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. 1-5 desi[OCRerr]ed to react with the user so as to insure that a given search request becomes an e££ective representation 0£ the userts in£ormation needs'in terms 0£ the system1s capabilities, the request £ormulation process z[OCRerr][OCRerr]st be considered a critical system £unctione11'12 Chapter 3 develops this premise and considers various means £or optimization 0£ user search requests' in terms 0£ system parameters and the objectives 0£ the retrieval process as a whole. The third £unctional aspect 0£ document retrieval relates to the nature 0£ the matching criteria used to select source documents `in 13,14 response to a userts input query. In Chapter 4 the in£luence 0£ * the structure 0£ the in£ormation representations on the `matching criteria is developed. [OCRerr]jor emphasis is placed in this chapter on the relation 0£ the matching[OCRerr]£unction to document clas'si£ication and searchin[OCRerr],' and' in this context an automatic classi£ication al[OCRerr]orithm * is developed. This ai[OCRerr]orithm is s'peci£ically desi[OCRerr]ed to increase search e££iciency and `is shown to be applicable to a certain class 0£ matching £unctions. Chapter 5 considers several aspects 0£ the [OCRerr]eneral problem 0£. the evaluation 0£ document retrieval systems, particularly as they relate to the £unctional model. - In addition to examining the statistical basis £or' evaluation parameters, some novel measures are derived which have several advantages over those in current use. - Some 0£ the salient £eatures 0£ the S[OCRerr]ART automatic document retrieval system are presented in an appendix. The [OCRerr] system is used both as a concrete model, and as a sizrnliation device £or the