FAQ

FAQ

What is Akinity ?

Akinity is an open system specification. The specification can be used to create interoperable information management systems having certain key properties.

In what ways is the specification open?

The Akinity specification is published and maintained at akinity.org. It and other related documentation on this site is published under the Creative Commons Attribution license.

The Akinity specification and related documents are provided free of charge. There is no restriction on commercial and / or non-commercial use.

There are some restrictions on how the Akinity specification may be used. These solely concerned with attribution for the work of the system's designers.

The specification of Akinity is maintained through a process of collaboration between interested parties. Any contributors are welcome to participate in improving the form and content of the Akinity specification. Contributions will be accepted based on the organisation's consensus view of what is in the best interests of promoting and maintaining the Akinity system.

Why is it called 'Akinity' ?

The etymology of the word akin in the English language is from kin or blood family. The word usually conveys similarity. The speaker is expressing a view that certain things appear similar, just as blood relatives appear similar. Akin, as generally used, is not very specific about the manner or reason why akin things are similar.

Akinity has an equivalent property. The system quantifies similarity of data, which appears to be caused by an actual kin relation.

Another good reason is that the wife of the system designer is named Aki.

What does 'cTag' stand for?

The preferred pronunciation of cTag is with a silent c. Note that the second letter is capitalised.

The derivation of the word cTag is deliberately left unclear. Certainly the word tag is a key component, since cTags are really a special kind of tag as commonly understood and used in computer systems. The presence of the four letters that comprise the four bases of DNA is also intended.

Neveretheless, the correct backronym of the small c in cTag is indeterminate. This was done intentionally to illustrate an important point about definitive and partial information in Akinity. It is hoped that the following example will help the reader to understand a fundamental property the system.

Bound to the definition of 'cTag' in the glossary is a cTag which (for the system's designer at a point in time) communicates a range of meanings for the first letter in cTag's backronym. This does not imply, however, that alternative cTags, presumably highly akin to the one in the glossary, would be any less good. Ultimately, the fitness of one or another cTag variant will be determined rather by the prevalence and frequency of their reproduction in the evolving landscape of subjects' cultures, than by fiat.

To peruse this particular cTag in the glossary using a culture comprising parts of the English language should reveal that its author intended to signal the relevance of terms such as: culture; cultural; conception; communicate; common; common currency and computer. There were in fact many more words included; some in English and a few in other languages too. Not all beginning with c.

The inclusion of one particular English word, was an act of non-gratuitous self-vandalism. It is a word that might be found written on the door of a public toilet. Subjects having this word in their culture could reasonably be inferred to have implicitly accepted that they would not find this word's use to be offensive. More importantly, those for whom the word is absent from their culture would neither see nor presumably would they care to see this word in any context. Thus Akinity can take a single data item and communicate different information to different readers. The different information held by the subjects is pertinent to their readers' different interpretations of the communicated cTag data.

Isn't tagging so 2004?

It surely must be the case that the hype cycle for natural language Tags has plateaued by now. One consequence of this is that tags are commonplace on the web, document management systems, operating systems and the like. Such resources are, by and large, potential hosts for Akinity's cTags.

For many forms of data (URI for instance) it is inconsequential to the system what form the data takes, so long as it is syntactically correct and the data is functionally identical wherever used. For Akinity though it matters greatly how data is represented. The system cannot just use some arbitrary symbol to express a concept since, in a sense, for Akinity the medium is the message. Therefore the binding of your original data to the Akinity cTag becomes key. Tags are an appropriate way to bind a data descriptor to a document.

Functionally too, tags seem a natural home for Akinity. After all, Akinity aims to achieve much the same goals as natural langauge or semantic web tags.

Is Akinity a web application?

Akinity is not an application. It is a technology which might be used in applications. Some of which might, but need not, be on the internet.

An application that demonstrates Akinity can be downloaded from Ingendex. This application does use web protocols and technologies. But it is unlike many web applications with respect to its architecture.

How did the system come to be?

The principle underlying Akinity was discovered while its designer was exploring two-parent inheritance relations. A commonplace amongst sexually reproducing organisms, it appeared anomalous that such a design had not yet been used in information management systems.

The designer initially made no reference to previous work. He established from first principles the possibility of some desirable properties of such a system and went on to outline a system design again without reference to previous work.

Is Akinity really novel?

Well, without having conducted an exhaustive examination of all the literature, in all honesty, we believe that the principles of Akinity have not been documented nor developed elsewhere.

