ArticlePDF Available
The question of effectively managing historical knowledge in an appropriate
fashion remains, despite modern computer techniques, an open question.
Historians have amassed enormous amounts of historical interpretation that are
amenable to textual search, but not to anything more semantic. The limitations of
textual search become apparent when one considers the differences between
feudal Japan and feudal Western Europe, between African-American slaves of the
Antebellum Period and Muslim slaves of the 10th-13th century as reported in the
documents of the Cairoer Geniza. Tools that treat documents as bags of words are
no help for researching these problems.
One tool that could assist historians1 is symbolic knowledge bases. Symbolic
knowledge bases take sentences authored in a knowledge representation language
akin to mathematical logic (Brachman and Levesque, 2004) as their input. The
meaning of the knowledge comes from the way the individual concepts, are
related to each other. The automatic reasoning operations permissible with these
concepts bottom out in the rules of inference, as defined for the mathematical
logic that provides the semantics for the knowledge representation language.2
However, using formal knowledge representation languages requires that
historians give a formal account of their interpretations; for it is this formal
account that can then be represented symbolically, and it is for this formal account
that the rules of inference are defined. This requirement is notorious: For any
discipline, there are enormous difficulties in capturing expertise in formal terms
(Russell and Norvig, 2002), (Klein, 1999). Historians compound these issues by a
research stance that is averse to “theories”.3
Towards a formalizable Account of Historical Interpretation
The “theory”-aversity of the historical community is understandable. In general,
interpretation attempts to assign meaning to a set of data treated as given. Inter-
pretation assigns meaning by selecting from a store of pre-existing conceptual
models and modifying them to bring them into alignment with the data requiring
1 The argument applies, mutatis mutandis, to other Humanities or Social Science
researchers, but for the purposes of this exposition, focus will rest on the historical
2 The theory underlying symbolic knowledge bases is the theory used in semantic web
technologies, such as DAML, OWL/DL and the like.
3 Consider pars pro toto (Horden & Purcell, 2000, 44): “To specify at this early stage
the forms that our efforts to outwit [historiographical] tradition will take would be to offer
something in the nature of a detailed programme or methodology. And that we prefer to
avoid: there have already been too many in the history of Mediterranean studies.”
interpretation.4 These appropriately adjusted conceptual models are then in turn
useful for identifying more data or adjusting the meaning of the “given” data. Put
differently, the “Hermeneutic Circle of Interpretation” (Gadamer, 1990) as the
fundamental paradigm of interpretation states that it is the continued interaction
between the existent conceptual models of the researcher and the sources (in the
broad sense of the term) which generates understanding—as well as better data
and conceptual models for the next round of interpretation.
For example, a historian might at first interpret documents from feudal Europe
with conceptual models developed in the research of feudal Japan. Sources from
Western Europe's feudal period will describe problems that are alien to the
Japanese context, such as the presence of an increasingly stronger Roman
Catholic Church. As the historian adapts the conceptual models to make sense of
the Western European sources, some of the adaptations will impact the inter-
pretation of Japanese feudal society. Thus, the historian might return to search
Japanese feudal documents for structural equivalents to the roles of the Roman
Catholic Church. This research agenda cannot be stated in the context of the
Japanese sources alone.
During the process of hermeneutic research, then, the data and the conceptual
models co-evolve, as the researchers uncover over-generalizations or omissions in
their prior conceptualizations, or recognize new data in the sources, flushed out by
their adapted conceptual models. “Theories” that are resistant to this dynamic
feedback cycle between data and conceptual models are suspicious from the
beginning. Formalization attempts such as Hempel's covering law unintentionally
sent the wrong signals to a community already rebelling against notions of
“eternal laws” from orthodox Marxist historians.
The question then is whether, and if yes, how historical knowledge could be
represented appropriately in symbolic knowledge bases.
Guidance on Formalizing Historical Knowledge
Those interested in formalizing historical knowledge receive a helping hand from
Arthur C. Danto's seminal The Analytical Philosophy of History (1968). Danto has
the credentials to bridge the gap between the two communities: He is an
accomplished art historian and an analytical philosopher of the first rank.
