Showing papers on "Abductive reasoning published in 2004"
Book•
01 Jan 2004TL;DR: 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, and offers the first true synthesis of the field in over a decade.
Abstract: 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.
*Authors are well-recognized experts in the field who have applied the techniques to real-world problems
* Presents the core ideas of KR&R in a simple straight forward approach, independent of the quirks of research systems
*Offers the first true synthesis of the field in over a decade
Table of Contents
1 Introduction * 2 The Language of First-Order Logic *3 Expressing Knowledge * 4 Resolution * 5 Horn Logic * 6 Procedural Control of Reasoning * 7 Rules in Production Systems * 8 Object-Oriented Representation * 9 Structured Descriptions * 10 Inheritance * 11 Numerical Uncertainty *12 Defaults *13 Abductive Reasoning *14 Actions * 15 Planning *16 A Knowledge Representation Tradeoff * Bibliography * Index
938 citations
07 Jun 2004
TL;DR: This work presents an approach by which a formal representation of a system, based on the event calculus, can be used in conjunction with abductive reasoning techniques to derive the sequence of operations that will allow a given system to achieve a desired goal.
Abstract: As the interest in using policy-based approaches for systems management grows, it is becoming increasingly important to develop methods for performing analysis and refinement of policy specifications. Although this is an area that researchers have devoted some attention to, none of the proposed solutions address the issue of deriving implementable policies from high-level goals. A key part of the solution to this problem is having the ability to identify the operations, available on the underlying system, which can achieve a given goal. This work presents an approach by which a formal representation of a system, based on the event calculus, can be used in conjunction with abductive reasoning techniques to derive the sequence of operations that will allow a given system to achieve a desired goal. Additionally it outlines how this technique might be used for providing tool support and partial automation for policy refinement. Building on previous work on using formal techniques for policy analysis, the approach presented here applies a transformation of both policy and system behaviour specifications into a formal notation that is based on event calculus. Finally, it shows how the overall process could be used in conjunction with UML modelling and illustrates this by means of an example.
178 citations
TL;DR: The authors compared inductive, deductive, and abductive reasoning, and argued that abductive inference is the proper technique when nothing is known about the research at the outset, and compared the three methods from scratch.
Abstract: This article looks at the process of doing research ‘from scratch'. The author began a project investigating children of Ethiopian origin living in Israel to see how ones who attended a kindergartern program years earlier differed from those who had not attended. However, the problem from the outset was that there may not be a difference to find. In this article, the author compares inductive, deductive, and abductive reasoning, and argues that abductive reasoning is the proper technique when nothing is known about the research at the outset.
105 citations
TL;DR: In this paper, an overview of various usages of abduction, from Eco to Bhaskar, is followed by a critical pragmatist assessment of its practical usefulness in daily affairs in determining that "which is the case".
Abstract: Among the classical pragmatists Charles Sanders Peirce is often regarded as the most removed from practical affairs; he is seen as the most ‘scientific’. The aim of this article is to show his practical usefulness in daily affairs in determining that ‘which is the case’. The abductive inference had a pivotal role in his pragmatist theory of inquiry. As Umberto Eco has shown, abduction is as equally central in the writing of detective stories as it is in the making of science. Recently, abduction has gained a renewed interest not only in the idea that science is basically ‘conjectural’ (Ginzburg and Popper), but also in different versions of critical realism. Here abduction gains interest as a ‘theoretical inference’ of special concern for the social sciences. An overview of various usages of abduction, from Eco to Bhaskar, is followed by a critical pragmatist assessment.
78 citations
TL;DR: The concept of manipulative abduction is devoted to capture the role of action in many interesting situations: action provides otherwise unavailable information that enables the agent to solve problems by starting and performing a suitable abductive process of generation or selection of hypotheses.
