You should be able to answer each of the following questions.
- How might one implement a retrieval system with natural
language queries, ranking and relevance feedback?
- What are the sources for new terms in query expansion, and what
are the ways to screen or narrow down to find the best ones to add?
- What combinations of weighting factors and similarity functions
seem best?
- What is the interaction between retrieval models and their ability
to do ranking? query expansion? relevance feedback?
- What type of weighting is based on discrimination value?
- Which formula for relevance feedback, developed in connection
with the probabilistic model, seems to perform best?
- What is the effect of document length on choice of ranking
systems?
- What special data structures and algorithms are needed to support
an efficient implementation of ranking and relevance feedback?
- What data structures are most useful for recording information
relating to query expansion?