Fortuitously, Danto’s Analytical Philosophy of History is foremost a demolition of
any “eternal laws” approaches to historiography. In thought experiments such as
the “ideal chronicler” (149-181) and in the identification of the historical tool of
the narrative sentence (143-181), Danto argues that historical interpretation is
fundamentally open toward the future (195-198) – unlike eternal laws. At the
same time, Danto provides formal logic interpretations for the kinds of natural
language descriptions that historians utilize in their work. Specifically, Danto
4 “Both ancient and modern perceptions should, in the first instance, be seen as
belonging equally to the history of ideas; before we test their applicability, that is, we
should interrogate their sources.” (Horden and Purcelle, 2000, 13).
establishes three critical insights that form the foundation for any serious effort to
formalize historical knowledge: interpretative expectations, historical laws and
preliminary types. These contributions provide a blueprint for formalization,
derived from the kinds of problems historians deal with and the natural language
description forms they employ.
The remainder of this essay is devoted to a discussion of these three critical
insights. This will include a preliminary survey of the representation approaches
that the knowledge representation community already has developed that could fit
the formal structures Danto identifies. While much is available, a critical lacuna
remains. In the conclusion, we will therefore describe a missing representation
structure we term a recurrent5 that captures the hermeneutic circle in its essence
and is the appropriate amalgamation of existing knowledge representation
infrastructure and Danto’s insights.
Interpretative Expectations
Danto describes the role of interpretative expectations in the context of the
thought-experiment of the “ideal chronicler” (149ff). The “ideal chronicler” is an
abstraction that simulates the predicament that the historian faces vis-a-vis the
historical sources. Put differently, the descriptive "granularity" of a particular
source is fundamentally atomic and cannot be subdivided by the historian using
that source alone. Consider Samuel Johnson's remarks on the Irish education of
Jonathan Swift—“… [Jonathan Swift] was sent at the age of six to the school at
Kilkenny, and in his fifteenth year [1682] was admitted into the University of
Dublin.” (Johnson, 1781, vol.I); they treat Swift's school time at Kilkenny as a
unit that permits no introspection. It is the literary convention of the biography
that allows Johnson to employ this atomic unit as a building block for telling of
the education of Swift.
This operation—inserting informational units into larger interpretative
structures shared by the historian and the audience—is one of the fundamental
tools of historiography. But in the rigorous analysis of what he terms “project
verbs” (161ff), Danto shows how complex this operation is from the perspective
of formalization. The example that Danto uses to expose these problems bears
quoting in full:
5 The German equivalent, Rekurrenz, is a technical term from textual linguistics and
denotes the accurate or allusive repetition of a language surface structure, as in: “My
opponent increased the taxes. My opponent increased the influence of the Federal
Government.” The notion of applying recurrents as a technical notion on the problem of
interpretation, especially of social and historical data, was inspired by Hermann Lübbe's
analysis of historiography (Lübbe 1977), which in turn looks at the contributions of Arthur
C. Danto's Analytical Philosophy of History to philosophy of history. However, the way the
term “recurrents” is employed here intentionally does not follow Lübbe.
“... [S]uppose Jones in temporal succession puts a seed into the ground,
scratches his head, strikes a match, blows a smoke ring, thinks of his wife and
shifts his foot. Asked at any moment during this stretch of time, *what* he is
doing, Jones will answer correctly, 'Planting roses.'” (162) 6
The first observation is that the project of “planting roses” that Jones is pursuing
is not observable. What is observable is merely that Jones' self-interpretation (as
reported) of planting roses is not immediately rejected by the description of the
acts Jones is performing.7 Notice how weak a statement “not immediately
rejected” is. And the statement needs to be weak by necessity; as Danto points out
(163), it is in general extremely difficult to disprove that Jones was “planting
roses”. At the same time, the description is exceedingly believable: Human beings
often do a multitude of things when they actualize a project—Swift was not just at
Kilkenny's school during those years from 1673 to 1682. Such “multi-tasking” is
especially true for long-term projects, as discussed below.
