IT21@SYSB.SALFORD.AC.UK (09/03/88)
Date: Thu, 1 Sep 88 12:36 EDT From: IT21%SYSB.SALFORD.AC.UK@MITVMA.MIT.EDU To: ailist@AI.AI.MIT.EDU Subject: Newell's Knowledge Level From: Andrew Basden, I.T. Institute, University of Salford, Salford. Please can anyone help clarify a topic? In 1982 Allen Newell published a paper, 'The Knowledge Level' (Artificial Intelligence, v.18, p.87-127), in which he proposed that there is a level of description above and separate from the Symbol Level. He called this the Knowledge Level. I have found it a very important and useful concept in both Knowledge Representation and Knowledge Acquisition, largely because it separates knowledge from how it is expressed. But to my view Newell's paper contains a number of ambiguities and apparent minor inconsistencies as well as an unnecessary adherence to logic and goal-directed activity which I would like to sort out. As Newell says, "to claim that the knowledge level exists is to make a scientific claim, which can range from dead wrong to slightly askew, in the manner of all scientific claims." I want to find a refinement of it that is a bit less askew. Surprisingly, in the 6 years since the idea was introduced there has been very little discussion about it in AI circles. In psychology circles likewise there has been little detailed discussion, and here the concepts are only similar, not identical, and bear different names. SCI and SSCI together give only 26 citations of the paper, of which only four in any way discuss the concepts, most merely using various concepts in Newell's paper to support their own statements. Even in these four there is little clarification or development of the idea of the Knowledge Level. So I am turning to the AILIST bulletin board. Has anyone out there any understanding of the Knowledge Level that can help in this process? Indeed, is Allen Newell himself listening to the board? Some of the questions I have are as follows: 1. Some (eg. Dennett) mention 3 levels, while Newell mentions 5. Who is 'right' - or rather, what is the relation between them? 2. Newell says that logic is at the Knowledge Level. Why? I would have put it, like mathematics, very firmly in the Symbol Level. 3. Why the emphasis on logic? Is it necessary to the concept, or just one form of it? What about extra-logical knowledge, and how does his 'logic' include non-monotonic logics? 4. The definition of the details of the Knowledge Level is in terms of the goals of a system. Is this necessary to the concept, or is it just one possible form of it? There is much knowledge that is not goal directed. Alexander et. al. and Clancey both question Newell's adherence to logic and goals, but do not discuss the case. Can anyone shed any light? I have further questions, which I will put directly to some of those who reply. Or (please tell me) should I put them on the board? And would anyone like a summary from me of my findings? Thank you, in advance. Andrew Basden Information Technology Institute, University of Salford, Salford, UK. JANET: abasden@uk.ac.salf.b Phone: (44) 61 736 5843 x510; Telex: 668680 (Sulib); Fax: (44) 61 745 7808
jbn@GLACIER.STANFORD.EDU (John B. Nagle) (09/05/88)
To: comp-ai-digest@decwrl.dec.com
Path: labrea!glacier!jbn
From: John B. Nagle <jbn@glacier.stanford.edu>
Newsgroups: comp.ai.digest
Subject: Re: Newell's Knowledge Level
Date: Sat, 3 Sep 88 11:06 EDT
References: <19880903034508.9.NICK@HOWARD-JOHNSONS.LCS.MIT.EDU>
Reply-To: John B. Nagle <glacier!jbn@labrea.stanford.edu>
Organization: Stanford University
Lines: 22
Much the same idea has been referred to as "deep understanding" by
the rule-based knowledge representation people. The term "deep structure"
is sometimes used by those working on natural language understanding. In
both cases, the limitations of the superficial representations in use today are
being recognized. The remark "Mycin doesn't know about bacteria" dates
from the previous decade, but is still applicable. Many critics of AI,
from Weitzenbaum to Dreyfus, have noted this problem, which some refer to
as the "knowledge representation problem". This is a key unsolved problem
in AI. A recent posting here by McCarthy indicates that he considers it
the key unsolved problem, and that effort should be directed toward the
development of a formal language suitable for the representation of
"deep understanding" of the real world.
I have not heard of any system where "deep understanding" or "deep
structure" or a "knowledge level" were implemented in any general way.
In a very few systems, always ones where the underlying domain is formalizable,
there is some notion of deep understanding. Eurisko (Lenat) comes close.
When people use these terms, they are usually talking about the parts of the
problem for which no useful approaches are known.
John Naglemohan@BOC.RUTGERS.EDU (Sunil Mohan) (09/05/88)
To: comp-ai-digest@paul.rutgers.edu Path: boc.rutgers.edu!mohan From: Sunil Mohan <mohan@boc.rutgers.edu> Newsgroups: comp.ai.digest Subject: Re: Newell's Knowledge Level Date: Sun, 4 Sep 88 11:53 EDT References: <19880903034508.9.NICK@HOWARD-JOHNSONS.LCS.MIT.EDU> Distribution: usa Organization: Rutgers Univ., New Brunswick, N.J. Lines: 34 The Knowledge Based Software Development Environment (KBSDE) group at Rutgers University are strong believers in the separation of the specification of knowledge from the specification of its use. I believe that that is the underlying theme of Newell's "Knowledge Level". Marr has also talked about the specification of a system in different levels, separating knowledge from algorithm from implementation. This allows a partitioning of the concerns involved in developing a system. As a simple example, it allows one to decide whether inability to solve a particular problem is due to lack of knowledge or an inherently `incomplete' algorithm that uses that knowledge. Describing your research along these levels will also help you and the reader decide where the contribution lies. See for example the paper "Learning At The Knowledge Level" by Dietterich (I think). How many levels you choose to have depends entirely on how finely you wish to partition your concerns. There is no "right" partitioning. The eventual aim is clarity. As far as logic belonging at the Knowledge Level is concerned, in so far as logic is used as a declarative specification of knowledge, and its implications, that is the purpose of the knowledge level. I would tend to think that logic may also be used to specify the algorithm at the symbol level, thus allowing the capability of reasoning about the algorithm. I don't know what you mean by "extra-logical". Could you perhaps be taking the terms too literally? Remeber that the Knowledge Level in itself is not interesting. It is interesting because of what it achieves (viz. clarity, focussing attention). Logic is just a specification and reasoning device. Any form of logic should do, so long as you are aware of its capabilities and limitations. _ Sunil