Nexus a Linguistic Technique for Precoordination

dete rmina t ion , a sequence rule goes into effect which combines t e r m s based on the i r syntax. A var ie ty of co rpo ra was used to tes t and develop the NEXUS precoord ina to t . Data bases consis t ing of legal information, computer p rog ram d e s c r i p t ions and NASA l inear tape s y s t em documentation were used. More variety was p r e s e n t in the NASA documents which made the re su l t s of the application of NEXUS to this collection m o r e significant than the o thers . Also, a fuller ba t te ry of ru les was developed by this t ime, increas ing the power of the p r o g r a m . S u m m a r y NEXUS is a r e s e a r c h project which is concerned with input p r o c e s s i n g of natural language for information re t r ieva l . The com pu t e r p r o g r a m used to do this task cons i s t s of l inguist ic rules that opera te on the suffix por t ions of printed words , and the o rde r of these words as they appear in a sentence. NEXUS accepts l i s ts of index t e r m s that have resul ted f rom the appl ication of an au toindexer p rog r a m to titles and abs t r ac t s . Thcsc t e rm l is ts a r e p r o c e s s e d by NICKUS in o rde r to form combinat ions of t e r m s which a re J s t o r e d as subject headings. Such subject headings o r precoord ina t ions aid the s e a r c h e r in finding information when they a re used in a bibliographic pr in tout . As opposed to coordinateindexed pr intouts , consis t ing of l i s t s of individual t e r m s and the access ion numbers of the sou r ce documents, those pr in tou ts of NEXUS-processed t e r m s contain word combinat ions that have been precoordina ted , saving t ime and increas ing accuracy for the s e a r c h e r . It m u s t be s t r e s s e d that NEXUS opera tes on genera l ru les . There a r e o c c u r r e n c e s in language that a re not covetable by this method. Storage by individual t e r m s is effected in conjunction wRh NEXUS so that nothing is m i s s e d because of ru le exceptions. Compar ison tes ts have been run us ing the full NEXUS p r o g r a m , a par t ia l application of the p r o g r a m using sequence ru les (SEQS), and human analys is of the s a m e data. Although falling shor t of human analysis in some r e spec t s (except for consistency), the NEXUS approach is more effective than SEQS in producing effective combinat ions. Although some suggest ions a rc made for applying this technique along with a poss ib le output fo rmat fo r a bibliographic application, the chief value of this effort, however, has been to fu r the r study those aspects of language that are amenable to computer ized analysis for the pu rpose of improving input and output functions in informat ion re t r ieva l .