Translation of retrieval requests couched in a “semiformal” English-like language
Thomas E. Cheatham, Stephen Warshall · Communications of the ACM · 1962
~' l tremely sophisticated scientific theory (which, in a sense, it is), to be simplified where eireumsta~mes allow and where practicalities demand.I t is almost certain that the full complexity of the representation process could not tie mechanized; it is ahnost certain thal~ substantial portions can b,.l:xperiencc s~t,sgests that, evert without any meehanizaii(m of represeni.ation,results satisfacto W for some purposes are obtainable with a relatively short processing time and relatively untrained analysts.In a n y case, the a m o u n t of mechanization t h a t is desirable will vaEy with the requirements of the situation.We have not considered meehanizability for its own sake to be a particularly valuable aspect of arty retrieval language.I t seems likely that the level of analysis most fruitful for most pmt)oses will involve simultaneous, extended "conversations" between several analysts and a eomp uter.Other tools, in which syntactic and semantic information are stored, will certainly p l a y a p a r t in both document analysis and searching procedures within a complete retrieval system.To burden the content representatiou itself with all the context.ualovertones of a particular usage runs the risk of making the representation lose its relevance as conditions change.In the ftfllest sense, rio representation less than the d o c u m e n t itself in the hands of the user ean be guaranteed to be adequate indication of a d o c u m e n t ' s relevance in all contexts.W h a t a syslem of repres(,,~tation such as the above, together with its attxiliary tools, offers is a normalized expression of a document's content which is formally correct and materially adequate with respect to some given set of criteria, yet, which need ~ot he discarded as those conditions of adecllm