From - Tue Feb 20 18:08:16 2001 X-UIDL: 489bf33cfaad4242 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 70185 invoked by uid 800); 20 Feb 2001 14:11:02 -0000 Date: 20 Feb 2001 14:11:02 -0000 Message-ID: <20010220141102.70184.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: <"volume00/9"%general-digest-list-request@onemodel.org> Subject: archive retrieval: volume00/9 Precedence: bulk Content-Disposition: inline; filename="volume00/9" Content-Type: message/rfc822; directory="volume00"; name="9" MIME-Version: 1.0 From: general-digest-list-request@onemodel.org Subject: general-digest-list Digest V00 #9 X-Loop: general-digest-list@onemodel.org X-Mailing-List: archive/volume00/9 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 9 Today's Topics: Re: Markets for a Unified Model of K [ Lee Howard ] Re: Modeling Relations -- Re: use ca [ "Tom and other Packers" To: "Tom and other Packers" Cc: "Jared (h) Norman" , "OM List" Subject: Re: Markets for a Unified Model of Knowledge Message-Id: <3.0.6.32.20000815222028.0080c150@server.deanox.com> Content-Type: text/plain; charset="us-ascii" > 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. Using BCC would have done that... Tom, I've never noticed you using BCC before. What's gotten into you? Um, actually I figured as much... only I thought it strange that you'd address Jared and then BCC him. Kind of a "I'm talking to someone, but you can't feeling." Hi Jared. :) > 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.) I could probably tolerate enough data entry myself to have created this database on my own, but honestly, I don't know or understand the basics of the data organization. We used to call them internodes. I don't know what they are or what they mean, and I don't know how to create a data file without that. Sure, I could plug in my family genology. That wouldn't get me far with the notice-a-pattern objective, though. ... >What can an >algorithm acting on Om do that another algorithm acting on a library can't >do? Isn't this my question? (or are you reiterating it?) > 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: .... (snip a very long and provocative requirement) .... > 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. Well, don't get me wrong. I'm not heralding libraries as the source of all knowledge. However I am suggesting that in practice, the current ambition of OM seems mighty similar to a library with a neat search utility. No, there are no libraries that I know of that allow you to do such a search, and I'm in complete agreement that libraries aren't particularly intended to do such things... at the moment. I do think, though, that with the evolution of the library along with the rest of everything that the library will pose a significant threat to OM independent success. Think about it. If you made electronic texts of all the books and then provided a search engine that allowed you to search on context as well as literals in a very sophisticated way, you wouldn't be far from OM's goal. At least one could accomplish very similar things in a very similar amount of time. I'll bet that you've used a library far more than I have, Tom, but it's been a long time since I used a library for finding literature. If I want literature I buy a book. When I'm in a library I'm doing research. I don't read books in a library, and I don't check books out from a library very often (I go buy them instead), but I do roam from book to book to book looking for a piece of knowledge. When I find what I'm looking for, I put the book away, having hardly read more than a chapter. I seem to treat libraries as the internet in that way. And if I had an electronic search engine that could do that leg-work for me... believe me, things would go much faster. Today's library is more than a source for literature. If you include the archive of periodicals, all you have to do is begin including the right periodicals, and Tom, you could have called the library and they would have been able to tell you about that non-fireside that you attended. As far as fiction vs. nonfiction we've got a problem. Don't get me wrong, I'm completely in agreement that there is fiction and that there is nonfiction (truth exists), but the whole purpose of OM is to provide the user an ability to discern that. Isn't OM supposed to help filter out fact from fiction? Again, to do that there must be a pattern to follow, and that pattern must be allowed growth. Although rudimentary, a library already has that pattern established by its organization. Nobody goes to a library and gets Huckleberry Finn believing it to believing it to be a true story. However, how will OM know when it's being fed with Huckleberry Finn-like information? As it now seems, the superuser has to tell it that it is fiction. And what's the point of having to tell OM something that it will tell us back? Nobody needs OM to tell them something that they already know. How do you discern fact from fiction? These are all illustrations of the need for a model, a pattern, an algorithm... and we all know that this is the crutch of the matter. We could talk until we're blue in the face about this, and we could program the prettiest program the world's ever seen for storing and retrieving 'lil bits of data. We could even spend our lives filling that database chock full of library-like information. Yet without a theory of what this all means, though, who cares? Or is my thinking wrong, and is OM only a database with a cool