From - Tue Feb 20 18:08:14 2001 X-UIDL: bf1457390894cbb7 X-Mozilla-Status: 0001 X-Mozilla-Status2: 00000000 Return-Path: Delivered-To: lacall-onemodel:org-lacall@onemodel.org X-Envelope-To: lacall@onemodel.org Received: (qmail 70147 invoked by uid 800); 20 Feb 2001 14:11:02 -0000 Date: 20 Feb 2001 14:11:01 -0000 Message-ID: <20010220141101.70146.qmail@uruz.pair.com> To: lacall@onemodel.org References: <3A927AE2.4010408@onemodel.org> In-Reply-To: <3A927AE2.4010408@onemodel.org> X-Loop: general-digest-list@onemodel.org From: general-digest-list-request@onemodel.org Reply-To: Please.write.a.new.mail.instead.of.replying@FIRST.WORD.archive Content-ID: <"volume00/8"%general-digest-list-request@onemodel.org> Subject: archive retrieval: volume00/8 Precedence: bulk Content-Disposition: inline; filename="volume00/8" Content-Type: message/rfc822; directory="volume00"; name="8" MIME-Version: 1.0 From: general-digest-list-request@onemodel.org Subject: general-digest-list Digest V00 #8 X-Loop: general-digest-list@onemodel.org X-Mailing-List: archive/volume00/8 Precedence: list MIME-Version: 1.0 Content-Type: multipart/digest; boundary="----------------------------" To: general-digest-list@onemodel.org Reply-To: general-list@onemodel.org ------------------------------ Content-Type: text/plain general-digest-list Digest Volume 00 : Issue 8 Today's Topics: Re: Markets for a Unified Model of K [ Lee Howard ] Re: Quite honored indeed. [ "Tom and other Packers" ] Re: Markets for a Unified Model of K [ "Tom and other Packers" To: "Tom and other Packers" Cc: general-list@onemodel.org Subject: Re: Markets for a Unified Model of Knowledge Message-Id: <3.0.6.32.20000814211455.0080ad30@server.deanox.com> Content-Type: text/plain; charset="us-ascii" Wisely so, SmartList strips all of the recipients from the original e-mail and only reveals the sender and the list (not CCs), so Jared will miss out on most conversation here unless he subscribes (because Reply-To-All won't catch a non-present address). Anyway, Tom, I think that you've done a sufficient job of illustrating the usefulness (can you pluralize that? properly? I think uses is the plural, anyway...) of OM. If my line of questioning led you to believe that I thought such an endeavor was without use, then I miscommunicated. I do not doubt (and even moreso now) the value of having a pool of all knowledge at one's fingertips. And I'm not drawing equivalencies, but in many ways I've viewed OM (well, MW before) as an attempt to reinvent a library. Assume, that I had an infinitely large library; assume that its organization was perfectly consistent; assume that it was completely cross-referenced in each page of every book and in each index and catalog (or catalogue for those Brits out there); assume that this library were electronic; and assume that information was continually dumped into it by everyone. Well, essentially this is (what appears to me) to be the goal of OM as far as being a reservoir of knowedge. And, on a small scale compared to OM's goals, and on a much larger scale than currently done by OM, many large libraries already do (or have the goal to do) what OM's intent is... and frankly, that's competition, and frankly I think they're doing a mighty fine job of it now. And as the books become more and more electronic and searchable and referenced by the readers, I think they'll outrun OM... at least as far as I can foresee. In order for OM to overcome the odds here (as far as being a repository of knowledge) and beat out the traditional library, then its data gathering would have to be automated... and in order for that to happen there's got to be a pattern of gathering established and between having knowledge and gaining more there is that requisite link of establishing and internalizing a pattern found among the knowledge already had... and in order to learn completely new things, the ability to autonomously alter or expand that known pattern must develop. (Use model in place of pattern, if you prefer.) Anyway, the underlying issue here, and the source of my critique is that there is an assumption being made that the knowledge pattern among one traditional subject will have any relevance or coherence with the pattern found among a traditionally completely unrelated subject. I do not doubt the value of having my source of information about mathematics being the same source of information about zoology. Certainly that's simple to understand... however, surely