francia@nmt.edu (Guillermo A. Francia) (05/31/91)
In as much as there are a number of means to represent (or code) knowledge in AI, I would like to know if anyone has done any research on a UNIFIED (mathematical) model of knowledge representation. Does anyone out there know of any work done on this? Will appreciate any help that I can get. Thanks. -- -------------------------------------------------------------------- francia@jupiter.nmt.edu Guillermo A. Francia III P.O. Box 2335 CS francia@minos.nmt.edu New Mexico Tech
aarons@syma.sussex.ac.uk (Aaron Sloman) (06/02/91)
francia@nmt.edu (Guillermo A. Francia) writes: > Date: 31 May 91 16:46:08 GMT > Organization: New Mexico Tech > > > In as much as there are a number of means to represent (or code) > knowledge in AI, I would like to know if anyone has done any > research on a UNIFIED (mathematical) model of knowledge representation. > Does anyone out there know of any work done on this? > > -------------------------------------------------------------------- > francia@jupiter.nmt.edu Guillermo A. Francia III > P.O. Box 2335 CS > francia@minos.nmt.edu New Mexico Tech Some of the people who have worked on first order logic have thought it could serve as the universal (= unified?) notation for representing knowledge. I don't know if this is what you mean by a model. However, it seems pretty obvious from the history of science and culture that different formalisms are useful for different purposes, including, logic, algebra, pictures, maps, tables, flow-charts, musical notation, 3-D models, natural languages, etc. One reason for this is that different kinds of notations have different kinds of variability, which limits their expressive power in different ways. This can sometimes be important when exploring large search spaces. If the structure of the notation is such that it won't let you express things that would only be rejected anyway, it can have great heuristic power. Equally if the structure of the notation is such that certain frequently used forms of inference can be done with simple algorithms, that is also useful. (E.g. roman numerals are very good for addition: simply concatentate the numerals, provided that the numbers you are adding total no more than III ! Arabic numerals are an ad-hoc but very useful notation, good for a wider range of operations, but not all that powerful when it comes to taking square roots, dividing large numbers that are not powers of 10, etc.) AI has so far explored only a tiny subset of possible forms of knowledge representation, mostly the ones that are easiest to manipulate using current computers and programming languages. I don't think that anyone will ever find a "UNIFIED (mathematical) model of knowledge representation" unless it takes the form of a sort of meta-theory explaining why different models are needed for different knowledge domains and different kinds of uses of knowledge. If anyone claims to have a unified model, ask him/her if it will serve for the purposes of representing knowledge about the contents of the current optic array in a robot's visual system, and for transforming that knowledge in the process of discovering what's out there (e.g. discovering binocular disparities for stereo vision) or for fine control of posture and actions, etc. Aaron Sloman, School of Cognitive and Computing Sciences, Univ of Sussex, Brighton, BN1 9QH, England EMAIL aarons@cogs.sussex.ac.uk or: aarons%uk.ac.sussex.cogs@nsfnet-relay.ac.uk
sticklen@pleiades.cps.msu.edu (06/03/91)
In article <5255@syma.sussex.ac.uk> aarons@syma.sussex.ac.uk (Aaron Sloman) writes: >francia@nmt.edu (Guillermo A. Francia) writes: > >> Date: 31 May 91 16:46:08 GMT >> Organization: New Mexico Tech >> >> >> In as much as there are a number of means to represent (or code) >> knowledge in AI, I would like to know if anyone has done any >> research on a UNIFIED (mathematical) model of knowledge representation. >> Does anyone out there know of any work done on this? >> >> -------------------------------------------------------------------- >> francia@jupiter.nmt.edu Guillermo A. Francia III >> P.O. Box 2335 CS >> francia@minos.nmt.edu New Mexico Tech > >Some of the people who have worked on first order logic have thought >it could serve as the universal (= unified?) notation for >representing knowledge. I don't know if this is what you mean by a >model. > >However, it seems pretty obvious from the history of science and >culture that different formalisms are useful for different purposes, >including, logic, algebra, pictures, maps, tables, flow-charts, >musical notation, 3-D models, natural languages, etc. > >One reason for this is that different kinds of notations have >different kinds of variability, which limits their expressive power >in different ways. This can sometimes be important when exploring >large search spaces. If the structure of the notation is such that >it won't let you express things that would only be rejected anyway, >it can have great heuristic power. > >Equally if the structure of the notation is such that certain >frequently used forms of inference can be done with simple >algorithms, that is also useful. (E.g. roman numerals are very good >for addition: simply concatentate the numerals, provided that the >numbers you are adding total no more than III ! Arabic numerals are >an ad-hoc but very useful notation, good for a wider range of >operations, but not all that powerful when it comes to taking square >roots, dividing large numbers that are not powers of 10, etc.) > >AI has so far explored only a tiny subset of possible forms of >knowledge representation, mostly the ones that are easiest to >manipulate using current computers and programming languages. > >I don't think that anyone will ever find a > "UNIFIED (mathematical) model of knowledge representation" >unless it takes the form of a sort of meta-theory explaining why >different models are needed for different knowledge domains >and different kinds of uses of knowledge. > >If anyone claims to have a unified model, ask him/her if it will >serve for the purposes of representing knowledge about the contents >of the current optic array in a robot's visual system, and for >transforming that knowledge in the process of discovering what's out >there (e.g. discovering binocular disparities for stereo vision) or >for fine control of posture and actions, etc. > >Aaron Sloman, >School of Cognitive and Computing Sciences, >Univ of Sussex, Brighton, BN1 9QH, England > EMAIL aarons@cogs.sussex.ac.uk >or: > aarons%uk.ac.sussex.cogs@nsfnet-relay.ac.uk > I'd like to comment on Aaron's thrust that "unified mathematical models of representation" may