search utility? Are we just looking for patterns here, or are we trying to define and circumscribe truth? And if we're seeking truth, how do we program it to discern truth from error without telling it beforehand? There's a basic theory here that we're missing, and if we expect this basic theory of truth to withstand growth and hordes of misinformation and yet absorb and encompass accuracies, then we cannot base this theory from observed patterns or we perpetuate the degree of truth found in that pattern, which may be great, but in the extremeties of our searching a small degree of inaccuracy will amount to great failure. I'm asking to be disproved. That would motivate me. I want to know that there is a definable pattern. I want to know that there's a way to tell a machine to learn truth without communicating with the source. Tell me there is; show me its possiblity and I will believe that OM can be more than a futuristic library. .... > 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". In essence, telling you what you already know - perpetuating a possibly flawed or incomplete 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. .... (snip a number of good illustrations of the failure in current search utilities) .... Which is what I meant by a "context search". The essential problem is that the search is carried out over the text of a descriptor rather than the text of the book. And the basic reason for this is to somehow achieve a vague type of actual context search which content searching doesn't do because context and content are different things. Could someone sit down and develop a better context search method. Yes. And when that happens, watch out OM. > 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. Like I said above, as far as knowledge goes, I use a library for finding more. What was OM's goal, did you say? Is it truly different? > About patterns, just let me finish mathesis. Then you'll see a pattern, >a very useful pattern. A universal pattern among knowledge. Which is just what I need. > I'm sure I didn't qute answer your concern. Please refresh my memory. I hope this helps. And I hope that all of this is seen as debate rather than argument. I'm taking a position and expressing it. I'm not necessarily expressing anything that's dear to my heart. Lee. ------------------------------ Date: Wed, 16 Aug 2000 10:00:45 -0600 From: "Tom and other Packers" To: "OM List" Subject: Re: Modeling Relations -- Re: use cases Message-ID: <004c01c0079b$8f68a8a0$1d0aa8c0@oemcomputer> Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Luke MetaInternodes can be pretty messy. I haven't completely worked them out in my own mind yet, so ... don't take this as "gospel truth". But this should give Lee the beginnings of my "model of truth" called Mathesis, if not an introduction to a small part of my "theory of truth" called "Em-Veh"; the full introduction will come later. First, I must explain that there are two types of languages, according to Em-Veh, (1) geometric and (2) symbolic. I propose that all knowledge is representable using one, or a combination, of these two types of languages, AND I propose that all knowledge can also be modelled using the three primary construct in mathesis. (These two propositions are called "conjectures" in mathematics, so don't ask me for the "proof" yet.) Therefore the mathetical constructs nodes, internodes, and metainternodes have analogues in the two language types. Thinking geometrically, (1) nodes are points (or collections of points, like lines), (2) internodes are generalisations of vectors, (3) metainternodes are generalisations of the combination of the following two things: (a) angles between vectors and (b) a thing that resembles a vector, which happens to record the displacement between vectors instead of points. Thinking symbolically, (1) nodes are symbols, (2) internodes are symbolic relations between symbols, (3) metainternodes are symbolic relations between symbolic relations between symbols. (By the way, Mathesis is a metalanguage, of both geometric and symbolic languages. And, by the way, I've not discovered a need to go beyond the three constructs; e.g. meta-meta-meta-meta-meta-internodes would be more intellectually costly than beneficial. But there is something called MetaMathesis, which is like a meta-meta-language, in which there is something like a meta-meta-internode defined, but it applies to all three constructs equally, and therefore is not an extension in the same linguistic dimension as metainternodes, but is an extension in an orthogonal meta-linguistic dimension; but that's another story.) One example of a metainternode is the following (facilitated by the geometric perspective): We have two simple and rather specific temporal relations, like R1 = "10 minutes after", and R2 = "20 minutes after". A third relation can easily be invented which is a generalisation of these two relations: R3 = "N minutes after". It's a generalisation of R1 and R2 because it lacks specificity. All three internodes can be represented by vectors pointing in the "before" direction of the time dimension, but the lengths of these vectors vary among the two precise values of "10 minutes" and "20 minutes", and the "N minutes". R1 and R2 can be related to R3 through internodal generalisation. R1 and R2 can be related to each other with something that happens to look a lot like a vector: "R2 is '10-more-minutes-after' than the 10-minutes-after of R1". The thing that relates