you can see that (even though it may be somewhat archaic) the traditional library already serves this purpose. In order to serve the purpose of something better than a library, OM must automate the knowledge acquisition process and significantly improve, filter, or streamline it. And for what I know, the way to do that is for a pattern to be established. The third learned language is easier than the second, which was easier than the first because the mind had already developed similar language patterns which it could follow, and the mind could observe when new knowledge did not fit that pattern, and then the pattern was adjusted so that the new knowledge could be retained as part of that particular language pattern. And, general (or specific) language models are already developed, and they could even apply to mathematics (another language) in many ways... but the crutch of my argument is that the importance of this project lies not in finding patterns of one language in another, but lies in finding patterns of language in the Earth's geology and climate. And frankly any correlations between them seem to me to be coincidental. Surely I believe that all knowledge comes from One Source. Surely I believe that all knowledge has common ties, but I have my doubts that the pattern of knowlege among one field of study has any real interpretation among every other field. If that makes sense. So, my question from a week ago could better be stated, "What has mathematics got to do with zoology?" (In any useful way. The fact that Tom studies both isn't very useful. Studying both may be useful, and surely they have their commonalities, but to say that the knowledge structure of one mimicks or is remotely similar to the knowlege structure of another is... presumptuous - surely there's a hundred more dissimilarities than there are similarities.) I'm being redundant. Anyway, if there is no common pattern then there is no value to the pursuit of OM because it won't serve for much more than a library. Lee. At 10:52 PM 8/13/00 -0600, you wrote: >2000.08.06/Sun: > >Hello OM List, > > Synopsis of the Following Long Letter which I wrote a week ago: > > An informal essay on the usefulnesses and potential markets for a >unified model of knowledge, including inference, (UMKI). Four overlapping >areas of application are discussed, with examples of each given. There >won't be a test afterwards, so don't read this if you don't want to. > > ~~~~~~~~~~~ > ------------------------------ Date: Mon, 14 Aug 2000 21:53:05 -0600 From: "Tom and other Packers" To: "Jeremy P. Almond" , Cc: Subject: Re: Quite honored indeed. Message-ID: <000501c00675$fc35d2e0$1f02a8c0@oemcomputer> Content-Type: multipart/alternative; boundary="----=_NextPart_000_0016_01C0063A.06692DC0" This is a multi-part message in MIME format. ------=_NextPart_000_0016_01C0063A.06692DC0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Silly. tomp ----- Original Message -----=20 From: Jeremy P. Almond=20 To: general-list@onemodel.org=20 Sent: Monday, August 14, 2000 11:34 AM Subject: Quite honored indeed. Hello all you funny people, I am quite honored to be a one model mailing list member. I am not = much into deep thinking but I am a very open minded person from the land = of the Orient. The trees there are beautiful and much desirable to look = upon. =20 If any of you people are interested in the where-about of the Ten Lost = Tribes I would be delighted to enlighten you. I am looking forward to hearing from the authorities. =20 Thanks, Jeremy Almond = -------------------------------------------------------------------------= --- Have a nice day. = -------------------------------------------------------------------------= --- ------=_NextPart_000_0016_01C0063A.06692DC0 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Silly.
 
tomp
----- Original Message -----
From:=20 Jeremy P.=20 Almond
To: general-list@onemodel.org
Sent: Monday, August 14, 2000 = 11:34=20 AM
Subject: Quite honored = indeed.

Hello all you funny = people,
 
I am quite honored to be a one model = mailing list=20 member.  I am not much into deep thinking but I am a very open = minded=20 person from the land of the Orient.  The trees there are = beautiful and=20 much desirable to look upon.  
If any of you people are interested = in=20 the where-about of the Ten Lost Tribes I would be delighted=20 to enlighten you.
I am looking forward to hearing from = the=20 authorities.