If we were attempting to patent this system (incidentally, it is not intended that Akinity be patented), we might declare the key points of innovation to be :

A data format which includes a sufficiently large sample size of discrete binary bits so that statistical analysis can be deployed.
- Each bit in the data format optionally to reference a value which conveys additional information about prior generations.
An algorithm to convert ad hoc digital data into the standard data format.
- An algorithm to dynamically expand the breadth of a Synthetic cTag, in order to be able to process more or less data according to the requirements of the application.
An algorithm to iteratively combine pairs of discrete bits, packaged in two instances of the standard data format, so as to optimise similarity through successive generations of the function.
- This algorithm produces output in the standard data format.
An algorithm to read off the distance between a pair of cTags based on their quantity of bits in common and to present the result as entropy; an information theoretic normalised form.
A method to dynamically expand the breadth of a Synthetic cTag, or to contract any cTag, in order to be able to process more or less data according to the requirements of the application.
Consideration that a cTag be functionally identical in two polarities and for meiosis to use reversal to prevent unwanted accumulation of low entropy at any bit.

It is certainly possible that prior art exists for some of this. Trivially, synthesis is little more than a hash function, which is well documented elsewhere. Clearly though, SHA-256 alone does not deliver the properties of Akinity.

In our view, an earlier system that demonstrably includes equivalent functions for both meiosis and akin should probably truly be considered prior art to Akinity. Absent either of these two functions, though, and that system should not be considered prior art. The reason for this choice is that akin delivers the principal output of the Akinity system while meiosis is the key method of generating the conditions for that output.

What is Akinity's theoretical heritage?

Akin has Shannon's Information Theory at its heart. For Akinity, noise is the appearance of a kin relation where none actually exists. Signal is a pattern of data such that an inference of kin relation may be correctly inferred from it.

Meiosis is inspired by living systems; specifically sexual reproduction.

Who is the target audience for this web site?

Many of the documents at akinity.org are quite technical in nature. This is because the system specification and related documents are mainly intended for application designers, who may wish to incorporate Akinity into their own systems.

Other material on this site is of more general applicability. Please suggest ways that we can use the web site to better target our audience.

What is Akinity for?

Akinity is a general purpose, open, subjective information management system. Principally, akin provides classification or filtering services.

Though it does not have specific use, it is believed that Akinity may be used in a variety of ways to add functionality to existing systems or new applications can be created based on the special features Akinity delivers.

Akinity shares many of the same objectives of various Semantic Web and tagging projects.

What is the Akinity reference application ?

Ingendex is a client-side web application, built in Javascript on top of (Google) Gears.

The software for this reference application is once downloaded and thereafter is run from the user's local machine. Data is held exclusively on the user's local device.

Where is the Akinity server ?

Though some applications may well offer their services as a client / server architecture, Akinity does not require a centralised information repository.

cTags may be stored on a local machine, in the cloud, replicated across machines on a local network, hosted in documents across a user's many places of interest on the web or in enterprise databases.

Ingendex has no centralised server. The intent was to demonstrate de-centralised yet interoperable nature of the Akinity system.

How is semantic content in Akinity created and curated?

A significant difference between Akinity and other semantic systems is that Akinity uses the process of interpreting data into information as a means to facilitate improvements in the system's data. This process of gradual evolution is how semantics can 'emerge' from ordinary use.

For instance, without any action by userA (uA), the semantic power of uA's data (dA) can be enhanced over time, if system usage by another user (uB) should conceive new descendants of dA through meiosis having as input: dA; any of dA's ancestors; any of dA's descendants; even descendants of dA's ancestors.

Is there a definitive set of 'reference' data?

There is nothing definitive about information in cTags, which individually are merely data.

It is the interpreting activities of subjects where information emerges through a bottom up process. Subjects having common data (e.g. some subset of words in the English language) may concur over certain terms, while holding nothing in common over their other data terms. The system would not work if such choherence were entirely absent (consider Wittgenstein's private language).

What is the difference between data and information ?

Akinity makes the following distinctions. These definitions are narrower than might be used in general speech.

Data can exist independently of information. Data has no inherent meaning. In principle, data may take any form. In practice, data is defined as any potential input to the synthesis function. So, in the Version 0.1 of Akinity, data is amenable to the SHA-256 hash function.

Information is produced in the confluence of two or more instances of cTag data and a subject who interprets their relative entropy to assign meaning to the data (or more likely to the tagged data items). Information is thus the output of the process of interpretation of cTag data by a subject.

Must a subject be human?

A subject may be any process having access to a culture that is capable of using akin to interpret cTag data as information, through making judgments concerning distance and uncertainty.

What are the special characteristics of Akinity ?