The second observation is that the conceptual model of “planting roses” used
in interpretation functions like a filter (165): It classifies the actions that Jones
takes into actions conducive to this goal and actions that occur because Jones is
also other things than a rose planter, e.g. bipedal, married or a cigarette smoker.
Cultural knowledge shared between Jones, Danto and the reader that supports this
classification. Such knowledge colors the perception. Jones might belong to a
horticultural tradition that uses cigarette ashes as fertilizer. He may believe in foot
shuffling and/or head-scratching as bringing about good fortune. If Jones adhered
to a fertility religion, he might expect the thoughts expended on his wife to
amplify the generative powers of the rose seed. Typically, historians discover
mismatches between their cultural expectations and the ones presumed by the
sources through atypically frequent occurrences of actions their conceptual model
would have filtered: If several sources on gardening mentioned foot shuffling,
historians would suspect that this was part of the conceptual model of flower
planting. Alternatively, parallelizing constructions might indicate that these
actions were meaningful. If a source contained a simile such as, “Next to the bar,
Miller was smoking his cigarette as if he was planting daffodils,” historians would
emend their conceptual models accordingly.
The third observation highlights the ubiquity of these conceptual models, to the
point where their use almost goes unnoticed. Careful reading of the rose-planting
example reveals that Danto does not give all the information for a conceptual
model of “smoking.” Nothing is ever put into the mouth; the match is never
brought into contact with anything smokable. Danto never even tells his readers
6 Notice that Danto in this example also plays the role of a source, forcing us to accept
his atomic units as the basis of the discussion.
7 For this analysis, we ignore the problem of how the source knows that Jones was
thinking of his wife.
what it is that Jones might be smoking.8 Thus, even the assumption that the
striking of the match and the blowing of the smoke ring belong together to match
a conceptual model of lighting a smokable substance is just that: an assumption
licensed by a particular conceptual model, whose appropriateness is open to
challenge. Since the range of conceptual models familiar to Danto's readership
will not allow associating the striking of the match with any other activity—e.g.,
as a contribution to the fertility thoughts, perhaps—the “chunking” of the atomic
actions is such that match striking and smoke-ring blowing are paired off as
partial descriptions of a specific conceptual model.
Some might protest that the proximity in the description of the match-striking
and the smoke-ring-blowing lends strong credence to the conceptual model of
Jones lighting a smokable substance; linking the match-striking with thoughts of
Jones’ wife, as in the fertility model, would have to explain why the author chose
to insert the smoke ring blowing between these two actions.9
But notice that proximity is primarily useful as a preference mechanism once
conceptual models have been identified; the proximity of the head scratching and
the match striking, for example, is deemed irrelevant because fitting conceptual
models are absent. Furthermore, the demand for an explanation might be simple
to justify: The very mention of an ill-fitting action might in fact be a pointed
criticism of Jones, illustrating the laxity of his religious practices or the uncouth
nature of his upbringing. This indicates that the earlier description of the filtering
mechanism was overly sketchy: Some actions might be neutral with respect to the
project, but others positively counter-productive from the point of view of the
conceptual model. We would expect such counter-indicators to be part of the
conceptual model.
So far the discussion has assumed that Danto as our source culturally shares
the conceptual models that Jones employs and applied them all as Jones would
have.10 This is the problem of the conceptual bias of the source and the fourth
observation to make. Danto, as the observer of Jones, is in the same position as
Danto's reader when reading Danto's narrative. Danto also decoded Jones'
behaviors in terms of conceptual models, albeit at a lower, more sensory level that
8 It is not even certain that the match lighting succeeded; only the match striking is
recorded. However, the following discussion will simplify matters by assuming—without
warrant—that the failure to light would have been reported in the source.
9 This claim partially depends on the fact that historians prefer to interpret their
sources as maximally intentional. While the possibility of a clumsy author always exists,
the interpretative stance is very difficult to control technically: It becomes difficult to sepa-
rate the ignorance of the author from the ignorance of the interpreter.