be misguided (my word, not Aaron's) and that if some unification is possible it will take place at the level of an architecture that selects problem solving methods for given use. The point of what in knowledge-based systems is being called "task specific architectures" (TSAs) is to suggest that problem solving is best analyzed and "mimiced/implemented" by using primitives that are used in the domain of the problem solving. For example, in a diagnostic domain, it is natural to both analyze diagnostic problem solving, and to build computer versions of diagnostic problem solving, by using the concept of "diagnostic hypothesis". Specializations of the TSA concept like Chandrasekaran's generic tasks (GTs) take one additional step of saying that there exists a finite set of problem solving strategies that are generally useful. These GTs are defined by giving both a knowledge representation template, and a control strategy. For example, there is a GT for classification problem solving, one for simple ("routine") design, one for function-based model level reasoning... At the same time as researchers seek to extend the capability of TSAs like those in the generic tasks, grand architectures for problem solving are being developed to handle any problem solving situation - SOAR may be the easiest example. SOAR generates problem spaces appropriate for a given problem. Although not a developed capability yet, the SOAR architecture may one day be able to generate very tailored problem spaces after analyzing a given problem situation - perhaps problem spaces not unlike GTs. If that happens, then it will be a unification. But a unification along the lines that Arron pointed to, not a unification at the base level of saying "one representation fits all." Jon Sticklen AI/KBS Laboratory
ISSSSM@NUSVM.BITNET (Stephen Smoliar) (06/04/91)
In article <5255@syma.sussex.ac.uk> aarons@syma.sussex.ac.uk (Aaron Sloman) writes: > >Some of the people who have worked on first order logic have thought >it could serve as the universal (= unified?) notation for >representing knowledge. I don't know if this is what you mean by a >model. > >However, it seems pretty obvious from the history of science and >culture that different formalisms are useful for different purposes, >including, logic, algebra, pictures, maps, tables, flow-charts, >musical notation, 3-D models, natural languages, etc. > >One reason for this is that different kinds of notations have >different kinds of variability, which limits their expressive power >in different ways. This can sometimes be important when exploring >large search spaces. If the structure of the notation is such that >it won't let you express things that would only be rejected anyway, >it can have great heuristic power. > >If anyone claims to have a unified model, ask him/her if it will >serve for the purposes of representing knowledge about the contents >of the current optic array in a robot's visual system, and for >transforming that knowledge in the process of discovering what's out >there (e.g. discovering binocular disparities for stereo vision) or >for fine control of posture and actions, etc. > In article <1991Jun3.110003.13773@msuinfo.cl.msu.edu> sticklen@pleiades.cps.msu.edu (Jon Sticklen) then runs with this ball, writing: > >The point of what in knowledge-based systems is being called "task specific >architectures" (TSAs) is to suggest that problem solving is best analyzed and >"mimiced/implemented" by using primitives that are used in the domain of the >problem solving. For example, in a diagnostic domain, it is natural to both >analyze diagnostic problem solving, and to build computer versions of >diagnostic problem solving, by using the concept of "diagnostic hypothesis". >Specializations of the TSA concept like Chandrasekaran's generic tasks (GTs) >take one additional step of saying that there exists a finite set of problem >solving strategies that are generally useful. These GTs are defined by giving >both a knowledge representation template, and a control strategy. For example, >there is a GT for classification problem solving, one for simple ("routine") >design, one for function-based model level reasoning... > >At the same time as researchers seek to extend the capability of TSAs like >those in the generic tasks, grand architectures for problem solving are being >developed to handle any problem solving situation - SOAR may be the easiest >example. SOAR generates problem spaces appropriate for a given problem. >Although not a developed capability yet, the SOAR architecture may one day be >able to generate very tailored problem spaces after analyzing a given problem >situation - perhaps problem spaces not unlike GTs. > >If that happens, then it will be a unification. But a unification along the >lines that Arron pointed to, not a unification at the base level of saying >"one >representation fits all." > I think this approach may be missing the point Aaron was trying to make. Generic tasks and problem spaces are, once the dust of surface appearance is swept aside, just as much mathematical objects as are the constructs of first-order logic. The difficulty lies in attempting the act of reductionism itself, rather than in the particular construct which is the target of your reduction. Aaron offered up a nice list of different formalisms which have been engaged for different purposes of reasoning. However, the more we begin to recognize the importance of viewing reasoning as SITUATED, the more difficult it becomes to carve off "the reasoning itself" as an object of study. This opens the door to the intimidating prospect that we need to be concerned with more general issues of BEHAVIOR, which means that the formalisms in Aaron's list are only scratching the surface. Consider this problem from another point of view. Regardless of whether or not we think it is feasible to build machines which satisfy "behavioral criteria for intelligence" (whatever those criteria may be), let us simply worry about whether or not we can DESCRIBE intelligent behavior. Let me say right off that I shall be the first to raise a skeptical eyebrow at any claim that this problem has been solved in any practical way. The fact is that psychology is still, to a great extent, floundering around simply trying to DESCRIBE many of the phenomena it wishes, ultimately, to explain. For many limited domains of scientific reasoning, we can at least fall back on foundations of formalisms such as those Aaron has enumerated; but when we try to take on behavior (even when intelligence isn't directly involved), we might as well be in a raft in the middle of the North Atlantic Ocean. I would