Representation II: Automatic Indexing
This lecture gives an overview on Information Retrieval. It explains why documents are ranked the way they are. The lecture explains the most relevant ways for content representation: Automatic indexing and manual indexing. For automatic indexing, the frequencey of word is of special relevance and their influence on the weighting of term are discussed. The most relevant models are introduced. The session on evaluation discusses new metrics like the Normalized Discounted Cumulative Gain. The session of information behavior provides a brief overview and explains the relation to IR. The session on optimization mainly introduces term expansion and fusion methods. The session on Web retrieval is concerned with the quality aspects and gives a basic insight to the PageRank algorithm.