ISR10 Introduction chapter Joseph John Rocchio Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. 1-4 In document retrieval, one does not necessarily desire to extract and represent the `1content" 0£ a document, but rather to characterize that content in a manner which can consistently lead to the recovery 0£ its primary representation, namely the document (the representation 0£ the natural language) T[OCRerr]e £irst £unctional aspect 0£ a[OCRerr] document retrieval system, there£4'[OCRerr]re, involves the means £or representing or characterizing the in£ormation content 0£ source documents. [OCRerr]aditionally, this is the process 0£ subject indexing. Use£ul re£erents to documents in a retrieval system may be indicative 0£ attributes other than in£ormation * content. In particular, re£erents such as the author's name, *.publication date, journal or publisher identi£ication, cited 10 re£erences, etc., can be use£ul in several contexts. For the current *purposes, however, those re£erents not[OCRerr]directly indicative 0£ in£ormation content'will be ignored with the understanding that their * `practical use£ulness to the retrieval process as a whole nuist be considered i[OCRerr] special circumstances. Chapter 2 0£ this. study considers the role 0£ indexing in document retrieval systems. The index £unction is discussed in terms 0£ its &oals, as well as in terms 0£ the linguistic aspects 0£ its mechanization, and 0£ the possibilities 0£ optimization'6£ automatic indexing techniques. The second £unctional aspect 0£ the retrieval system, [OCRerr]hat is the search request £ormulation, is primarily a user £unction. In the broad sense it is also a system ,£unction in that a retrieval system includes the user. In a narrower sense, however, when the system is I