like to pose a possibly radical explanation for why we find ourselves so lost here, and that is that THERE IS NO SUCH THING AS OBJECTIVE DESCRIPTION. Description is highly subjective to the person doing the describing. If anyone else wishes to draw upon that description, he is obliged to enter into a relatively sophisticated process of negotiation which is known as communicating in a natural language. (Anyone who is interested in seeing blatant examples of how subjectivity lurks in seemingly objective accounts should take a look at "Apes R Not Us," a review of primate studies by Lord Zuckerman which appeared in the May 30 issue of THE NEW YORK REVIEW. Also relevant is Brian Smith's response to Lenat and Feigenbaum, "The Owl and the Electric Encyclopedia," published in Volume 47 of ARTIFICIAL INTELLIGENCE.) In the face of such a daunting proposition, it may be that the quest for a "unified representation" is not so much misguided because of the "unified" attribute as because of the very goal of representation itself. We are now entering a period of skepticism regarding the issue of representation. Such skepticism is encouraged by the work of Smith, Rodney Brooks, and others far too numerous to mention. If any of these researchers can demonstrate results which scale up from small experiments to practical problems of getting on in the world, our current obsession with representation may ultimately be dismissed as a distracting side-track. =============================================================================== Stephen W. Smoliar Institute of Systems Science National University of Singapore Heng Mui Keng Terrace, Kent Ridge SINGAPORE 0511 BITNET: ISSSSM@NUSVM
ewc@hplb.hpl.hp.com (Enrico Coiera) (06/04/91)
Steven Smoliar's follow-up contains an interesting suggestion ie that 'there is no such thing as subjective description. Description is highly subjective to the person doing the subscribing'. A similar view is shared by Paul Compton (see P. Compton, R. Jansen, A Philosophical basis for Knowlege Aquisition, Knowledge Aqusition, (1990), 2, 241-257). He suggests that there is little point in attempting to structure the knowledge elicited from experts because each expert 'makes-up' the knowledge as it were in the context of a particular problem. While each individual has an internal representation of some form, the way in which it is communicated to others is highly context dependent - the knowledge elicited in context only has validity in that context - it makes no sense to attempt to represent it in a way in which it can be considered context independent. Enrico Coiera Hewlett Packard Labs Filton Rd Stoke Gifford, Bristol BS12 6QZ United Kingdom
jj@medulla.cis.ohio-state.edu (John Josephson) (06/05/91)
Steven Smoliar> THERE IS NO SUCH THING AS OBJECTIVE DESCRIPTION Careful lest next you decide that no description is any better than any other, since they are all subjective, in which case (paraphrasing Bob Dylan) there is no point talking to you, it's just the same as talking to anybody. If, indeed, some descritions are better than others, how can this be? Maybe some are more objective than others. .. jj
ISSSSM@NUSVM.BITNET (Stephen Smoliar) (06/06/91)
In article <JJ.91Jun5122616@medulla.cis.ohio-state.edu> jj@medulla.cis.ohio-state.edu (John Josephson) writes: > >Steven Smoliar> THERE IS NO SUCH THING AS OBJECTIVE DESCRIPTION > >Careful lest next you decide that no description is any better than >any other, since they are all subjective, in which case (paraphrasing >Bob Dylan) there is no point talking to you, it's just the same as >talking to anybody. > The trouble with emphasizing an entire sentence is that it distracts the casual reader from the subsequent elaboration of that sentence. I fear I had better repeat the following two sentences, since I feel they are more to the point than John's attempt at a witty cautionary remark: >Description is highly subjective to the person doing the describing. If >anyone >else wishes to draw upon that description, he is obliged to enter into a >relatively sophisticated process of negotiation which is known as >communicating >in a natural language. In other words if you want to talk with me (or anyone else) you first have to WANT to engage yourself in this "process of negotiation." If you approach dialog without such a disposition to negotiation, then John is quite right. In that frame of mind, there is no point in your talking to anyone; the experience will be tantamount to talking to yourself. >If, indeed, some descritions are better than others, how can this be? >Maybe some are more objective than others. > In any given situation of discourse, we, as outside observers, may be able to say that certain exchanges involve LESS negotiation than others. However, I would be reluctant to say that those exchanges are based on "better" descriptions. The amount of negotiation which is required is more dependent on the context of the discourse than on any characteristic attributes of the descriptions being invoked. Take a look at the cited Brian Smith article before taking your next shot from the hip, John. :-) =============================================================================== Stephen W. Smoliar Institute of Systems Science National University of Singapore Heng Mui Keng Terrace, Kent Ridge SINGAPORE 0511 BITNET: ISSSSM@NUSVM "The funny thing about being smug about health and purity all the time is that it can turn you into a fascist."--Paul Theroux
is_s425@kingston.ac.uk (Hutchison C S) (06/10/91)
re: Steven Smoliar> THERE IS NO SUCH THING AS OBJECTIVE DESCRIPTION I missed earlier messages, so maybe I'm barking up the wrong tree here. I'm starting work in the area of descriptions and truth. Basically, the problem is this: given four descriptions of the same event (in this case, events which took place on 2nd June 1975 on the outskirts of what was then Salisbury, Southern Rhodesia) -- "Rioting blacks shot dead as ANC leaders meet" (Times of London) "Police shoot 11 dead in Salisbury riot" (Manchester Guardian) "Rebels kill 11 ANC men" (Times of Zambia) "Racists murder Zimbabweans" (Tanzanian Daily News) each report is either true (corresponds to a state of affairs in the world) or false (does not correspond). By virtue of the meanings of the words in the sentences, they cannot all be true at the same time. Yet they all report the 'same' events. My hunch is to say that the physical circumstances under- determine possible linguistic descriptions, enabling various ideological interpretations of the events. What then do readers 'know' about the events? How is knowledge in this sense distinct from mere belief? Any thoughts?