10 We bracket here the problem that Jones, as the "author" of a sequence of actions,
is no privileged interpreter of that sequence (Eco, 1998) as well as all of the psycho-
historical issues this assumption raises. While these are valid issues, it is the assumption
of the position presented here that these issues only raise more problems of the same
types, but not qualitatively different ones. A similar argument holds for all issues of
philology, which have been bracketed here.
is even more challenging to unravel. This is how, even though historians prefer to
treat their sources as authoritative as possible, additional information sources
might allow them to question the correctness of the application of conceptual
models as recorded in the source. Under any conceptual model that interprets the
striking of the match as a necessary component of the proceedings, a historian
might suggest that Danto incorrectly decoded the extinction of the match with the
foot as a foot-shift. The historian does this by generalizing the observed action to
a more generic one, such as "foot motion", and then looking for more specific
action types that are more strongly connected to the local context.
The fifth and final observation to make is tied to the problem of long-term
projects. A conceptual model for a “project verb” comes with an expectation of
the amount of time it will take for the project to reach completion. Danto
illustrates this point with reference to writing a book or courting another person
(165), both activities that can take significant amounts of time. Long term projects
have a much more permissive identification function with respect to actions that
do not contribute nor disrupt or negate the long-term project. Most historians'
conceptual model of book writing would permit for someone to “take time off”
without considering the project abandoned.11
Representing Conceptual Models of Interpretation
The knowledge representation community recognized early on the need for con-
ceptual models to assist in the interpretation of natural language and of the every
day world. Knowledge representation pioneer Marvin Minksy (Minsky, 1974)
borrowed British psychologist Sir Frederick Bartlett's notion of schemas (Bartlett,
1932) to define frames of knowledge, highly correlated pieces of information in a
sense that generalized to physical structures as well as to situations. Cognitive
scientists and psychologists (Schank and Abelson 1977; Rumelhart 1980), coming
to the problem from the side of natural language understanding, proposed the use
of a hierarchy of information-organizing constructs—scripts, plans, goals and
themes—to capture the conceptual models that formed the backdrop of reading
comprehension in story understanding.12 Since in natural language discourse,
scenes from conceptual models are referenced pars pro toto for the conceptual
model directly (e.g. “striking a match” instead of “lighting a cigarette”), the
contribution that striking a match can make to lighting a cigarette needs to be
represented to use such knowledge accurately.
Knowledge that is heavily dependent on events, such as historiographical
information, can in addition benefit from a proposal by Donald Davidson
(Davidson 1967) for formalizing the semantics of action verbs. Davidson made
the event as such an integral part of the represented knowledge and treated
additional information about the event, such as who performed an activity or how
11 Danto proposes “temporal structures” (161) for situations where the covering event
is temporally discontinuous while the individual sub-events are tempoally contiguous.
12 This description follows (Brewer, 1999) and (Brewer et al, 2000).
the activity was performed, with the use of which implements, etc. as property
statements about the event. The so-called Davidsonian event representation aligns
well with the intentions behind Schank and Abelson's script representation, as a
set of universally quantified statements about the event types; the roles and the
actors that can play in them; the pre-conditions for instances of this event type;
the relationships between various role assignments across sub-scenes; and the
post-conditions and world state changes affected by the script's execution.
As far as the knowledge representation community is concerned, the proposals
of Schank, Minksy and Rumelhart failed to make sufficient dents into the hard
problems of artificial intelligence, such as natural language understanding. The
needed level of detail for representing the scripts, for example, to process
arbitrary newspaper articles proved to be forever elusive.13
However, for the purposes of the representation of historical knowledge, where
the humans are authoring the knowledge, this is less of a concern. Suffice it to
say then that the knowledge representation community has developed types of
representational vocabulary (Forbus et. al., 2005; Kahlert et. al., 2006) appropriate
to the needs of the historical community for dealing with the interpretative
expectations, as outlined by Danto.
