Indexing

Abstract

Once the processing tokens and other metadata that is associated with items have been identified, the system then needs to construct the searchable index that will be used with the user’s queries to return results. Before discussing the detailed alternatives, the concept of indexing is discussed along with what its objective. The automatic textual indexing techniques are discussed in the context of statistical, natural language and concept indexing approaches. Statistical is the major technique used in current systems and the major algorithms used are discussed (e.g., term frequency/inverse document frequency with item length normalization). Finally the indexing techniques associated with multimedia items (audio, image and video items) are described.