cpshelley@violet.waterloo.edu (cameron shelley) (06/11/91)
In article <1991Jun10.094754.3303@kingston.ac.uk> is_s425@kingston.ac.uk (Hutchison C S) writes: [...] >My hunch is to say that the physical circumstances under- >determine possible linguistic descriptions, enabling various ideological >interpretations of the events. What then do readers 'know' about the events? >How is knowledge in this sense distinct from mere belief? > >Any thoughts? Eduard Hovy produced a text generation system "PAULINE" that generated different accounts of the same event (a student action at Yale) from different points of view. You should consider reading about that system. The lesson to be learned from that system, is that there many ways of phrasing a single proposition, the choice of expression is then determined by what stylistic and political (read 'social') content you wish to place in the description. The situation itself (the material state-of- affairs?) does not contain such information, but the people concerned with it do have such views. I guess that if you want to try and filter out how people's views affect their descriptions, then you'll have to come up with a model that systematically relates the two. That's not so much a matter of 'truth' as socio-linguistics. Cam
ISSSSM@NUSVM.BITNET (Stephen Smoliar) (06/11/91)
In article <1991Jun10.175110.22654@watdragon.waterloo.edu> cpshelley@violet.waterloo.edu (cameron shelley) writes: >In article <1991Jun10.094754.3303@kingston.ac.uk> is_s425@kingston.ac.uk >(Hutchison C S) writes: >[...] >>My hunch is to say that the physical circumstances under- >>determine possible linguistic descriptions, enabling various ideological >>interpretations of the events. What then do readers 'know' about the events? >>How is knowledge in this sense distinct from mere belief? >> >>Any thoughts? > >Eduard Hovy produced a text generation system "PAULINE" that generated >different accounts of the same event (a student action at Yale) from >different points of view. You should consider reading about that system. > >The lesson to be learned from that system, is that there many ways of >phrasing a single proposition, the choice of expression is then determined >by what stylistic and political (read 'social') content you wish to >place in the description. The situation itself (the material state-of- >affairs?) does not contain such information, but the people concerned >with it do have such views. > >I guess that if you want to try and filter out how people's views affect >their descriptions, then you'll have to come up with a model that >systematically relates the two. That's not so much a matter of 'truth' >as socio-linguistics. > PAULINE was basically an exercise in rhetoric, demonstrating that an allegedly objective propositional account could be realized in a wide variety of texts, where that variety could be delimited according to some underlying rhetorical structures. I find it an excellent demonstration of the scope of rhetoric, but I question the premise that one can start with that objective propositional account. This may sound solipsistic, but the only accounts we can give of events of the world are based on perceptions, be they our own, those of "credible sources," or interpretations of devices. (Note that key word "interpretations." Sophisticated devices do not PERCEIVE the world. WE perceive the world through our ability to use those devices. For example, cloud chambers do not "perceive" subatomic particles but simply provide us with evidence of their existence which we are then obliged to interpret.) I think it is the lack of such an objective starting point that supports Cam's position that the issue here is not "truth" (so, yes, Hutchison IS "barking up the wrong tree") but sociology--specifically the role the inter-personal behavior contributes to communication. We cannot expect to run the rhetorical mechanisms of PAULINE "in reverse" in order to "distill" an objective account out of a given presentation. The best we can hope for is to strip out the rhetorical "interference" and get a more compact statement of what that particular individual perceived; but those perceptions are only "meaningful" to the extent that we know how that individual interpreted his sensations. Because this is asking a bit much for even HUMAN intelligence, I have been arguing my position based on NEGOTIATION: The only way I can figure out what you are talking about is to engage in dialog. In the course of that dialog, I develop hypotheses about your perceptions which I can then test by "probing" you with appropriate questions and remarks. This process never really "converges" to my having a "total" model of your perceptual interpretations; but it tends to provide enough information for the two of us to share our experiences of a common world. Hutchinson's example of four separate newspaper accounts makes negotiation a bit more tricky, since you cannot engage in a dialog with a newspaper. Fortunately, there are some viable alternatives. Newspapers tend to have editorial policies. Frequent exposure to the TIMES, supplemented by knowledge of their editorial pieces, can also be used as a source of hypotheses about their perceptions of world events. Such hypotheses can often be tested by the very basis of Hutchinson's experiment--comparing their account of a story with those of other papers. This is the sort of reading which is required in order to extract the news from a newspaper (as anyone who has ever been the subject of a newspaper articles knows); and I would argue that it is the "logical equivalent" (using those words VERY informally) of the way we use dialog in normal discourse. =============================================================================== Stephen W. Smoliar Institute of Systems Science National University of Singapore Heng Mui Keng Terrace, Kent Ridge SINGAPORE 0511 BITNET: ISSSSM@NUSVM "He was of Lord Essex's opinion, 'rather to go an hundred miles to speak with one wise man, than five miles to see a fair town.'"--Boswell on Johnson
byland@iris.cis.ohio-state.edu (Tom Bylander) (06/11/91)
In article <9106110020.AA17886@lilac.berkeley.edu> ISSSSM@NUSVM.BITNET (Stephen Smoliar) writes: >...I question the premise that one can start with that objective propositional >account. This may sound solipsistic, but the only accounts we can give of >events of the world are based on perceptions, be they our own, those of >"credible sources," or interpretations of devices.... > >I think it is the lack of such an objective starting point that supports Cam's >position that the issue here is not "truth" (so, yes, Hutchison IS "barking up >the wrong tree") but sociology--specifically the role the inter-personal >behavior contributes to communication.... I agree that "perceptions" are not completely objective, but if they did not deliver some element of truth, then a lot of things become hard to explain. How do humans and other animals with sophisticated sensors survive if their perceptions do not provide appropriate information about the world, i.e., something true about the world? Are scientific arguments about cold fusion, ozone depletion, dinosaur extinction, cholesterol, and so on merely sociological? I think a more reasonable position is that perceptions provide evidence about many aspects of the world, and that the quality of our assertions about the world depends on the quality of the evidence that supports them. Tom Bylander byland@cis.ohio-state.edu
