Architected Futures™

Tools and strategies ... for boiling the ocean

Architecture Information Repository (AIR)

Repository Definition

The Architecture Information Repository (AIR) is the facility within the Architected Futures™ framework that provides the data store for the architecture knowledgebase. It is where the awareness of the architecture is stored. The representation and storage format for the knowledgebase has undergone multiple transformations over time as the concept of the framework has evolved.

The most recent scheme for the knowledgebase is to define the knowledgebase in terms of a semantic knowledgebase. One of the strong motivators in this direction is that this form of data representation would enable the use of current and emerging semantic tools to provide, enable or facilitate desired operations on the knowledgebase within the framework.

Design Element

Prior to this point we have used a GUID of our own design as a universal identifier for all elements. This has the benefit of providing a fixed size element which provides some efficiency in automated operations. However, a problem with this approach is the complete obfuscation of all meaning connected to the internal GUID when it is viewed by a person. In that sense, there is a strong motivation to move toward the use of either a RDF namespace type notation. However, one aversion to the standard RDF type notation has always been the issue of language. Part of that issue has to do with choice of language (English, etc.) for words and the other part of that aversion has to do with the choice of words as selected to represent concepts. The use of GUIDs avoids these two issues and allows for translations to be employed for tailoring the representation of the GUID to a language and word choice that is most appropriate to the target audience.

Athena Notes

(See font page, column 3, near the bottom, I think.) For everything. Soup, kitchen, nuts, bolts, aircraft, the Panet(?) Pluto, and every member of the community and their grandmothers. We find out about stuff that we care about, which is a critical feature, then we filter, and then we catalog it. We care about stuff, we catalog it. We don't care about stuff, we don't catalog it. But we don't assign GUIDs until we anticipate a need to deal with an element as a unity. [Most GUIDs are assigned and used internally and automatically by the system. EATS. We only tend to look at GUIDs to verify stuff if the system looks a little flaky, to debug it, or to validate historical accuracy. GUIDs are hard on your eyes. People don't use GUIDs unless they have to.Personal note. Looking back to the 1960's when I was at Cal Poly, there was a "revolt" about people being assigned numbers. I remember being handed an "IBM Card" to be able to register for classes. I can almost remember my number, 7 something I think. or, maybe that's my CDL. You had to write both numbers on your checks at the book store. Now I'm inventing schemes, yet again, to assign mumbers (GUIDs) to people. But, at least its only 1, it's completely unreadable, and it's the same as all of the other animals, plants, m inerals, vegatables, atomes, forces and conjectures in the universe, so maybe its not so bad.) Creating GUIDs early, is not a problem. When our current size specification becomes an issue, there are easy solutions because it was "designed to grow." There isn't supposed to be any issue, numbers are an infinite set. We can catalog whatever we want and keep track of it. Forever. Off and running.

We don't assign any GUIDs until we identify something. And then we assign a GUID if its something we didn't already know.  And we start to classify it. (If we've never seen you/it before, and never heard of you: "Oh, you're a new member" or "Oh, that's a new idea!", we give out new GUIDs. If we've heard about you before you arrive, there might already be a GUID waiting in your name1.) Community member GUIDs are no different from any other GUID. All GUIDs consist of a lot of mathematics gibberish that is used to try to assure that they are unique in the universe as we make them up, yes, the big, big, big universe2 So, every thing and every one has a unique identifier. But, at the same time, "they are all the same" so there is no way you can just look at one and tell what it's about, if it's a "who" or a "what" or a "where" or a "how" or a "when" or a "why." There is no pattern. It's just not there. Whether it's an object or an action. Whether its a noun or a verb. Just that it's something that we have identified, and it's in our model. Honest. So, like everything else, you get lost in the system. Big deal, you've got a tag. Don't worry about it. It's a whole lot less conspicuous that an "@facebook" or "" identifier. It helps you get free badges. (But you have to work for them. They need to be earned. Except for the one that says "member." That's free, or so they say.) [Warning normal people, weird or geek science ahead --> happens at these3-> [The reason for the extra stuff above is because this works both ways. Not the security, the tagging. That same tagging that "gets things lost" is the mechanism for putting them back together, for "un-losing" them." Multiplexing and demultiplexing. [And "weaving", aka creating new patterns, aka intelligence [ [{Bug Report:4} Intelligent Un-losing + pattern matching + recognition would be the next frontier[]Anyone there yet? That sounds like security in the movies, how the cops track and spot the guy instantly by knowing his patterns and doing facial recognition on the monitor as he goes through the subway. So a frontier must be further out.]] EATS at its current design level is about pattern matched multiplexing as an Augmented Intelligence Toolkit. This is core to how it's done.[Core Architecture Concept]] That other stuff in there, those were maybes. This is how I do "architecture." It's how I also bias my algorithms. By asking, what would my next moves be? How else could we help the whole of what we're doing? Lot's of simple compare contrasts that would be done faster, and be more accurate, and inclusive, if they were done by somebody or something else if it mattered and I wanted it done to a measureable standard. [Horizontal Search for inclusion] A computer is fully capable of searching a lot more stuff than I am, a lot faster. And I could just decide from that right/wrong, good, better, best, yes/no, "around 5," or "probably okay, but it would be advisable to monitor the situation. Pease let me know if it gets ourside these parameters."  That's called "Augmented Intelligence" if its adone as a cooperative exercise between a human and a machine, at least in my parlance, and that's what EATS is about. Put that on your iPhone and tell Siri to take a vacation. But first you'd have to down load Ani to do the job., or leave Siri there as an assistanct for Ani, a lot of peole would probably like that option. You talk to Siri, Siri will talk to Ani, and Ani will talk to Ani will talk to Ani until she figures it out, or tells you "I don't Know, but here's some possibilities."]