The Clarification of Historical Laws
Danto's contribution to the nomological debate of the late 1960s is effectively a
clarification of the debate that makes the problem go away. This clarification
interprets the different positions of nomologists like Hempel on the one side and
the historians like Dray on the other side as parts of an overall coherent argument,
with each side emphasizing their contribution. Danto localizes the source of
confusion in the problem of interpretative explanations.
Danto agrees with Hempel that explanation means alignment of data with a
model, in this particular case, with universally quantified variables in a deductive
rule. The determination of the relevant data and its alignment with the model is
the historian's interpretative contribution and therefore compatible with Hempel's
analysis. However, what makes the discussion confusing is the linguistic encoding
of that alignment. When historians write about these explanations, they do not
give deductive proofs; they use conceptual shorthand that gives enough of the
relevant data and the model alignment for the readers to work out proof for
themselves (222f). Thus, one source of the confusion is that the writing is merely
detailed enough to satisfy the interpretative expectations of the readership. Danto
shows (223) that there is always a reformulation of a historical explanation that
13 For Wendy Lehnert’s assessment of her graduate student days in Roger Schank’s
AI lab at Yale, see Lehnert (1994, 150-163).
has the property that it makes the alignment of the data with the deductive rule,
and thereby the deductive inference and its proof steps, explicit.14
It is easy to underestimate the contribution that Danto made to the discussion
of historical laws, mainly because the context is no longer considered relevant.
The theoretical programs that attempted to formalize the philosophical under-
pinnings of the empirical sciences are no longer appreciated (Laudon, 1996) and
as programs, have been largely abandoned. The heavy focus on deduction,
inspired by the supporters' connection to the Wiener Kreis, as the normative mode
of generation of evidence has given way to a plurality of quantitative and
qualitative approaches and a philosophy of science discourse based around task-
appropriate model building (Giere, 1988) and conceptualizations (Van Frassen,
1980), with which the data is now aligned, without much regard for the
epistomological status of such models and conceptualizations.
But for the dialog between the historiographical community and the knowledge
representation community, Danto's observations are a relief. Though not in these
words, Danto effectively identified the intermediate points of stability in the
hermeneutical process, the structure of the conceptual models that confront the
data and vice versa, as bundles of deductive rules with universal quantification.
Such rules form the logical backbone of the knowledge representation endeavor
and are therefore capabilities that the knowledge representation community has
much expertise with and that its tools support well. Other than the historians'
protesting against a nomological approach to historiography might have
suggested, historical explanation is rife with the deductive rules the knowledge
representation community knows how to handle so well.
The preliminary Structure of Historical Explanation
The third and final contribution that Danto makes to the discussion between
historians and the knowledge representation community is an analysis of the
preliminary properties that command the structure of historical explanation. And
to complicate matters further, these preliminary properties have projections onto
the temporal structure of historical arguments as well.
In the context of a reconstruction of how narratives are structured by
preconceptions (120-129), Danto proposes the thought experiment of imagining
that as little was known about Leonardo da Vinci as is about the Greek painters of
Antiquity (123). The critical observation that Danto makes is that this thought
experiment is not really executable, because one's notion of what an artist is
contains so much of what Leonardo da Vinci was (123). Or put differently,
exemplary instances of specific types have the property that they modify their
type in such a manner as to color it with their peculiarities. But this has important
ramifications for the temporal structure of historical explanation of an artist.
14 Danto warns that the recovery of the deductive rule from the narrative qua proof
fragment need not be straightforward (223) and possibly not amenable to an algorithmic
discovery process (250).
There are now conceptual aspects of the type that are distinctly Renaissance,
despite the fact that there are commonalities that make it sensible to categorize
both the painters of Antiquity and Leonardo da Vinci as artists. These conceptual
aspects cannot have held prior to the Renaissance and eventually ceased to hold.
