ERROR-TOLERANT DESIGN IN COMPUTER-AIDED DISPATCH (FROM CARNAHAN CONFERENCE ON CRIME COUNTERMEASURES PROCEEDINGS, 1976, BY JOHN S JACKSON - SEE NCJ-57707)

• NCJ Number: 57726
• Author(s): B C SANGSTER
• Date Published: 1976
• Length: 5 pages
• Corporate Author: University of Kentucky
• Publication Format: Document
• Publication Type: Program/Project Evaluation
• Language: English
• Country: United States of America

Annotation

THIS PAPER PRESENTS A SOFTWARE MODULE DESIGNED TO SOLVE PROBLEMS CREATED BY ERRORS IN THE COMPLAINT ENTRY PORTION OF A COMPUTER-AIDED DISPATCH SYSTEM.

Abstract

ONE OF THE GREATEST SOURCES OF DIFFICULTY IN ANY AUTOMATED LAW ENFORCEMENT SYSTEM IS THE MAN-MACHINE INTERFACE. ERRORS MAY BE INTRODUCED BECAUSE OF FEAR AND EXCITEMENT ON THE PART OF THE COMPLAINANT, ORAL TRANSMISSION OF INFORMATION ALONG TELEPHONE LINES, UNFAMILIARITY WITH LOCAL NAMES, IGNORANCE OF THE CORRECT SPELLING, OR HASTE. PROMPT AND ACCURATE IDENTIFICATION OF THE INTENDED LOCATION IS ESSENTIAL, AS HELP MUST BE DISPATCHED QUICKLY; IN MANY SYSTEMS, THE MOST TRIVIAL DEVIATION FROM THE CORRECT FORM, SUCH AS A MINOR SPELLING ERROR, WILL RESULT IN A TOTAL ABSENCE OF COMMUNICATION. A SOFTWARE MODULE CAN ALLOW FOR AUTOMATIC INTERPRETATION OF A WIDE RANGE OF INPUT ERRORS, SUCH AS MISSPELLING AND STANDARD ABBREVIATIONS, AND PROVIDES THE CAPABILITY FOR AUTOMATIC INTERPRETATION OF FREELY FORMULATED MATERIAL. FOR EXAMPLE, IN DEALING WITH ABBREVIATIONS, OPERATORS MAY ABBREVIATE WHATEVER THEY WISH IN ANY REASONABLE MANNER, AND DEVISE ALGORITHMS TO IDENTIFY THE INTENDED NAME FROM THE INPUT ABBREVIATION; THIS REQUIRES ONLY SLIGHTLY MORE PROGRAM STORAGE. IN ADDITION, IF THE SOUNDEX ALGORITHM, BASED ON THE PRINCIPLE THAT SELECTED SETS OF THE LETTERS OF THE ALPHABET MAY BE USED TO SPELL WORDS THAT MAY BE MISSPELLED DUE TO PHONETIC SIMILARITIES, IS MODIFIED SO THAT IT CAN HANDLE A GREAT PROPORTION OF THE TARGET VOCABULARY, AND IF KEYS ARE INTRODUCED TO SUPPLEMENT THE SOUNDEX-BASED SYSTEM, IT CAN NULLIFY THE EFFECT OF MISSPELLING. AN INPUT ALGORITHM CAN BE DESIGNED FOR SYNTACTIC AND SEMANTIC ERRORS BY APPLYING STRAIGHTFORWARD TECHNIQUES FOR PARSING AND SEMANTIC ANALYSIS, A SYSTEM WHICH APPEARS TO BE THOROUGHLY PRACTICAL WITHIN THE CONSTRAINTS OF A MINICOMPUTER-BASED SYSTEM. IN THIS WAY, BOTH THE SPELLING-ERROR CORRECTION AND THE SYNTATIC AND SEMANTIC INTERPRETATION ROUTINES ARE HIGHLY MODULARIZED. IT IS SUGGESTED THAT THIS MODULAR, TABLE-DRIVEN STRUCTURE CAN PERMIT COMPARATIVELY STRAIGHTFORWARD RE-IMPLEMENTATION FOR OTHER NATURAL-LANGUAGE MEDIATED SEARCHES, SUCH AS FOR INFORMATION ON STOLEN PROPERTY OR WANTED PERSONS, OR FOR HANDLING LANDMARK INFORMATION AND NONSTANDARD ADDRESSES. REFERENCES ARE PROVIDED. (AUTHOR ABSTRACT MODIFIED--MHP)