any two of these internodes is a metainternode. All of these metainternodes indicates that there is a commonality among all three relations, namely direction; and that there is a variability in length. The metainternode which relates R1 and R3 thusly specifies that the direction of the two vectors is the same, but the length is different, e.g. length is more precise in R1, and less precise in R3. In this, case the "angle" aspect of metainternodes does not come into play (or you could say that the angle-component of these metainternodes is zero); rather, the "generalisation of something like a vector" is what I would use to describe these metainternodes. By "generalisation of a vector", I usually mean that there is information added to what a vector normally contains. In this case, information is added to the length-component of the vector, such that there is a range of possible lengths, not just one precise length. Another example of a metainternode: An "-is-a-type-of-" relation and an "-is-a-component-of-" relation can both be said to point from a node of small volume in mathetical space to a node of larger volume. E.g. they are nodal generalisations, or generalisation relations. These two relations don't point in exactly the same direction, as vectors, but given some convention of combining these two dimensions into one many-dimensional geometry, there is a way to relate the two dimensions. The angle between these two dimensions is probably going to be a right-angle, but it will be a right angle in some predefined angular direction, to distinguish it from other orthogonal pairs of dimensions. This illustrates one possible use of the "angle" aspect of the metainternode. I wanted to find a more symbolic example, but it didn't work out this time. I guess I get stuck in geometry a lot lately, which is good and bad. If we can find a well-founded geometric system in which all knowledge can be modelled, we could make use of a lot of Euclidean geometry, non-Euclidean geometry, analytic geometry, etc. to make precise, concrete inferences. But I fear that this is not practical for certain parts of a large body of very subjective, abstract knowledge, in which case we will be forced to use the more abstract structure of symbolic language some of the time. Each of these two types of language has its strengths and its weaknesses, and I believe that the strength of human conception and reasoning is accomplished by what the psychologists call "lateral thinking". And yes, by this, I am implying that the symbolic and the geometric languages correspond to the left and the right hemispheres of the human cerebrum, respectively. Actually, here's a quick example of a symbolically-facilitated look at metainternodes: You have a relation R1, "-is-the-brother-of-", and the relation R2, "-is-the-grandparent-of-". These two relations would be classified as "family relations", and therefore could be related to a more generic "-is-related-to-" relation -- using metainternodes, of course. They could also be related to each other using another sort of metainternode. Does this work? Reading more of Luke's letter, I see that you pretty much understood this stuff already. As for more examples of queries ... I keep going back to the same type of example: a query which uses a generality-metainternode. I'm sure there are other examples, but this one is very useful. I'll now put it in the context of my last example, familial relations. Say you wanted to locate each person B, defined as being close relatives of person A. I.e. "find each B who is related to A, in a family-tree sort of relation, within a certain distance D". (There would have to be some graph-theoretical convention established first, for comparing the genealogical "distances" or "lengths" of relations which connect people in different directions, e.g. "vertically" and "laterally" in the family tree, but that's a different story.) You also want the query to list exactly how each B is related to A, e.g. "-is-the-mother-of-", "-is-the-brother-of-", etc. Your database is not going to record two relations between A and every B-candidate, but the program needs two relations for this particular query: it needs (1) the generic "-is-related-to,-within-a-distance-D-" relation, so it knows when to return B as a result of the query, and it needs (2) the exact relation between A and B. There are two levels of relational generality here. Instead of storing both types of relations for B, the program would probably store the more precise, specific relation, and then related it to the more general relation specified in the query, at the time of the query. Personally, I think you should relate the generic and the specific relation with metainternodes, but I'm sure someone out there in the genealogy business has come up with another solution for this minor problem, (right Mark?). More importantly, if you have incomplete information, such that the relation between a B-candidate and A is not known explicitly, you will have another use for metainternodes. You may have enough peripheral information, in the form of relations between A and other people, C's. (In graph theory, this would be a path of length two between A and B, instead of the ideal path of length one.) You could then find the specific relation, the path of length one from B to A, recorded implicitly within these other relations, (through deductive (syllogistic) inference in symbolic language, or through vector addition in geometric