The type then, initially treated as the given, becomes a preliminary statement that
allows to identify exemplars, but in turn sees its conceptual semantics modified by
the examples. The type becomes a cursor that sweeps over a temporally ordered
example sequence, conforming to the exemplars or statistical averages of the
times and places in turn. 15
As mentioned before, in a narrative sentence (143-181), a past event is
analysed from the vantage point of a more recent (but still past) time point that
explains the significance of the prior event.16 Danto explicates this concept with
the example of referring to the birth house of Isaac Newton as the birth house of
the author of the Principia17 (158). But the same select principle holds in the
modifications to the conceptual semantics of a type. When deciding which
conceptual elements Leonardo da Vinci contributes to the type “artist”, large
subranges of the temporally ordered example sequence, starting with Leonardo
and potentially extending as far up as the historian’s own time, are taken into
consideration. What is shared, over periods and regions, becomes part of the type;
what is not becomes curiosity. Artists still fill sketch books; they do not write in
mirror-image cursive.
For the knowledge representation community, this is a usual situation, because
the type information usually forms the backbone of the inference process. Type
reasoning is computationally fast and can reduce the inferential search space
through early pruning. Specifically, the oscillation of the type’s extension over
time is also unusual, to say the least. Danto’s observations trigger no recognition
of ready-made solutions, only the parameters for a detailed description of what is
Towards Recurrents as Formal Building Blocks for Historiography
While the knowledge representation community is able to describe the inter-
pretative expectations historians require in a universally quantified form—as type
level script representations in the Schankian tradition, for example—, this
15 This observation differs from an ideal-type (in the spirit of Max Weber’s Objektivi-
tätsaufsatz) in that Danto is explicitly interested in the temporal properties.
16 Danto initially develops this argument to show that the interpretation of the past
must be open toward the future, because future events can change how the past will be
perceived and interpreted in the future. Danto illustrates this fact through an example with
reference to the verb "anticipate" (169).
17 The reference is to Newton’s main work, the Mathematical Principles of Natural
Philosophy, in Latin Philosophiae Naturalis Principia Mathematica, of 1687, which
contains (among other insights) the Newtonian laws of motion,
capability only covers the intermediate steps in hermeneutical interpretation,
which pretend to interpretative quiescence.
As the description of the derived types shows, however, the types that the
knowledge representation community prefers to consider primary and fixed are
more fluid and provisional in the changing process of interpretation, where data
and models can influence each other.
This is where the recurrents come in. A recurrent is conceptual model, encoded
using universally quantified interpretation expectations, that retains knowledge
about its own evolution, specifically, about the data set that drove its hermeneutic
transformations. It has information about how it was abstracted away from the
data set and thereby can justify its appropriateness. The features that recurred in
the data make up the conceptual semantics of the recurrent.
As Danto’s analysis of the temporal structures in historiography show, these
data sets exhibit change over time. Therefore, the types must be revised as the
data sets change. But in historiography, data sets are fundamentally incomplete.
Few experiments can generate new data; yet new documents, new research or new
excavations could bring to light new information at any point in time. And for
historiography, new information means new data.
Such new data might require re-doing the abstraction process and force a
reconceptualization of what it means, for example, to be an artist. This in turn
could influence the data set that the recurrent contains: Some data might now be
eliminated as no longer applicable; alternatively, new data, excluded earlier for
lack of relevancy, might be added. And the conceptual re-adjustment of the
recurrent would not be limited to the internal composition of and the data
supporting the recurrent. The recurrent of a Renaissance artist would be linked to
such recurrents as funding, military and political views, artistic styles and
fashions, materials and techniques, and the like—each appropriately temporalized
and abstracted from data sets.
Changes to any of these recurrents would propagate outward to the other
recurrents, causing them to undergo reconsideration themselves, consequently be-
coming the source of reconsiderations in their own right. Adjustments would con-
tinue to trickle through the semantic network of recurrents until some form of
quiescence were achieved, i.e. there being no remaining way to make progress on
either the conceptual models or the data items.
The knowledge representation community has much to offer to the
historiographic community in terms of its formalisms for capturing interpretation
expectations and its facility with formal rules. It is in the realm of representing the
temporally mutable and preliminary, however, that the necessary infrastructure is
missing. Elements, such as forward-concluding rules or automatic truth
maintenance systems to support the type modifications, exist. But a holistic model
of the recurrent as the formal building block of a knowledge representation
approach to historiography remains a desideratum.
