IR Homework

1
IR Homework 2

• By J. H. Wang
• Apr. 1, 2009

2
Programming Exercise 2 Query Processing and
Searching

• Goal to search for relevant documents
• Input a query
• (simple search keyword, Boolean)
• Output a ranked list of search results from
  Reuters-21578 collection
• (details to be described later)

3
Input User Query

• Simple Search
• Keyword
• Ex Malaysia, Nuclear,
• Free text
• Ex United Nations, Nuclear Submarine Fleet,
• Simple Boolean search
• Ex Israel OR Pakistan,

4
Output Ranked List

• A ranked list of search results from
  Reuters-21578 collection
• Term weighting scheme TF-IDF
• The ranking is done in vector space model, i.e.
  the cosine similarity between query and document
  vectors

wij (1 log tfij) log (N/dfi)
5
Example Output

• Ex
• Query Bangladesh
• Result ltDocgt ltsimilarity scoregt
• 244 0.85
• 443 0.67
• 413 0.3

6
Optional Features

• Optional functionalities
• Better user interface for query
• Complex queries phrase, substring, proximity
  search, combinations of Boolean operators,
• Stopword removal, champion lists,
  impact-ordering, tiered index,
• Different ranking/term weighting schemes
  variants of TF-IDF,
• They should be able to be turned off by a
  parameter trigger

7
Submission

• Your submission should include
• The source code (and optionally your executable
  file)
• A one-page description that includes the
  following
• Major features in your work (ex high efficiency,
  low storage, multiple input formats, huge corpus,
  )
• Major difficulties encountered
• Special requirements for execution environments
  (ex Java Runtime Environment, special compilers,
  )
• The names and the responsible parts of each
  individual member should be clearly identified
  for team work
• Due two weeks (Apr. 15, 2009)

8
Submission Instructions

• Programs or homework in electronic files must be
  submitted directly to the TA as follows
• Team members list please e-mail your team
  members list to the TA (t6598006 _at_ ntut. edu. tw)
  even if youre the only team member
• Preparing submission file one single compressed
  file named as ltTeamID-HWn.zipgt, for example,
  IR0901-HW1.ZIP
• Remember to specify the names of your team
  members and student ID in the files and
  documentation
• E-mail or online submission TBD
• If you cannot successfully submit your work,
  please contact with the TA or the instructor

9
Evaluation

• Some example queries will be submitted to your
  program, and the ranked list will be checked for
  effectiveness (recall and precision)
• Optional features will be considered as bonus
• Various query forms, weighting schemes, efficient
  scoring and ranking,
• You might be required to demo if the program
  submitted was unable to run by TA

10
Questions?