language). For example, you may not have the fact that B is the uncle of A; you may only have the two facts that C is the father of A, and B is the brother of C. I believe you need metainternodes to meta-relate all of these relations in such a way that a transitive-closure graph can be temporarily constructed for the purpose of the query. In this graph, the exact, single relation between A and B can be displayed. "B-is-the-uncle-of-A" is the relation constructed from the transitive closure of two relations: "B-is-the-brother-of-C" and "C-is-the-father-of-A". But how is the program going to know this unless you have some way of relating relations? Is this not way cool? Going on ... I agree that we should be allow inconsistencies to exist in the model, and let the user correct them if/when he wants to. Fact and Fiction. More on that later. ciao, tomp ----- Original Message ----- From: Luke Call To: Sent: Tuesday, August 15, 2000 7:53 AM Subject: Re: Modeling Relations -- Re: use cases Tom, can you give a detailed example or two of meta-relations? 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? 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. ------------------------------ Date: Wed, 16 Aug 2000 09:40:45 -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: <004b01c0079b$8c649060$1d0aa8c0@oemcomputer> Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit 2000.08 Lee A few disagreements: I think we're using the word "literature" differently. I'm using it in a broad sense, in which it means every instance of written language, including books, magazine articles, pamphlets, etc. I don't think any library would have told me that there was not a CES fireside that Sunday, except perhaps the family history library or the BYU library, and in those cases only because I'd expect one of the employees to happen to know, on his own, and not because the library itself actively sought such "temporary" knowledge. Furthermore, it was just an example. We can all think of information that we can't find anywhere, even though it's theoretically plausible that someone in the world knows this knowledge. I completely disagree with the statement that there is no value in Om telling us back what we already know. From one perspective, that's all Om or any other model of knowledge ever does. The value is not in coming up with new knowledge, per se, but in having all the previously existing knowledge in one place, and in having algorithms which automatically find instances of one particular bit of information in a context that we hadn't considered before, (I don't mean bit in the binary digit sense). In one use there is a reorganisation of information, sure. And you could say that it comes up with new information. But in the other, there is no fundamental reorganisation. It's supposed to look the same as how it was entered; the difference is the person who is looking at it is not the same person who entered it. There is value in that. But if that doesn't motivate you ... you're sounding like the animals in the story about the red hen. You want the finished project without the work required in making the less-useful intermediate model. We have to start somewhere, Lee. Simple, redundant (library-like) model of information first. Massively powerful theory of truth and oracle of the gods, second. And about fact and fiction, I fear you are guilty of a cardinal sin: compartmentalisation. If there were no truth in Huckleberry Finn at all, no correspondence in any way to reality, what value would it have? It would not only be of no value, it would be meaningless and unintelligible. And what of parables, and allegories like "Animal Farm", and the children's moralistic story about the red hen and the lazy animals, and all other instances of the blurred line between fact and fiction? Single human beings can't always take the time to compare a story to (1) the authors life history, (2) world history, (3) gospel doctrine, (4) any other arbitrary branch of knowledge. And what of theories and hypotheses, which are supposed to be factual, but which are mostly fictional, aesthetically-appealing simplifications of the data someone has gathered? And what of old knowledge, such as "the world is round", which was fact, but then became disbelieved, and which later became fact again? And what of any "fact"? Are we 100% certain of anything? Sure, we have our personal convictions, but I believe that it is impossible to prove to someone else even those things which we, ourselves, feel we have 100% certainty about. We ought to have the ability to model fact and fiction for three reasons: (1) just to keep a record of all possible instances of knowledge, just so these ideas are around and testable, and we don't loose any support for or against an idea, so that in the end we may have amassed enough information that the thought in question can safely be regarded as fact or fiction, with or without the help of a special truth-finding algorithm, (2) so we can use the subtle and oft-times hidden instances of truth, even in fiction, for the development of theories of psychology, mythology, and everything else that "fiction" may be found to be relevant to, (3) just so we have knowledge of what art, or what fiction, could be beneficial or enjoyable solely for the purpose of the enjoyment of fiction. So, right now, there are values in telling the OM that it is "fictional", manually. And later, there will be added benefit when it can figure this out itself (using other