     =
Thanks,
 
Jeremy Almond
 
----------------------------------------------------------------= ------------
Have=20 a nice=20 = day.
-----------------------------------------------------------------= -----------
------=_NextPart_000_0016_01C0063A.06692DC0-- ------------------------------ Date: Tue, 15 Aug 2000 07:53:30 -0600 From: Luke Call To: general-list@onemodel.org Subject: Re: Modeling Relations -- Re: use cases Message-ID: <39994B5A.8080509@onemodel.org> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit I believe I've understood some basics things from Tom & Mark about modeling; keep an eye on developments to make sure the concepts you've outlined make it into the specs as we go throught this. At what phase do you think it would be best to address tracking certainty? Maybe after we get the basic scheme and operations working well. Replies/questions on Tom and Mark's email are embedded below. I've mixed quotes from each of them: > Relations like "is-a" are often hard-coded because they are very useful. > But if we hard-code this one, and not others, we risk writing algorithms > which only make use of a narrow subset of the possible relations, and > therefore make a model which is not fully generalised, not able to take > advantage of the other relations as well. If we don't add any distinction > between relations, and are forced to define all relations the same, using > meta-relations, we are more likely to be able to handle any arbitrary > knowledge the same. Tom, can you give a detailed example or two of meta-relations? Sounds like a relation is another attribute that the user defines (key relationships being predefined); in essense, a relation is another entity. The part where I get confused is where they are compared to "inodes" and "meta-inodes", so if you could find some examples to illustrate all that, it could help clarify this for me. It sounds like since relationships are entities on their own, they have types, and each given relationship is an instance of that type, "is-a" being one example of a relationship type, having many instances where it connects specific objects. Is this what you have in mind? > I still don't know what you mean when you ask "what is the 'is-a > pointing' to"? Since "is-a" is intransitive, it will of course by > "directed", so instead of a line, it would be represented by an arrow. So, > if we define this relation strictly enough, such that there is an order > specified (which "is-a" implies, but you have to remember that there are > three very closely related relations that I've been talking about when I say > "is-a": (1) the "subtype-to-supertype" relation, (2) the > "supertype-to-subtype" relation, (in both of which, order does matter), and > (3) the union of these to sets of relations, "subtype/supertype", (in which > .... >Yes. Except a class may also define strict rules that must be >assumed to be correct for all members of a class. e.g. If I > say that {A.weight < 100 lbs} and B is A then I must be >able to say that B.weight < 100 lbs. So, when a user creates some form of is-a relationship between entities, perhaps the system can assume it is an instance of an existing type unless they change the attributes. So, as mentioned before, each class is defined by a "type-definer" instance (an archetype), which has additional type-specific data defined, such as constraints, default values, and the type name. These could be moved, if someone prefers, to a different entity, which then becomes the archetype in its place. But then when you add or remove attributes from a member of that class, does that automatically define a new class, modify the existing class (we could prompt the user for which), or is a new type created only when you sub-type it by creating another is-a that points to it as the parent? > Second Solution: inclusion of "worldviews" into the model. A worldview > is assigned a certainty, and all relations would be classified as belonging > to this or that worldview (and perhaps more than one -- redundancy is very > good in promoting certainty). >Definitely - there must be checks to locate inconsistencies >and allow removal or refinement. Attributes and >relationships can be considered to be a set of >equations in multiple variables - if you can't >find simultaneous solutions for all attributes of an >entity then one of the following applies: >1. The entity is not real >2. One or more of the attributes are not real >3. One of the relationships / constraints is false >and needs to be revised. Could we handle worldviews and inconsistencies in data by allowing inconsistent data, but being able to show where the inconsistencies are, and allowing a user to choose the data set they prefer (or choose it for a given query), given the conflict? > Also, multiple "is-a" internodes are desired, and necessary, I believe. > The way to handle this (if I know which sense you mean, by "multiple is-a > relations"), is using meta-internodes. You define the specific type of is-a > relationship, and the, for logic requiring the ability to know of all "is-a" > relationships, regardless of specificity, as in a query with little > constraint, it would use the meta-internodes, and