Brachman, Ronald J., Hector J. Levesque (2004). Knowledge Representation and
Reasoning. San Francisco : Elsevier.
Danto, Arthur C. (1968). Analytical Philosophy of History. New York: Columbia
University Press.
Davidson, Donald (1967). The Logical Form of Action Sentences.
Reprinted in:18 Donald Davidson (1980). Essays on Actions and Events. Oxford :
Clarendon Press, p.102-122.
Eco, Umberto (1998). Lector in Fabula: Die Mitarbeit der Interpretation in erzählenden
Texten. Stuttgart : Deutscher Taschenbuch Verlag (DTV).
Forbus, Ken, Larry Birnbaum, Earl Wagner, James Baker, Michael Witbrock (2005).
Combining analogy, intelligent information retrieval, and knowledge integration for
analysis: A preliminary report. In: Proceedings of the 2005 International
Conference on Intelligence Analysis, McLean, VA, May 2005.
Gadamer, Hans-Georg (1990). Wahrheit und Methode. Tübingen : Siebeck-Mohr.
Giere, Ronald N. (1988). Explaining Science. A cognitive Approach. Chicago – London:
University of Chicago Press.
Horden, Peregrine, Nicholas Purcell (2000). The Corrupting Sea. A Study of
Mediterranean History. Oxford: Blackwell, 2000.
Johnson, Samuel (1781) .Lives of the Poets (Addison, Savage, Switf). (approached 2007-07-25)
Kahlert, Robert C., Ben Rode, David Baxter, Michael Witbrock, Ken Forbus, Larry
Birnbaum, Purvesh Shah, Dave Schneider, Kathy Panton, Alan Belasco, David
Crabbe (2006). Tracking Quantity Fluctuations using STT. In: Proceedings of the
2006 AAAI Fall Symposium on Evidence Extraction, Arlington VA.
Klein, Gary (1999). Sources of Power. How People Make Decisions. : MIT Press.
Laudon, Larry (1996). Beyond Positivism and Relativism. Theory, Method and Evidence.
Boulder, Co.: Westview Press.
Lehnert, Wendy (1994). Cognition, Computers and Car Bombs: How Yale Prepared Me
for the 1990s, 143-174. In: Roger Schank, Ellen Langer (ed), Beliefs, Reasoning and
Decision Making: Psycho-Logic in Honor of Bob Abelson, Hillsdale, NJ: Lawrence
Lübbe, Hermann (1977). Geschichtsbegriff und Geschichtsinteresse. Analytik und
Pragmatik der Historie. Basel – Stuttgart: Schwabe & Co.
Minsky, Marvin (1974). A Framework for Representing Knowledge. MIT AI Laboratory
Memo 306, June 1974.
Russell, Stuart J., Peter Norvig (2002). AI: A Modern Approach. New Jersey: Prentice
Schank, Roger, Robert Abelson (1977). Scripts, Plans, Goals and Understanding. An
Inquiry into Human Knowledge Structures. Hillsdale, NJ: Lawrence Erlbaum.
Van Frassen, Bas C. The Scientific Image. Oxford : Carendon Press.
18 Originally published in: Rescher, Nicholas (ed.) (1967), The Logic of Decision and
Action. Pittsburg: University of Pittsburg Press.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Introduction The Sins of the Fathers: Positivist Origins of Postpositivist Relativisms Theory And Evidence Demystifying Underdetermination Empirical Equivalence and Underdetermination (with Jarrett Leplin.) Methods And Progress A Problem-Solving Approach to Scientific Progress For Method: Answering the Relativist Critique of Kuhn and Feyeraband Reconciling Progress and Loss Choosing The Aims And Methods Of Science Progress or Rationality? The Prospects for Normative Naturalism The Rational Weight of the Scientific Past: Forging Fundamental Change in a Conservative Discipline Normative Naturalism: Replies to Friendly Critics History And Sociology Of Science The Pseudo-Science of Science? The Demise of the Demarcation Problem Science at the BarCauses for Concern Dominance and the Disunity of Method: Solving the Problems of Innovation and Consensus (with Rachel Laudan.).
