Databases and Information Retrieval

1
Databases and Information Retrieval Lecture
1 Basics of Databases and Information Retrieval
Instructor Mr. Gautam Das University of Texas at
Arlington Email gdas_at_cse.uta.edu
2
Database
IR

• Data
• Collection of Documents Unstructured piece of
  information
• Follows Rank and Relevance query model
• Output is the document

• Consist of Schema
• Relational Model
• Data stored in form of tables
• Follow typical Query Model and Joins
• Output in form of tuples which are made of joins
  from one or more tables

3
Types of Queries

• Conjunctive Queries
• Car , Accident
• Will search for the word either Car or
  Accident.
• General Boolean Queries
• Car Accident Arlington
• Will Search for words Car and Accident
  but should not have word Arlington.

4
Retrieval Models of IR

• Boolean Retrieval Model
• Ranked / Relevance Retrieval Model
• One which is missing in databases

5
Parameters Used for Ranking in Typical
Information Retrieval System

• Parameter 1
• Occurrence and Frequency
• The number of times the specified word occurs in
  the document decides the rank
• The position it occurs at e.g. Title, Sub Title.

6
Parameters Used for Ranking in Typical
Information Retrieval System

• Parameter 2
• Proximity
• If two or more words are specified in the search
  string then the documents containing those words
  near to each other should be ranked higher.

7
Parameters Used for Ranking in Typical
Information Retrieval System

• Parameter 3
• Stemming
• Uses various verbal forms of word for seraching.
• E.g. Run gt Ran, Run over, Running
• Exact match of word should be ranked higher
• E.g. If the word info is searched then the
  document containing word infotech should be
  ranked after the document containing exact match
  as info.

8
Parameters Used for Ranking in Typical
Information Retrieval System

• Parameter 4
• Frequency across Documents
• The words like a, an, the etc. should be
  suppressed as more probability is that those are
  irrelevant as far as searching criteria is
  concerned.
• If we are searching for Microsoft Corporation
  then the specific word Microsoft is more
  important than the general word Corporation

9
Parameters Used for Ranking in Typical
Information Retrieval System

• Parameter 5
• Page Access Frequency
• If the page is accessed more number of times i.e.
  If the page is popular then it should be ranked
  higher
• This kind of ranking requires to maintain log
  about the frequency of page access
• Useful in case of systems which store News,
  Stories or readable articles.

10
Parameters Used for Ranking in Typical
Information Retrieval System

• Parameter 6
• Number of In-Links to the Page
• It is the number of times other pages on web are
  having links to the page be ranked.
• Again a parameter for deciding the popularity of
  a page.