  • 1. Have you ever seen an "if this is you..." link in a query result?
  • 2. If we ever do meet somebody on the SETI network, we will have an ideal mapping system to try to demultiplex whatever it is they are talking about. It could be that we are going to pick up some of those tapes we've been collecting, thank you Carl, and find out that there is a background intergalactic, party-line, phone call that has been going on for a long, long, long time; and we are just the new comers to the conversation. (Visitors from the past. If this is the Star Wars contingent, maybe we can find out what the real end to the Empire was. There might be some history lessons left from C3P0 to R2D2. No I'm not psycho. If alien [computers] were talking to other alien [computer] across space, the demultiplexed signal would reveal a historical record of the conversation.) In the meantime, as they used to say "back at the ranch," the same system of cataloged, relationsional knowledge can be used for problem solving back here on earth, in "Real World" and "Vision World," our simulation and awareness monitoring modules.
  • 3. If I had better software I wouldn't have to write like this, and you wouldn't have to read it like this, whoever you are. I hope you are making your way through this. At whatever level you are reading it. It's one document, aimed at a lot of levels, from a lot of different perspectives, all at the same time. The editor I have constructed on my latest prototype EATS model would be a much better way for me to write this documentation. And that's what all this stuff is folks, documentation. Anyway, the editor works better, and clears up massive (well, not massive but problems) with getting it to be both write-friendly and read-friendly and appropriated "componentized" and "cataloged"  while I write it. This is bad on both sides.[Architectural Analysis] I'm taking the time, from time to time to try to include notes. Notes that a simple, reasonably simple, computer algorithm could follow. ..
  • 4. I don't know how many of you are aware of how text processing systems worked in 1970, but that's when I learn to use some of this stuff. Improvements send things forward, but there's also back slides, and side slides. What I'm doing as I write this, is essentially the same as writing HTML with a text editor. For the record, I'm using a variant of CKEditor. Whick is ok for what it was designed for, but for what I'm doing, it needs som elove, and I don't have time to it myself, so I work with what it's got. Sound familiar? The point about HTML in a text editor is that it's a language inside a language. You tell which "level" you are working on, with symbols to help the computer understand. Like the different between HTML and content. Content is what you are trying to communicate. The HTML "wraps it" and directs formatting. all of which is "wrapped" by your web page or whatever. Two levels, not so hard, but HTML editors do a nice job. I'm trying to use my two-level HTML editor for a "three or more" level documentation stream capture. Symbols in the text are supposed to be able to help direct a parser in EATS to disambiguate the levels and demultiplex the text. A "smart EATS" with assistance from AnI should be able to parse what I'm writing and clean up some of these web pages. And, feed the repository. In the meantime, I'm trying to write for both human consumption and later machine parsing consumption. The symbols for those are getting confused sometimes and I need to change my symbols to keep from having the CKEditor go crazy. My use of '[' symbols needs adjustment. A big problem is grammar and puctuation, which has to be correct at each level, and at the same time you are trying to get the symbolic punction correct at the human-to-human level. It's a pain, that could and should be corrrected.

Add new comment

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.