cpshelley@violet.waterloo.edu (cameron shelley) (06/12/91)
In article <133090@tut.cis.ohio-state.edu> byland@iris.cis.ohio-state.edu (Tom Bylander) writes: [...] >I agree that "perceptions" are not completely objective, but if they >did not deliver some element of truth, then a lot of things become >hard to explain. How do humans and other animals with sophisticated >sensors survive if their perceptions do not provide appropriate >information about the world, i.e., something true about the world? >Are scientific arguments about cold fusion, ozone depletion, dinosaur >extinction, cholesterol, and so on merely sociological? Actually, I think the question (to which I responded before) has not so much to do with efficacy of perceptual capabilities so much as our compulsion to interpret what we perceive. This is further compounded by our subsequent encoding of information for communication to others. I believe it is reasonable to assume that our perceptual capabilities are evolved to help achieve behavioural success, and when dealing with 'everyday' physical data we have little to gain from adopting particular political viewpoints. On the other hand, the original poster seemed more interested in events which are politically charged (ie. perceived in a social context), and filtering "truth" from ideology is unlikely to get far. This is especially true of reports of the motives of other people and the social effects of their actions. >I think a more reasonable position is that perceptions provide >evidence about many aspects of the world, and that the quality of our >assertions about the world depends on the quality of the evidence that >supports them. Again, the subject really wasn't the quality of our perceptions, but that our method of communication contians biases which affect the content of messages. I would point out also that most scientific arguments deal with what generally happens (distributions), not about a narrowly "truthful" account of any individual event. At that point, communication systems like mathematics (with biases we've all been trained to accept) are used, rather than natural language monologs. Cam
ISSSSM@NUSVM.BITNET (Stephen Smoliar) (06/12/91)
In article <133090@tut.cis.ohio-state.edu> byland@iris.cis.ohio-state.edu (Tom Bylander) writes: >In article <9106110020.AA17886@lilac.berkeley.edu> ISSSSM@NUSVM.BITNET >(Stephen Smoliar) writes: >>...I question the premise that one can start with that objective >>propositional >>account. This may sound solipsistic, but the only accounts we can give of >>events of the world are based on perceptions, be they our own, those of >>"credible sources," or interpretations of devices.... >> >>I think it is the lack of such an objective starting point that supports >>Cam's >>position that the issue here is not "truth" (so, yes, Hutchison IS "barking >>up >>the wrong tree") but sociology--specifically the role the inter-personal >>behavior contributes to communication.... > >I agree that "perceptions" are not completely objective, but if they >did not deliver some element of truth, then a lot of things become >hard to explain. Is a proposition having "some element of truth" sort of like a woman being "a little bit pregnant?" :-) I would argue that if you want to talk about propositional accounts at all, then you are basically buying into the rules of formal logic; and those rules require that a proposition be either true or false. (Even if you choose to use a fuzzy logic system, every USE of a proposition ultimately boils down to a commitment to the truth or falsity of that proposition. Making the logic fuzzy simply allows the proposition to flip back and forth with a bit more flexibility.) I am further arguing that getting on in the world does not require buying into those rules. (To repeat Minsky's position from THE SOCIETY OF MIND: Logic is all right for a POST HOC systematic explanation, but that does not mean it is any good for controlling your decision-making behavior.) > How do humans and other animals with sophisticated >sensors survive if their perceptions do not provide appropriate >information about the world, i.e., something true about the world? Your problem, Tom, is that you want to equate the adjectives "appropriate" and "true." I would argue that "true" is only a useful term in formal logic. Outside of that realm, it has been abused to death since (at least) the days of Pontius Pilate. Hopefully, you will agree with me that, as far as formal logic is concerned, truth need not have anything to do with appropriateness; so I suggest we just chuck the term altogether and try to refine what we mean by "appropriate." To give you some idea of the dangers of playing with truth without the support of formal logic, consider mirages. Is your perception of a patch of water on the highway on a hot day telling you "something true about the world?" It is certainly NOT true (even in an intuitive sense of the word) that the highway is wet up there, as you quickly discover when you get closer. Perhaps we have to get even more extreme and say that we cannot talk about whether any information about the world is "appropriate" until have we, as perceiving agents, have subjected it to the sort of interpretation I have been discussing in my previous articles. >Are scientific arguments about cold fusion, ozone depletion, dinosaur >extinction, cholesterol, and so on merely sociological? > An affirmative answer to this question would be sort of a REDUCTIO AD ABSURDUM reading of Kuhn. The funny thing is that buying into it no longer strikes me as absurd. There is that old joke attributed to Wittgenstein about what the sky must have looked like before people accepted the heliocentric model. >I think a more reasonable position is that perceptions provide >evidence about many aspects of the world, and that the quality of our >assertions about the world depends on the quality of the evidence that >supports them. > I guess my own position is that we do not make assertions about the world while we are behaving in it. Therefore, it is not an issue to ask how the quality of those assertions affect our behavior. Rather, there is a much tighter coupling between perception and action. This is what Minsky is trying to get at when he talks about "closing the loop" in THE SOCIETY OF MIND; and it appears that David Chalmers, Robert French, and Doug Hofstadter are trying to come with with a concrete implementation (such is my reading of CRCC Technical Report 49 from the Center for Research on Concepts and Cognition at Indiana University). Logic has quite enough to do in handling that POST HOC explanation of behavior. Don't try to burden it with more than it can handle! =============================================================================== Stephen W. Smoliar Institute of Systems Science National University of Singapore Heng Mui Keng Terrace, Kent Ridge SINGAPORE 0511 BITNET: ISSSSM@NUSVM "He was of Lord Essex's opinion, 'rather to go an hundred miles to speak with one wise man, than five miles to see a fair town.'"--Boswell on Johnson