perform a wider search. F > or narrower searches, such as for "makes and models" of cars, it would not > need the dispersing-effect of meta-internodes. Multiple inheritance as you've described above makes sense to me. But as to the meta-internodes and queries, detailed examples in layman's terms would help--describing specific entities, relationships,and how a search would traverse them. Maybe a handful of dissimilar examples would be best. This would communicate it not only to me, but would make use case fodder for us to build on. As Mark pointed out earlier, there could be inconsistencies in the multiple parents. We would need a systematic way to deal with those--perhaps checking for them at the time multiple parent types are related to an object, then forcing the user to reconcile or specify which side overrides the other, before saving the data. I'll try to get a new proposed list of use cases out, next. Luke ------------------------------ Date: Tue, 15 Aug 2000 08:39:17 -0600 From: "Tom and other Packers" To: "Lee Howard" Cc: "Jared (h) Norman" , "OM List" Subject: Re: Markets for a Unified Model of Knowledge Message-ID: <002c01c006cb$0fd1a860$1b07a8c0@oemcomputer> Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Lee Does this mean that you can't see Jared's address in this letter? Certainly it doesn't mean that he didn't receive this letter, right? I was using BCCs before. Nope, I knew that you knew that there were uses for OM; but when you said that you wanted an algorithm before you would get too excited about the project, I thought (and think) you needed a little more motivation to maybe help us find it. I'm sorry I didn't give you a very large database in your mw file format. As soon as we have some sort of format that I think we will be using for a while (or which will at least be _useful_ for a while, even via subsequent translation into a later version) I will probably start working on just such a database again. (Everyone working on the OM project will probably need such a database for testing.) Let me see if I understand yet. There are data structures, and there are algorithms to operate on data structures. Om is potentially one data structure. A library is another data structure. They each potentially could contain all knowledge relevant to any algorithm. What can an algorithm acting on Om do that another algorithm acting on a library can't do? In any library that you've ever heard of, existing or designed, can you perform a search/query which returns all instances of the following complex and abstract condition: Object A of type C contained object B1 at time T1, and the axiological value (the practical output or usefulness) of A was N1, with an epistemological value (certainty or probability) of M1. Object A contained object B2 at time T2, and the axiological value of A was then N2 with certainty M1. T1 > T2. N1 < N2. M1 > 0.5, and M2 > 0.5. Return all instances when the difference (growth) between N1 and and N2 was X. Let me explain what all that means. A, B, C, etc. are not given names, but are defined by their relations to each other. Given my personal value system and believe system, I am looking for any instances of when a component t was added to some object which increased its value. One example would be to infer the increase of value of a computer from its increased output after replacing a small RAM chip with a larger RAM chip. This query could be restricted to yield results more relevant to a particular subject, such as by constraining C with additional relations. This is very much a practical, realistic, and conceivable use of OM, but I've never heard of any library which even attempts such a utility. Libraries don't intend to model axiology, epistemology, and only to a very weak and disjointed extent ontology, the three necessary components of knowledge (according to Em-Veh, and according to a certain author of an old philosophy text who shares Mark's last name). A library's purpose is not to model knowledge or make inferences based on all recorded knowledge. A library's purpose is to make available literature, factual and fictional, and everything in between. It's the readers job to manually decide how factual the book is, and how valuable it is, and to make other inferences. If a library _will_ do this for me, then fine. Bonus points for that library. (I've been telling people that I might work on digital libraries when I grow up and get a PhD in cognitive science.) If the library intends to do everything that OM is intended to do, then it is either (1) truly competition, or (2) where I want to market OM. (Actually, that's where I would want to market _Informatica_, because personally, I don't expect that we will have something optimal the first go-around, especially since I don't think that any of this group is serious enough to work out the theory first, and start implementation later, perhaps much later. Hence this is a testing ground for some ideas. The rest of the ideas I've been gathering since 1993, and the ideas I will continue to gather through this project, I intend to implement in Informatica, which will be based more strictly on Em-Veh and structured using