knowledge which we had to "feed" it). There is _no_ way for OM to find truth, the way you seem to want. There. Is that what you've been waiting to hear? But there is value in OM, by it simply reconciling thoughts that we feed it, or even by showing us where the inconsistencies are, and letting us reconcile them manually. Good. I hope this isn't dear to your heart. In other words, why do we have to be in competition with the well-advanced library of the future? Why can't we take some [good] pride in being among those who may develop that perfect library? And as far as a theory of truth, we haven't been talking long enough for such a theory to come out, even if one of us had one. Actually, we all do have one. We all have an epistemology. Some of us like to systematise it more than others. You're just going to have to wait until I finish either (1) The Em-Veh Treatise, or (2) The MetaMathesis Essay before you can say that we (that I) don't have a _good_ theory of knowledge and truth. But before that, please give us the benefit of the doubt, unless there's something specific and already-vocalised that you can be the gadfly on. It's a bit disconcerting to feel I have to defend something that I think has been criticised before the gadfly has even been introduced to it. And, as Jared says, it's easier to tear (sp?) down than to build. If the fact that our theory of truth isn't finished means that you are disinterested in the project, I think you ought to have faith that it might turn out to be the best thing (next to the scriptures) that you've ever read -- and just help us now anyway. tomp ----- Original Message ----- From: Lee Howard To: Tom and other Packers Cc: Jared (h) Norman ; OM List Sent: Tuesday, August 15, 2000 10:20 PM Subject: Re: Markets for a Unified Model of Knowledge Well, don't get me wrong. I'm not heralding libraries as the source of all knowledge. However I am suggesting that in practice, the current ambition of OM seems mighty similar to a library with a neat search utility. No, there are no libraries that I know of that allow you to do such a search, and I'm in complete agreement that libraries aren't particularly intended to do such things... at the moment. I do think, though, that with the evolution of the library along with the rest of everything that the library will pose a significant threat to OM independent success. If I want literature I buy a book. When I'm in a library I'm doing research. Today's library is more than a source for literature. If you include the archive of periodicals, all you have to do is begin including the right periodicals, and Tom, you could have called the library and they would have been able to tell you about that non-fireside that you attended. As far as fiction vs. nonfiction we've got a problem. Don't get me wrong, I'm completely in agreement that there is fiction and that there is nonfiction (truth exists), but the whole purpose of OM is to provide the user an ability to discern that. Isn't OM supposed to help filter out fact from fiction? Again, to do that there must be a pattern to follow, and that pattern must be allowed growth. Although rudimentary, a library already has that pattern established by its organization. Nobody goes to a library and gets Huckleberry Finn believing it to believing it to be a true story. However, how will OM know when it's being fed with Huckleberry Finn-like information? As it now seems, the superuser has to tell it that it is fiction. And what's the point of having to tell OM something that it will tell us back? Nobody needs OM to tell them something that they already know. How do you discern fact from fiction? These are all illustrations of the need for a model, a pattern, an algorithm... and we all know that this is the crutch of the matter. Or is my thinking wrong, and is OM only a database with a cool search utility? Are we just looking for patterns here, or are we trying to define and circumscribe truth? And if we're seeking truth, how do we program it to discern truth from error without telling it beforehand? There's a basic theory here that we're missing, and if we expect this basic theory of truth to withstand growth and hordes of misinformation and yet absorb and encompass accuracies, then we cannot base this theory from observed patterns or we perpetuate the degree of truth found in that pattern, which may be great, but in the extremeties of our searching a small degree of inaccuracy will amount to great failure. I'm asking to be disproved. That would motivate me. I want to know that there is a definable pattern. I want to know that there's a way to tell a machine to learn truth without communicating with the source. Tell me there is; show me its possiblity and I will believe that OM can be more than a futuristic library. Which is what I meant by a "context search". The essential problem is that the search is carried out over the text of a descriptor rather than the text of the book. And the basic reason for this is to somehow achieve a vague type of actual context search which content searching doesn't do because context and content are different things. Could someone sit down and develop a better context search method. Yes. And when that happens, watch out OM. Like I said above, as far as knowledge goes, I use a library for finding more. What was OM's goal, did you say? Is it truly different? -------------------------------- End of general-digest-list Digest V00 Issue #9 **********************************************