is_s425@kingston.ac.uk (Hutchison C S) (06/12/91)
I am aware of the PAULINE program; also of a program by (I think) Cornelius Wegman which is similar; also of Carbonell's POLITICS. I still believe that 'truth' is an issue. A semantic theory will specify the meanings of well-formed sentences in the language. A semantics built upon a correspondence theory of truth has a commonsensical appeal to it: what, after all, are sentences expressing propositions about if not about the world in which language users live? Conversely, if sentences do not express propositions about the world that can be true or false (referring instead, for example, to speakers' "perceptions" or internal representations of the world), then how can conversants ever know that they are talking about the same thing(s)? A correspondence theory of truth, and a semantics dependent upon it, rescues the theory of semantics from the vagaries of 'mentalism' and 'solipsism' that Stephen Smoliar fears. Let's go back to my four newspaper headlines: Rioting blacks shot dead by police as ANC leaders meet Police shoot 11 dead in Salisbury riot Rebels kill 11 ANC men Racists murder Zimbabweans One might argue (as has one of my correspondents, John Bradshaw) that all four sentences may be true, albeit each expressing only a partial truth. That is, it may be that the predicates 'police', 'rebel', and 'racist' all apply to the agents of the four sentences. That they *may* apply does not, of course, mean that, in a specific case, they *do* apply. Thus I may be happy about the truth of the proposition expressed by the last sentence; others may not be, though be quite happy about the truth of the first and second. Now let's assume that the greater part of what I or anybody else knows is knowledge derived from text (e.g., I know that Bogota is the capital of Columbia because I have read it; I have never been to Columbia to check; perhaps Bogota doesn't even exist). What do readers of each of the four sentences know? 'Know' is a factive verb; does this then fall short of a bona fide sense of knowledge? Does the reader of headline 1 know the same things as the reader of headline 4? If the first and last sentences both express partial truths, why do I find it hard to persuade my fellows that the last headline represents knowledge of the world every bit as much as does the first headline? How is the knowledge that I and my fellow newspaper readers extract from text integrated into what I and they know already? It seems to me that talk of 'partial truths', 'negotiation', and so on, may not get us very far. If I'm negotiating with you, I'm really just trying to tell you why you are (mostly) wrong and I am (mostly) right. If I adduce evidence to support my claims, then we may end up negotiating what counts as evidence. We're stuck in a hopeless regress. (Try telling one billion Christians or one billion Muslims they're wrong -- especially if it is perfectly obvious to you that Humanistic Buddhism is the only right way. Try negotiating with the Jehovah's Witness on your doorstep. Try telling the free market liberal about the unspeakable suffering and brutality that capitalism has wrought upon the cheap labour markets of the Third World.) I mentioned Carbonell's POLITICS program earlier. It is claimed that the program "simulates human ideological understanding of international political events" by "an American newspaper reader" who may be on an occasion either a "US-liberal" or a "US-Conservative". The program actually does no such thing. Carbonell gives, not events, but *reports* of events; the text received by the reader is the content of an utterance act by some speaker who, as much as much as the reader, has "subjective interests, personal motivations, beliefs" (p.2). This being so, it would be naive and indeed counter-intuitive to assume that such interests, motives and beliefs will not have enetered into, first, the speaker's/writer's reasons for uttering the text at all, and secondly, his lexical and themtic choices in the formulation of his utterance. Here is an example form Carbonell (1981): Soviet-backed forces are scoring rapid gains against the Bhutan government. The US is diverting tanks and M-16s, ear-marked for the US army, in an emergency airlift to Bhutan. Carbonell indicates the different responses that the report (or, in his words, "event") would elicit from a "US-liberal" and a "US-Conservative". Clearly other responses would be elicited were the report worded differently. Here are some of my versions: Rebels' advance against government forces prompts US airlift of emergency arms package to Buthan. US pumps arms into Buthan as the Washington-backed regime loses ground to people's army. Buthanese people's gains in struggle to liberate homeland spark panic bid by US imperialists to bolster beleagured puppet dictatorship. Are the reports talking about the same event? If so, the reports are either true or false, and there should be ways of determining the truth value of the propositions expressed. If the reports represent merely the contents of 'perceptions' or 'interpretations', then how can we ever know that we are talking about the same things? (or: how can we be sure that we are talking about the real world at all, and that therefore there is any physical circumstance that can in principle decide the issue?) To get things in context, despite the political flavour that my question may appear to have taken on, my main concern is with automatic knowledge acquisition from text (whatever kind of text it may be). My problem is: is knowledge representation going to be about an intelligent agent's models of the physical world or of speakers' reports about the world? This is a technical rather than a philosophical issue since it impinges directly on what kinds of inference and what sources of knowledge are relevant to the reasoning process. Chris H.
