Abstract
SAL stands for Semantico-syntactic Abstraction Language, a symbolic, second-order language to which natural language readily maps. SAL is organized as a taxonomy (ontology) of supersets, sets and subsets, allowing it to represent virtually any expressed thought, any sentence, at multiple levels of abstraction. To date, SAL has been developed for Engish and German languages only. In what follows we show only the English language variant. SAL is not a metalanguage but approaches one. Because nouns denote things and concepts that are virtually universal, SAL codes for nouns are generally applicable to any natural language. For the other open-class parts of speech (verbs, adjectives and adverbs), SAL is apt to be metalinguistic only at the SAL superset level.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
See Part II Postscript for discussion and illustration regarding Logos Model use of numbers. Note that SAL mnemonics are employed strictly for documentation purposes.
- 2.
Mnemonically 1 34 2 would be expressed as N(COfunc; pl).
- 3.
The negative number −1 represents the universal case, i.e., any WC, Type or Form. Thus, −1 −1 −1 would represent any element whatsoever. Negative numbers below −1 in the Type field function differently. They are addresses pointing to an array containing an ad hoc set of Type codes for the given element.
- 4.
See Postscript Part II for typical diagnostic output showing use of numbers.
- 5.
Some slight variation in SAL superset ranges occurs for certain WCs.
- 6.
All of the other systems tested handled this construction correctly.
- 7.
Sentence is an extreme example of POS misresolutions by Logos Model.
Author information
Authors and Affiliations
Postscript
Postscript
1.1 Postscript 9-A
Although Logos Model’s use of numbers is foreign to linguistic convention, the practice has proven to have many advantages computationally, especially with respect to the storing, indexing, accessing, and matching of patterns when expressed numerically. In SAL for example, superset codes generally range from 1 to 17 set codes from 18 to 99, and subset codes from 100 to 997. The benefit of this is that these numbers allow developers to instantly recognize a pattern’s place in the taxonomy, i.e., its degree of semantico-syntactic specificity. This number arrangement facilitates the internal practice of matching SAL input against stored patterns first on the basis of subset codes, then on sets, and finally on supersets, thus insuring priority to more specific pattern-rules. Finally, this arrangement allows pattern-rules to be self-organizing and rationally indexable. Developers do not need to ponder where to place a pattern-rule or where to look for it.
Figure 9.29, is an extract from a Logos Model diagnostic showing use of numbers. The diagnostic shown becomes available at the end of the second of Logos Model’s six processing stages, i.e., at the end of the macro-parse.
Rights and permissions
Copyright information
© 2018 Springer International Publishing AG, part of Springer Nature
About this chapter
Cite this chapter
Scott, B. (2018). The SAL Representation Language. In: Translation, Brains and the Computer. Machine Translation: Technologies and Applications, vol 2. Springer, Cham. https://doi.org/10.1007/978-3-319-76629-4_9
Download citation
DOI: https://doi.org/10.1007/978-3-319-76629-4_9
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-76628-7
Online ISBN: 978-3-319-76629-4
eBook Packages: Computer ScienceComputer Science (R0)