Full-text available
We describe an approach to extracting and tracking events in which measurable quantities such as economic indicators undergo a change. The causes and consequences of oil price fluctuations are an example of such events, and are tracked in the STT (Situation Tracking Testbed) prototype. We propose a representation of these event types in the logical representation language CycL and apply this representation to the problem of identifying and extracting information from news sources in RSS feed repositories.
This volume collects Davidson's seminal contributions to the philosophy of mind and the philosophy of action. Its overarching thesis is that the ordinary concept of causality we employ to render physical processes intelligible should also be employed in describing and explaining human action. In the first of three subsections into which the papers are thematically organized, Davidson uses causality to give novel analyses of acting for a reason, of intending, weakness of will, and freedom of will. The second section provides the formal and ontological framework for those analyses. In particular, the logical form and attending ontology of action sentences and causal statements is explored. To uphold the analyses, Davidson urges us to accept the existence of non‐recurrent particulars, events, along with that of persons and other objects. The final section employs this ontology of events to provide an anti‐reductionist answer to the mind/matter debate that Davidson labels ‘anomalous monism’. Events enter causal relations regardless of how we describe them but can, for the sake of different explanatory purposes, be subsumed under mutually irreducible descriptions, claims Davidson. Events qualify as mental if caused and rationalized by reasons, but can be so described only if we subsume them under considerations that are not amenable to codification into strict laws. We abandon those considerations, collectively labelled the ‘constitutive ideal of rationality’, if we want to explain the physical occurrence of those very same events; in which case we have to describe them as governed by strict laws. The impossibility of intertranslating the two idioms by means of psychophysical laws blocks any analytically reductive relation between them. The mental and the physical would thus disintegrate were it not for causality, which is operative in both realms through a shared ontology of events.
Introduction. Acknowledgements. Note on References and Abbreviations. Lists of Illustrations. Part I: 'Frogs Round a Pond': Ideas of the Mediterranean: 1. A Geographical Expression. 2. A Historian's Mediterranean. Part II: 'Shory Distances and Definite Places": Mediterranean Microecologies: 3. Four Definite Places. 4. Ecology and the Larger Settlement. 5. Connectivity. Part III: Revolution and Catastrophe: 6. Imperatives of Survival: Diversify, Store, Redistribute. 7. Technology and Agrarian Change. 8. Mediterranean Catastrophes. 9. Mobility of Goods and People. Part IV: The Geography of Religion: 10.'Territories of Grace'. Part V: 'Museums of Man': The Uses of Social Anthropology: 11.'Mists of Time': Anthropology and Continuity. 12.'I also Have a Moustache' : Anthropology and Mediterranean Unity. Bibliographical Essays. Consolidated Bibliography. Index.
For both people and machines, each in their own way, there is a serious problem in common of making sense out of what they hear, see, or are told about the world. The conceptual apparatus necessary to perform even a partial feat of understanding is formidable and fascinating. Our analysis of this apparatus is what this book is about. —Roger C. Schank and Robert P. Abelson from the Introduction (
Knowledge representation is at the very core of a radical idea for understanding intelligence. Instead of trying to understand or build brains from the bottom up, its goal is to understand and build intelligent behavior from the top down, putting the focus on what an agent needs to know in order to behave intelligently, how this knowledge can be represented symbolically, and how automated reasoning procedures can make this knowledge available as needed. This landmark text takes the central concepts of knowledge representation developed over the last 50 years and illustrates them in a lucid and compelling way. Each of the various styles of representation is presented in a simple and intuitive form, and the basics of reasoning with that representation are explained in detail. This approach gives readers a solid foundation for understanding the more advanced work found in the research literature. The presentation is clear enough to be accessible to a broad audience, including researchers and practitioners in database management, information retrieval, and object-oriented systems as well as artificial intelligence. This book provides the foundation in knowledge representation and reasoning that every AI practitioner needs.