cpshelley@violet.waterloo.edu (cameron shelley) (06/13/91)
In article <1991Jun12.130817.3621@kingston.ac.uk> is_s425@kingston.ac.uk (Hutchison C S) writes: >I am aware of the PAULINE program; also of a program by (I think) Cornelius >Wegman which is similar; also of Carbonell's POLITICS. I still believe that >'truth' is an issue. > >A semantic theory will specify the meanings of well-formed sentences in the >language. A semantics built upon a correspondence theory of truth has a >commonsensical appeal to it: what, after all, are sentences expressing >propositions about if not about the world in which language users live? Perhaps the key word here is "about". How directly is a speaker's statement related to a 'real' thing? In other words, how 'about' do you want? I doubt anyone would question that speakers are interested in expressing propositions about the world. However, you seem to be asserting that such statements can be 'un-abouted' in order to arrive at an exact record of the sensations the speaker experienced at some time. To construct a model of the relationship between sensation and communication (realistically), you are going to have to accept some compromises, and thus let slip some of the absoluteness of the 'truth' about which you're concerned. Do you want "the truth" or "a truth"? >Conversely, if sentences do not express propositions about the world that >can be true or false (referring instead, for example, to speakers' >"perceptions" or internal representations of the world), then how can >conversants ever know that they are talking about the same thing(s)? We assume we have a common ground of understandings (culture, etc.) or we use some means to establish common ground (the negotiation Steve mentioned). Otherwise, we are very likely to misunderstand and I don't see how any theory can do better. >A >correspondence theory of truth, and a semantics dependent upon it, rescues >the theory of semantics from the vagaries of 'mentalism' and 'solipsism' >that Stephen Smoliar fears. > >Let's go back to my four newspaper headlines: > > Rioting blacks shot dead by police as ANC leaders meet > Police shoot 11 dead in Salisbury riot > Rebels kill 11 ANC men > Racists murder Zimbabweans > If you're after *the* truth, then you'll have to provide *the* definitions of "riot", "racist", and so on (as I guess John Bradshaw pointed out). These are perceptual, possibly unique categories to each speaker and not things you can measure, weigh, or otherwise legitimately encode in a formal language---not and capture all the variations involved. It might be tempting to define a 'correct' riot, and then assert others are incorrect, but then you've done nothing but become another interpreter with your own opinion. This situation doesn't preclude the possibility of understanding other speakers, but it does mean understanding will require work (in the literal sense). Occasionally, it may require the agent to shift the 'fixed-points' (axioms, assumptions) of interpretation, something I haven't observed truth-conditional theories to be good at. >It seems to me that talk of 'partial truths', 'negotiation', and so on, may >not get us very far. If I'm negotiating with you, I'm really just trying to >tell you why you are (mostly) wrong and I am (mostly) right. If I adduce >evidence to support my claims, then we may end up negotiating what counts as >evidence. We're stuck in a hopeless regress. (Try telling one billion >Christians or one billion Muslims they're wrong -- especially if it is >perfectly obvious to you that Humanistic Buddhism is the only right way. Try >negotiating with the Jehovah's Witness on your doorstep. Try telling the >free market liberal about the unspeakable suffering and brutality that >capitalism has wrought upon the cheap labour markets of the Third World.) What you seem to be saying is that the *process* of understanding (or failing to understand) is very hard in difficult cases. I doubt anyone would dispute that. Unfortunately, truisms don't support one position over another. Perhaps we should consider a concrete example. What distinction would a truth-functional theory make between the following pair of utterances? The cat sat on the mat. The mat was sat on by the cat. Although they are different, the usual theory would describe both with the same semantic form and say that either both were true or both were false. But if our communicative apparatus exists just to transmit the truth, why do we have more than one token for the same proposition? The only explanation comes from considering the speaker's desire to emphasize "cat" or "mat" (or conversely de-emphasize the other). Even in a description of an uncontroversial and simple event, point of view can play a role. Now, how true are the following two from previously? Rebels kill 11 ANC men Racists murder Zimbabweans We may judge that members of one group killed members of another, but the points of view have probably dictated the exact descriptions. Were the Zimbabweans really "murdered" or "killed"? Most people would say there is a difference between the two; the two terms are certainly used to have different effects. >To get things in context, despite the political flavour that my question may >appear to have taken on, my main concern is with automatic knowledge acquisition >from text (whatever kind of text it may be). My problem is: is knowledge >representation going to be about an intelligent agent's models of the >physical world or of speakers' reports about the world? This is a technical >rather than a philosophical issue since it impinges directly on what kinds >of inference and what sources of knowledge are relevant to the reasoning >process. Like Carbonell's (and Hovy's) systems, a model of the physical world will require 'objective' input at some point. Since this is not really possible, I would select option b) you give above. Also, as I hope I pointed out, by not modelling speaker's reports, you lose information about speakers, and speakers are presumably in the real world too. And then you have the problem of handling reports about things which don't exist in some way (ie. unicorns, Sherlock Holmes, rained-out ball games, etc...). A paper by Hirst I was forced to read lately (on KR of non-existence) suggests just going with a naive model. I would say, in summary, that using a truth conditional KR and model will get you somewhere, but maybe not where you would think at first blush. Sorry for rambling. :-( Cam
minsky@media-lab.media.mit.edu (Marvin Minsky) (06/13/91)
In article <1991Jun12.130817.3621@kingston.ac.uk> is_s425@kingston.ac.uk (Hutchison C S) writes: >Conversely, if sentences do not express propositions about the world that >can be true or false (referring instead, for example, to speakers' >"perceptions" or internal representations of the world), then how can >conversants ever know that they are talking about the same thing(s)? The reductio ad absurdum is appropriate. Conversants never do, in fact, know that they are talking about the same things. It is always a matter of convention, convergence, and good fortune -- even in the case of "mathematical truths". When you and I both talk about "that chair over there", our internal models differ substantially, but not enough to make most practical interactions too difficult. And the cchir itself changes imperceptibly from one moment to the next as it loses and gains atoms and suffers thermal agitations of its internal degrees of freedom. There is no chair, indeed, from a modern physical point of view, only boundaries imposed by observers; my decorator friend regards this chair and that other one as a possibly conflicting pair, my fried the carpenter sees it as a possibly unsound linkage of glue and sticks, and so on. Let's grow out of this unproductive idea of formal semantics, and low-level childish, religious, primitive ideas about truth, and get on with the work of making machines that can solve problems and communicate with one another as best they can.