Mathesis and mathetical language. Maybe you guys intend to work out the theory and the practice simultaneously, which might be fine, too. We'll have to see.) With OM (or Informatica), axiology and epistemology can, to a large extent, be automatically calculated for you, based on a few descriptors listed in your personal "worldview". Let me explain just one of the problems of my query example, problems which I think, in most libraries, would be considered to be more cost than benefit, would not fit their mission statement, and therefore would not be attempted. Cataloguing all objects as being "types" of some other object would be very silly because every concrete thing (and almost every abstract thing) is a type of something else, and is usually a type of many things at once. But to start a query that is as generic as the one I gave above, every instance of "type" would have to be considered and compared to the constraints of the query, otherwise you'd probably miss something. Do libraries store metarelational information, such as when one relation is a type of another? This goes back to my "make" and "model" example in another letter. Is a library going to classify the relation "make", which relates my car's subclass to it's superclass, as an instance of the "subtype" relation? If it didn't, the query in that library could then miss all instances of when an engine was replaced, and the value of the car went up. In this case, C is the make, A is a car, B is an engine. More likely, the library will let you search for author, subject, ISBN, etc. Oh, speaking of subject searches, libraries should develop this feature, at least the very least: I search for books about mammals. I expect it to return more books than I could count, let alone read, because it should give me titles of books which are about mammals in general, books about dogs, books about cats, and even books about restaurants which happen to have a small chapter on the problem of rat infestations. Do you think libraries do this? This is not metarelational. This is a simple relation. But I don't think they do even this much yet. The most advanced stuff I've heard about so far (and, I must admit I'm not studying this stuff much yet, so I really don't know, but ...) is the attempted use of natural language processing to "read" a book or journal article and automatically index it, and pick out subject words (and probably author, title, etc.). Just subject words. No relations that I'm aware of. Just key subject words. This was one of the important things that the world famous computational linguistics professor at Brandeis University, Pustejovski, was doing last time I checked (a year or two ago). Sort of pitiful, don't you think. I had a class not too long ago called "research methods in biology" which was held at the Weber State library, and in which I had to write a research paper on my subject (zoology), and at my fingertips I had some of the most important and renowned online indexes of primary and secondary scientific literature available in the world, like MedLine for medical information, and government indexes, and ... I forget the names. (I was obviously not that impressed.) Do you think any of these indexing services had a "semantic net" to work with, in finding books on my particular subject? Heck no. They relied on a line or two of descriptor terms which some guy probably had to type in manually in order for that journal article appear in at least some of the queries in which it would be relevant, and hopefully not too many queries in which it was irrelevant. These subject searches are by no means complete. Nor are they concise. But that's probably another story for another day. (For example, if there is not a good semantic net with lots of relations and metarelations, and if I had specified that C = VW auto, in my thought-experimental query above, and A = Fox, I might get a whole lot of information on the mammal of the same name.) Right now I'm just talking about subject searches, something that libraries should be good at. They are far from doing this well. How much longer do we have to wait for a library to perform a massive search on all my ancestors who are recorded as having a career in some scientific field, and any indication of how well they did? Personally, I don't think it will be hard to beat any library out there, by our rules, because we're not playing the same game. We have different purposes. About patterns, just let me finish mathesis. Then you'll see a pattern, a very useful pattern. A universal pattern among knowledge. I'm sure I didn't qute answer your concern. Please refresh my memory. ciao, tomp ----- Original Message ----- From: Lee Howard To: Tom and other Packers Cc: Sent: Monday, August 14, 2000 9:14 PM Subject: Re: Markets for a Unified Model of Knowledge In order to serve the purpose of something better than a library, OM must automate the knowledge acquisition process and significantly improve, filter, or streamline it. And for what I know, the way to do that is for a pattern to be established. -------------------------------- End of general-digest-list Digest V00 Issue #8 **********************************************