LBSC708A Test Notes

What is this class about?  A shared view of the most important points.

o Ranked Retrieval / Retrieval Models (Vector Space, Probabilistic, Boolean)
o Feature / Term Extraction / Segmentation / Stemming
o Evaluation Techniques / Recall and Precision / Desiderata
o System Models
o Machine Learning (Indirectly)
o Bag of Words / Terms
o Indexing
o Tradeoffs
o User Interface (QF, SI, EX)
o Summarization
o Social Filtering

All cross-referenced with:

o Cross Language
o Speech
o Video
o Document Images

Sample exam questions:

(45 minutes) Recent estimates suggest that the size of the World Wide
Web is now doubling about every five months.  Explain what effect this
rapid growth will have on the ability of search engines to give rapid,
complete, and accurate responses to user queries.

(3 minutes) How many angels can fit on the head of a pin?

(20 minutes) One criticism on the "bag of words" approaches to
information retrieval is that they don't capture evidence of
meaning that comes from the relationships among terms.  Explain the
meaning of "term" in the context of information retrieval and
explain how an appropriate choice of terms can help to overcome this
problem.

(25 minutes) How could collaborative filtering be used in
cross-language retrieval (If speakers of both languages are available
to provide user assessments).

(15 minutes) Discuss some of the ways IR systems process non-Latin
scripts and how this differs from processing of Latin characters.

(15 minutes) Identify the parts of the information retrieval model and
describe the relationships between them.

(20 minutes) Evaluate three methods of segmentation in text-based IR.
Which would be the best for use with multimedia retrieval, such as
video?

(20 minutes) Describe the advantages and disadvantages of each
retrieval model we have studied.