iOlite the AI language naturalisation programme that will make smart contracts accessible to non-coders
The potential uses of smart contracts are manifold: from simplifying insurance claims, to improving reporting on security ownership, a self-executing contract with a universally available data registry would streamline errors and inefficiencies across multiple sectors.
There is a gaping hole in the marketplace where easy, intuitive apps for non-coding professionals to design smart contracts should be. A few recent launches, like that of Fabric Token, profess to offer a platform where any commercial organisation can integrate automated process verification into its infrastructure. Such a contract would be contingent, say, on authorised parties giving their digital approval to a decision, which would trigger its execution.
However, this platform relies heavily on the use of pre-written templates, which are available to purchase for a price. The natural language translation machine under construction by iOlite is based on emerging AI technology which uses its proprietary Fast Adaptation Engine (FAE) on both coding languages like C, C++, Java and JS, and on for example legal syntax, underpinned by a Dependency-based Action Language (DAL) for each.
Using a logarithmic algorithm, the system maps possible alternative meanings of new phrases based on existing knowledge — within mathematically defined parameters — building its vocabulary. Through this process it is also able to abbreviate cumbersome coding instructions.
The aim is for translation to work both ways, so lawyers will be able to look at smart contracts and make adjustments to them without having to read and write code. The platform is based on an idea pioneered by a team of computer scientists from Stanford, outlined in a paper on ‘Naturalizing a Programming Language via Interactive Learning’[1], whereby different languages can be assimilated by the underlying DAL.
The applications of the iOlite platform, its creators believe, are almost infinite and will allow smart contracts to be assimilated seamlessly into organisational structure, making major efficiency savings as well as being a more secure contract validation solution with a DLT and hence indelible footprint.
”Smart contracting ’technology could reduce banks’ infrastructure costs attributable to cross-border payments, securities trading and regulatory compliance by between $15-20billion per annum by 2022’” Santander, The Fintech 2.0 Paper: Rebooting Financial Services, 2015
[1] Naturalizing a Programming Language via Interactive Learning Sida I. Wang, Samuel Ginn, Percy Liang, Christopher D. Manning Computer Science Department Stanford University
{sidaw, samginn, pliang, manning}@cs.stanford.edu
Naturalizing a Language
The language underpinning the programme is called Voxelurn, and is comprised of sets of directional and mathematical instructions.
Voxelurn is comprised of voxels (objects), with defined relations to each other, allowing you to create sets of rules. Some relational rules given as examples are “actions if S is non-empty, action while S is non-empty”; “perform actions sequentially”; “change property of selection.” When linked to a language like JavaScript, this allows the smart contract to set conditions contingent on online forms being filled to satisfaction, for example.
A function called scoping allows you to more specifically define a set of objects you are working with.
The grammar rules are grouped into four categories: domain-general action compositions, actions using sets, lambda DCS expressions for sets, and domain-specific relations and actions. Using lambda calculus allows you to abstract over variables with a range of values, thus is used in linguistic modelling to capture the ambiguity of semantic meanings.
This is done usings a logarithmic algorithm based on something called ‘semantic parsing,’ that expands the learned definitions for each program, leading to ‘child derivations’ within mathematically defined parameters. And that is how machines learn Latin, and other languages.
Using ‘Def:’ lets you list the characteristics of an object in terms of the core language e.g. its position in a sequence or its relation to other defined variables, so the system recognises it; then you can set a sequence that will be initiated once it is recognised.
The platform’s creators will issue tokens to stakeholders using the platform and contributing code to be reviewed and added to the library. The iOlite ICO is underway and listed on Token Market.
