A grammar base question-answering procedure

The letters we have generated were drawn o~ a specific machine, the Stromberg.-Carlson 4020 microfilm printer. A. brief description of the SC 4020 will assist in using the rector letters with other machines. The SCC 4020 has two modes of operation. First, it can draw vectors, which can start at aw raster point on its 1024 X 1024 grid and extend up to 64 grid spaces in either or both x and y directions. Secondly, it can produce a total of 64 different characters by shaping the electron beam with an appropriate nmsk in the cathode-ray tube. One character is a dot. This mode allows one to construct shapes using closely spaced dots, or any other available character, as building blocks. In the type fonts described here, only the vector *~iode ~>i operatior~, was used. Measurement of the widtt~ o:l i.i~, vector indicates that it is equal to 2.3 grid spaces. 'fi~ ... means that a character which is 23 grid spaces high t~a:~ :~ resoluti<~ of only 10 vector ~idths. Conclusions The three fo~ts of letters presented here are the begi~ ning of a great variety of possible fonts and characte-> which will be numerically described and computer dra.~-;,~. The generality of the representation is clear from the ease with which the vectors can be adapted to other compu}-e>> and other cathode-ray tubes. We believe the fonts wi!! have great utility. The subject of this paper is a procedure for the automatic retrieval of certain segments of stored information, either explicitly or implicitly represented, through questions posed in natural language sentences. This procedure makes use of a sentence recognition device for the class of grammars which will correctly decide between the grammatical and ungram-matical sentences of a natural language. It is possible to make use of a recognition device of this sort for the following reason: Much data is fully expressible as a set of sentences in a natural language, a set which can be exhaustively and exclusively generated by a grammar. Based upon the rules of this grammar, a sentence recognizer will evaluate sentences , questions in the normal situation. Since the recognition function succeeds just in case the posed question is drawn from the set of sentences expressing the data, or, more correctly, is grammatical in terms of the grammar for this set of sentences, sentence recognition itself is a procedure for retrieving information. When the recognition …