Error Detection, Recovery, and Repair in the Translation of Inorganic Nomenclatures. 1. A Study of the Problem

Abstract

The treatment of errors generated in the transmission of chemical information from humans to machines is examined. Every type of human-machine communication requires a translation subprocess to deal with the various possible representations of knowledge; the present study was designed to consider errors occurring in such a subsystem. Regardless of the model employed, translation of knowledge is performed over several successive stages; at each stage different types of errors may be detected. A method is proposed to classify these errors according to the stage at which they occur, thus facilitating the generalization of the process. An analysis of errors occurring during the translation of inorganic chemical names is also presented. The particular grammatical features of different nomenclatures would require a study of the errors appearing in the human-machine communication process in modern chemistry-oriented systems, a problem which has only been touched upon in the systems proposed to date. Attention here is centered on lexicographic errors, since that stage of the translation process is completely independent of the model employed and might thus be useful in a more general sense.

Error Detection, Recovery, and Repair in the Translation of Inorganic Nomenclatures. 2. A Proposed Strategy

“…In the case of chemical names, when the strings representing chemical substances in any of the established nomenclatures are subjected to a translation process in order to generate a different representation of the knowledge (be this representation abstract, symbolic or natural), the process may fail due to the appearance of lexical errors, errors due to the presence in the input string (the substance's name) of substrings (morphemes) that do not coincide with any of the terminal symbols or vocabulary of the source language (the nomenclature) …”

“…Here it can be clearly seen that we have a classical approximate matching problem where a pattern set Σ T has to be searched for the substring r ∈ Σ T which is most similar to the error string. The set Σ T is the lexicon, and in the present case comprises the terminal symbols or morphemes defined for the grammars of the nomenclatures, such that the string r is lexically correct …”

“…In the introduction and in the previous paper in this issue, the authors made reference to studies aimed at the calculation of edit distance for exact and approximate matching and the modifications proposed to “penalize”, with different weightings, differences due to insertion, deletion, and substitution of characters (transposition being usually regarded as a deletion and an insertion). , …”

“…Two types of algorithms have been developed to this end: one aims to find all occurrences of a pattern P in a text T with a limit of k differences ( k mismatches problem ); the second type attempts to find all substrings T‘ of a text T whose difference when compared with a pattern P is below a threshold k ( k differences problem ). Both algorithm types rely on measuring the edit distance in order to determine the similarity between strings; , in other words, the number of deletion, insertion, or substitution operations needed for the pattern and the text to be identical, as described in the first article of this series …”

Name=Struct: A Practical Approach to the Sorry State of Real-Life Chemical Nomenclature

“…The nomenclature used to describe chemical structures is a language, as others have noted. − A specialized one, surely, but one that is systematic and rule-based, with specific syntax and semantics that allow complicated terms to be built from simpler morphemes taken from an agreed-upon (but otherwise arbitrary) lexicon, just like English or German or Japanese or any other spoken language. Unlike English or the others, however, nobody speaks Chemist in day-to-day activities, and nobody grows up learning it.…”