Conceptual Issues

• What are good measures of effectiveness of retrieval systems?
• Explain the need for average precision/recall
• Plausibility of Foraging and Information Diet model.
• Indexing with phrases rather than single words.
• Shannon information and entropy
• N-gram and N-phone (when are these helpful)
• Inter-judge reliability for relevance.
• Why is relevance feedback different from just re-running the same query.
• Probabilistic model
• Inference net
• Text similarity vs. visual similarity
• tf-idf for summarization
• Features of textures (smoothness, periodicity, directionality)
• Explain how ``eignefaces'' works
• Use of parallel corpora in cross-language retrieval
• Query translation versus document translation
• Supervised versus unsupervised learning
• Relation between machine translation and cross-language retrieval
• Use of parallel corpora in CLIR
• Statistical methods for classification
• Adaptive profiles for filtering (implicit vs. explicit feedback)
• Processing documents images directly versus doing OCR
• Issues for IR on OCR
• Issues for duplicate document detection
• Motion detection in video, camera motion
• Scene and shot detection in video (identify some factors)
• Browsing video with VIBE (see w14 reading)

Basic Equations and Algorithms: Be able to work simple problems with the following

• Calculate Precision/Recall
• Basic statistics
• Use Bayes Rule
• Simple string matching.
• Vector model applied to color matching (use cosines on color bins)

Algorithms Students

• Evidence in support of foraging model
• Blob-world
• Simple probabilistic model (like homework)
• Coding (p.180 MIR)
• R-Tree