Text Representations and Their Use

Conclusions

In this chapter we described the traditional forms of indexing descriptions and abstracts and their advantages and disadvantages. We also outlined the intellectual process of indexing and abstracting and saw that these cognitive processes heavily rely upon text structures and reoccurring word patterns to identify the content of texts. In chapter 1 we extensively elaborated the need and the increasing importance for systems that automatically produce useful and correct text representations in the form of indexing descriptions and abstracts. The use of the text representations in browsing, retrieval, and question-answering systems confirms this necessity.

In the next part we give a detailed overview of existing techniques for automatic indexing and abstracting. They comprise techniques for identification of key terms (natural language index terms) in texts, assignment of fixed descriptors (controlled language index terms) to texts, and methods for text summarization.