ISSSSM@NUSVM.BITNET (Stephen Smoliar) (06/13/91)
In article <1991Jun12.130817.3621@kingston.ac.uk> is_s425@kingston.ac.uk (Hutchison C S) writes: >A semantic theory will specify the meanings of well-formed sentences in the >language. A semantics built upon a correspondence theory of truth has a >commonsensical appeal to it: what, after all, are sentences expressing >propositions about if not about the world in which language users live? >Conversely, if sentences do not express propositions about the world that >can be true or false (referring instead, for example, to speakers' >"perceptions" or internal representations of the world), then how can >conversants ever know that they are talking about the same thing(s)? A >correspondence theory of truth, and a semantics dependent upon it, rescues >the theory of semantics from the vagaries of 'mentalism' and 'solipsism' >that Stephen Smoliar fears. > I am willing to accept that mentalism bring along some vagaries which we would like to avoid, but the only reason I was sounding apologetic about solipsism is that I suspect it is not as bad as we have been conditioned to believe. Long before he began to develop his work on situated automata, Stan Rosenschein was entertaining the possibility that solipsism had a legitimate role in artificial intelligence; but back in those days it was still fashionable to poke fun at Bishop Berkeley. Now that we are beginning to get a handle on situated reasoning and build systems which can actually engage it, I see no reason to "fear" solipsism. Rather, it may rescue us from all the corners in which we keep painting ourselves with our obsessive belief that "knowledge representation" has something to do with that knowledge we engage to get along in the world. Let us try to pursue this point a bit further: > Now let's >assume that the greater part of what I or anybody else knows is knowledge >derived from text (e.g., I know that Bogota is the capital of Columbia because >I have read it; I have never been to Columbia to check; perhaps Bogota doesn't >even exist). I shall grant you this assumption even though I disagree with it. When you get too wrapped up in text, you tend to dismiss all the things you know that are NOT derived from that source (such as how to tie your shoes, how to cross a busy street, and probably even how to get to work in the morning). I would further argue that it is all this non-text knowledge which we never even consider articulating in text which is REALLY the "greater part" of what anybody "knows." > What do readers of each of the four sentences [the four headlines about an > African event] know? 'Know' is >a factive verb; does this then fall short of a bona fide sense of knowledge? >Does the reader of headline 1 know the same things as the reader of headline >4? If the first and last sentences both express partial truths, why do I >find it hard to persuade my fellows that the last headline represents >knowledge of the world every bit as much as does the first headline? How is >the knowledge that I and my fellow newspaper readers extract from text >integrated into what I and they know already? > Basically, I would argue that you are trying to make your point by asking a lot of ill-formed questions! It is not the QUANTITY of your questions that matters but rather their QUALITY! Rather than ask what a reader "knows," I would argue that you should be asking how that sentence impacts his behavior. At this point, you have to recognize that there is no such thing as a "generic" reader. You can only ask about the behavior of a flesh-and-blood (so to speak) INDIVIDUAL, rather than an abstract sentence processor. For example, for an international trader, "knowledge" is going to have to do with doing business in Africa. If he has an office in Salisbury, he probably has to entertain a decision to shut that office down and evacuate his personnel. On the other hand, a white middle-class reader in a relatively quiet town might start reflecting on his attitude towards the blacks who moved in down the lane, realizing that he has been unconsciously crossing to the other side of the street whenever he sees them. We are not talking about text-based propositions which are true or false here. We are talking about making decisions in the real world--a form of knowledge which, I believe, Donald Schoen has come to call "knowledge-in-action." (Since my books are still in transit, I may need to be corrected on this.) My sincere advice to you, Chris, is to get yourself a better set of questions before you proceed any further! >Here is an example form Carbonell (1981): > > Soviet-backed forces are scoring rapid gains against the Bhutan > government. The US is diverting tanks and M-16s, ear-marked for > the US army, in an emergency airlift to Bhutan. > >Carbonell indicates the different responses that the report (or, in his words, >"event") would elicit from a "US-liberal" and a "US-Conservative". Clearly >other responses would be elicited were the report worded differently. Here >are some of my versions: > > Rebels' advance against government forces prompts US airlift of > emergency arms package to Buthan. > > US pumps arms into Buthan as the Washington-backed regime loses > ground to people's army. > > Buthanese people's gains in struggle to liberate homeland spark > panic bid by US imperialists to bolster beleagured puppet > dictatorship. > >Are the reports talking about the same event? If so, the reports are either >true or false, and there should be ways of determining the truth value of >the propositions expressed. If the reports represent merely the contents >of 'perceptions' or 'interpretations', then how can we ever know that we >are talking about the same things? (or: how can we be sure that we are >talking about the real world at all, and that therefore there is any physical >circumstance that can in principle decide the issue?) > In this case I feel I can give you a straight answer: We can't! This is not as horrible as it may seem at first blush. There are very few absolute conclusions we draw as we go around in the world. Our "intelligence" (whatever that means) does not reside in our ability to determine the true of propositions but in our ability to adapt to changing positions in the conclusions we draw and the decisions we make. Much of dialog is actually a matter of dealing with the fact that two people are not really "talking about the same things." Since they still have to deal with each other, dialog becomes a tool for resolving matters; but there is never any ABSOLUTE resolution. Rather, there are these continuous streams of behavior. Were those streams not properly mediated, we would not be able to survive in this confusing world. =============================================================================== Stephen W. Smoliar Institute of Systems Science National University of Singapore Heng Mui Keng Terrace, Kent Ridge SINGAPORE 0511 BITNET: ISSSSM@NUSVM "He was of Lord Essex's opinion, 'rather to go an hundred miles to speak with one wise man, than five miles to see a fair town.'"--Boswell on Johnson