Every cTag inherently has a deterministic distance, vis-a-vis every other possible cTag.
Data held in the cTag format implies a measurable quantity of similarity to every other cTag, actual or potential, without necessary recourse to a separate look-up table. All the relationship information you will need is encoded in the cTags themselves and the algorithms which created them.
Interpretations of some cTag pairs are more pertinent than others. Interpretations with quantitatively more similarity reveal a greater probability of an underlying semantic relationship between the participating cTags.
Akinity tokenises information theory. Any data may be encoded, and subsequently interpreted as information using entropy-normalised similarity units of akin's results.
Information acquisition is always an act of interpretation.
Interpretation generally improves the fitness of data both to the interpreter and to other agents. Therefore read-only / copy is usually a less appropriate way to replicate data than is meiosis.
Pull-only interpretation. You need something to get something more. Ordinary data is push-pull as you always get the same, but interpretation is not in scope.

Is Akinity useful for encryption?

Synthesis, in Version 0.1 of Akinity, is a one-way cipher. A supplementary method of synthesis using a two-way cipher would be possible. However, there is no current plan to add this to the specification of Akinity.

If encryption is sought then a special-purpose algorithm and secure implementation is recommended.

An interesting property of Akinity is a propensity for steganography.

Why re-invent the wheel ?

Natural languages have evolved over tens of thousands of years to work under constraints which are less applicable to a global digital information system than they are to face-to-face communication. Many variables have influenced the form of languages. The range of sounds audible to the human ear; the shape of the human palate; background noise correction; the number of words that can be accessed in the mind in real time; the average number of interlocutors in a conversation; the variability in membership of every participant's conversations; the degree of experience shared by communicators; the typical length of time available to learn new words; the capacity of the human mind to use metaphors. And so on.

Though the forms of natural languages are somewhat different, these designs emerged through an evolutionary process which is consistent with optimisation for human-social constraints.

Contemporary communication practices have, in some fora, removed or altered many of these constraints, to the extent that it now seems worthwhile to re-examine the role of natural languages as the fundamental basis of information transfer in every fora. We believe that the Intercourse will eventually become a new form of communication - an Interlingua or a common currency for digital information transfer - by means of which other communication forms such as natural langauges will be readily integrated with each other.

That is a large vision, are there any limitations ?

Since the minds of humans, and other sentient creatures, appear to be unsuited to computing digital information directly, the utility of Akinity relies on the presence of computing interfaces as a tool to enable communication in such a form.

Does Conception produce unique cTags?

Synthesis
- is guaranteed to produce the same output from the same input data. It is sensitive to case, plural, punctuation etc.
- is not guaranteed to always produce different output from different input data. However, due to the hashing algorithm used, it is very unlikely that collisions would be encountered under normal conditions. Unless, of course, the input data had been identical.
Meiosis
- is guaranteed to produce the same output cTag from the same two input cTags
- is not guaranteed to produce different output cTag from different input cTags. Again, meiosis is vanishingly unlikely to produce a collision unless the colliding meiotic cTags were conceived from the same pair or a pair highly akin to the original input cTags.

A strong hint is that if you are using a cTag as a unique identifier then you are probably using it inappropriately. It is not so much a question of whether it is safe to use a cTag this way. Rather, that the whole point of a cTag is its semantic distance to another cTag, which may well have been generated in an external application. cTags are generally used to express the similarity (akin / like) relation. The identity (= / is) relation is probably better expressed using a special-purpose unique identifier, generated by the application in which it will be used.

How can I contribute ?

There are many different ways that people who are interested could contribute to Akinity. You could :

use the reference application and, by posting cTags online or writing a blog say, help spread the word about Akinity
use the reference application and submit bug reports
mirror the reference application download on your server
port the reference application to other platforms
help to implement and maintain a source control system and associated processes
help to develop and maintain ingendex.akinity.info using Gears and Javascript
build one-off interfaces to current and emerging technologies
develop your own independent implementation of the Akinity specification
contribute to writing supporting documents
help build the organisation
translate documents and applications into other languages
help plan and execute marketing campaigns
contribute to developing the specification of Akinity
suggest a source of funding for akinity.org

What does Ingendex do?

Ingendex has some similarity to an on-line bookmarking service like Delicious in that you can cTag documents that you have found on the web, in order to be able to easily find them again at a later date.

Unlike Delicious, which is a web service, Ingendex can work off-line. Application code and data is held in folders local to the machine, where it is not normally accessible to other users. cTags generated by the reference application may nevertheless be uploaded wherever a tag can be posted on-line.

Such an architecture might be thought to go against the grain of increasing connectivity. In part, the architecture was designed to demonstrate the possibilities of information sharing, without necessary recourse to centralised servers. As data our becomes increasingly connected, the trend for it to be replicated, sold on and mined is sometimes thought to be increasingly pernicious. This application is to demonstrate an alternative.

Akinity has many more potential applications than this reference application demonstrates.

Akinity

Deconstructured Data

FAQ

FAQ