ISR11 Operating Instructions for the SMART Text Processing and Document Retrieval System chapter M. E. Lesk Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. 11-17 Title and body differentiation is also performed by manipulating the weights. Usually, the weight associated with the body of the text is left at 1.0 and the weight of the title adjusted. However, to obtain only title processing, the user specifies BODYWT 0.0 and TITLWT 1.0. BODYWT x x is the weight associated with concepts fr[OCRerr]iL the body of the text. TITLWT x x is the weight associated with concepts from the title. These parameters apply to all sources of concepts (phrases, etc.), including the expansion concepts (hierarchy and concept-concept). LOGVEC this specification causes all weights to be either zero or one. Weighting proceeds normlly, except that after the c[OCRerr]utation of the weight, any nonzero weight is set to 1. causes the vectors to be printed for each document in the system. If concon or hierarchy expansions were specified, two printouts of vectors are made, one before and one after the expansion. In each printout the source of each concept and weight is shown in detail. 3.6. Request-Document Correlation Once the final set of vectors is formed, the request vectors are correlated against the document vectors. The documents with high correla- tions are considered to be the ca[OCRerr]puter-generated ttanswersl? to the queries. These may now be printed. They can also be compared with a set of "relevant'1 documents, obtained from some other source. SMART can automatically compare the relevant documents with its answers, and evaluate the performance of the computer system against the outside standard.