co-citation research.
• Co-citation is the mentioning of any two earlier
  documents in the bibliographic references of a
  later third document.

Document 1
cites
Later Document 3
?
cites
Document 2
14
Co-Citation Analysis

• The count of mentions may grow over time as new
  writings appear. Thus, co-citation counts can
  reflect citers changing perceptions of documents
  as more or less strongly related.
• Documents shown to be related by their
  co-citation counts can be mapped as proximate in
  intellectual space.

15
Co-Citation Mapping

• Detects patterns in the frequency with which any
  works by any two authors are jointly cited in
  later works.
• Only recurrent co-citation is significant The
  more times authors are cited together, the more
  strongly related they are in the eyes of citers.

16
A Map of Information Scientists
17
AuthorLinks
18
Link-Based Queries

• Hypertext Structure
• Is a link a query?
• http//www.google.com/search?hlenqinformationr
  etrieval
• This is called query-mediated link.
• It is also called soft link.
• Is a query a link?
• Many pages are dynamically generated from a
  database or a search engine.
• Your review pages

19
Queries, Links, Is there a difference SIGCHI97

• An experiment was conducted to compare browsing
  behavior in query- and link-based interfaces.
  Results suggest that query-mediated links are as
  effective as explicit queries, and that
  strategies adopted by users affect performance.
  This work has implications for the design of
  information exploration interfaces.

20
Query Structure

• Hierarchical Structure
• What does the user want when searching for
  substance abuse
• We may not know, but adding narrower terms of
  substance abuse will likely get better results
• Alcohol Abuse
• Drug Abuse
• Alcohol-Related Disorders
• Amphetamine-Related Disorders
• Cocaine-Related Disorders
• Marijuana Abuse

21
Automatic Expansion

• If there is a defined hierarchy, several search
  strategies may be defined to expand the query
• Search with the query term only
• Search with the query term and all the terms in
  its upper hierarchy
• Search with the query term and all the terms in
  its lower hierarchy.
• Search with the query terms and its all the
  sibling terms

22
(No Transcript)
23
Query Operations

• Query execution
• Query expansion
• Query translation

24
Query Expansion

• Improve the initial query through automatically
• restructuring the query or
• adding other new terms or
• Adjusting weights of each terms.

25

• Restructuring the query
• Identify key concepts through natural language
  processing
• Identify any field information that may be
  contained in the query
• Is this an author?
• Is this a journal?
• Reverse term orders in the query

26

• Adding new terms
• Synonyms
• Hierarchical terms
• Scope terms
• Does query Football retrieve information on
  football or on soccer?
• Relevant terms
• Selected terms from relevant documents
• Terms co-occur most often with the query terms

27

• Adjusting term weighting
• If relevant documents are known, increase the
  weights for terms assigned to the relevant
  documents and decrease the weights to terms
  assigned to non-relevant documents.
• Adjust term weights in a topic tree
• Fruit
• Fruit, 0.9 apple, 0.7 orange, 0.7 banana,
  0.6 . Macintosh, 0.1 Computer -.4.

28
Query Translation

• From natural language to queries
• AskJeeves
• From queries in one system to queries in another
  system
• From one natural language to another natural
  language
• Altavista

29
Other types of representation for users needs?

• Mind-reading?
• Non-text queries?
• Gesture/motion?

30
IBM Visualization Space

• This information system understands the user.
• It "hears" users' voice commands and "sees"their
  gestures and body positions. Interactions are
  natural, more like human-to-human interactions.

31
Multimedia Queries

• Content-based
• Text indexing
• Attribute-based
• Color, size, type, time period,
• Structure-based
• Location, shape, layout, etc.
• Cluster-based
• Semantic groups, physical groups,
  structure-groups,
• Example find a photo that has the White House in
  the center.

32
Project Discussion

• Idea 1 Install and implement an IR system
• Focus on system and technology
• Need to have a collection
• Need to have hand-on experience with systems
• Idea 2 Conduct an evaluation experiment on one
  or two selected IR systems
• Focus on interfaces and users
• Idea 3 Customize an IR system
• Focus on functionality and customization

33
Project Evaluation

• Topics
• Relevance
• Problems identified
• Technical difficulties
• Solutions/ideas
• The process
• Design
• Implementation

34

• The report
• Background
• Written
• Oral

35
Midterm

• Concepts
• What is information retrieval?
• Data, information, text, and documents
• Two abstractions principles
• Users information needs
• Queries and query formats
• Precision and Recall
• Relevance

36
Midterm

• Procedures problem solving
• How to translate a request into a query?
• How to expand queries
• for better recall or better precision?
• How to create an inverted indexing?
• How to create a vector space ?
• How to calculate similarities of documents?
• How to match a query to documents in a vector
  space?

37

• Discussions
• Challenges of IR
• Advantages and disadvantages of Boolean search
  (vector space, automatic indexing,
  association-based queries, etc.)
• Evaluation of IR systems
• With or without using precision/recall.
• Difference between data retrieval and information
  retrieval