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. 1-7 category terms in the input text. B. Request Processing Search requests in the experimental system are introduced into the computer in the natural language with no £ormat restrictions. The index language representations 0£ the search requests are identical in structure with those 0£ the re£erence documents and are derived by applying thesame trans£ormation rules to the request text.. It will be assumed, in[OCRerr]general, that a search request is to be interpreted as a description 0£ a si[OCRerr]le topic area, i.e. a request describing'topics r'A11 and 11B" will be assumed to be satis£ied by documents dealing with A and 3, or with A in relation to B, etc. A user interested in documents either about A or about B is, by assumption, required to submit two search requests. The implications 0£ this assumption are discussed in more detail in Chapter 3. C. Angalar Distance [OCRerr]tching A retrieval operation in the model system is per£ormed by matching the index vector representation 0£ the search request with the index vector representations 0£ all re£erence documents. The range 6£ the matching £unction is assumed to introduce at least a partial order on the re£erence collection. Since the length or absolute magnitude 0£ an index vector under the assumed index trans£orma'tion is a £irst order £unction[OCRerr]o£'the length (number 0£ words) 0£ the text which it[OCRerr]represents, rather than 0£ the content