Abstract: What I call theoretical abduction (sentential and model-based)certainly illustrates much of what is important in abductive reasoning, especially the objective of selecting and creating a set of hypotheses that are able to dispense good (preferred) explanations of data, but fails to account for many cases of explanation occurring in science or in everyday reasoning when the exploitation of the environment is crucial. The concept of manipulative abduction is devoted to capture the role of action in many interesting situations: action provides otherwise unavailable information that enables the agent to solve problems by starting and performing a suitable abductive process of generation or selection of hypotheses. Many external things, usually inert from the epistemological point of view, can be transformed into what I callepistemic mediators, which are illustrated in the last part of the paper, together with an analysis of the related notions of ``perceptual and inceptual rehearsal'' and of ``external representation''.
64 citations
27 Sep 2004
TL;DR: In this article, a new proof procedure for abductive logic programming is presented, which extends that of Fung and Kowalski by integrating abductive reasoning with constraint solving and relaxing the restrictions on allowed inputs for which the procedure can operate correctly.
Abstract: We introduce a new proof procedure for abductive logic programming and present two soundness results. Our procedure extends that of Fung and Kowalski by integrating abductive reasoning with constraint solving and by relaxing the restrictions on allowed inputs for which the procedure can operate correctly. An implementation of our proof procedure is available and has been applied successfully in the context of multiagent systems.
59 citations
01 Jan 2004
TL;DR: It is argued that the goals of the user should be taken into account when deciding what is a good explanation for a given Case-Based Reasoning (CBR) system.
Abstract: In this paper, we present a short overview of different theories of explanation. We argue that the goals of the user should be taken into account when deciding what is a good explanation for a given CBR system. Some general types relevant to many Case-Based Reasoning (CBR) systems are identified and we use these goals to identify some limitations in using the case as an explanation in CBR systems.
38 citations
05 Feb 2004
TL;DR: In this article, a man named Albert is in a bit of a hurry but could probably arrive on time going either way, and decides to take the scenic eastern route to get to Boston.
Abstract: Albert thinks about what route to take to get to Boston. He thinks that, while the direct western route is faster, the scenic eastern route is longer but more enjoyable with less traffic. He is in a bit of a hurry but could probably arrive on time going either way. He eventually reaches a decision. The reasoning Albert goes through in settling on what route to take is practical. He is deciding what to do. At about the same time, Albert's friend Betty tries to decide what route Albert will take. She thinks about what Albert has done before, what Albert likes in a route, and how much of a hurry Albert is in. Betty's reasoning is theoretical. She is trying to arrive at a belief about what Albert will do. Practical reasoning in this more or less technical sense leads to (or modifies) intentions, plans, and decisions. Theoretical reasoning in the corresponding technical sense leads to (or modifies) beliefs and expectations. There is also the possibility that reasoning of either sort leaves things unchanged. Any given instance of reasoning may combine both theoretical and practical reasoning. In deciding which route to take, Albert may have to reach theoretical conclusions about how long it will take to go by the eastern route. In thinking about
38 citations
17 Jun 2004
TL;DR: Knowledge representation is at the very core of a radical idea for understanding intelligence as discussed by the authors, which 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.
Abstract: 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.
*Authors are well-recognized experts in the field who have applied the techniques to real-world problems
* Presents the core ideas of KR&R in a simple straight forward approach, independent of the quirks of research systems
*Offers the first true synthesis of the field in over a decade
Table of Contents
1 Introduction * 2 The Language of First-Order Logic *3 Expressing Knowledge * 4 Resolution * 5 Horn Logic * 6 Procedural Control of Reasoning * 7 Rules in Production Systems * 8 Object-Oriented Representation * 9 Structured Descriptions * 10 Inheritance * 11 Numerical Uncertainty *12 Defaults *13 Abductive Reasoning *14 Actions * 15 Planning *16 A Knowledge Representation Tradeoff * Bibliography * Index
38 citations
Journal Article•
TL;DR: The exact nature of abduction is a controversial issue as discussed by the authors. But the concept has its critics: some have maintained that it is not a viable mode of inference, especially if abduction is presented as a logic of discovery (Frankfurt 1958; Kapitan 1990, 1992).
Abstract: Abductive inference is nowadays a controversial issue, despite its application in a variety of fields. In philosophy, logic, and artificial intelligence, some researchers have taken a great deal of interest in it (see, e.g., Josephson & Josephson 1994; Niiniluoto 1999a; Flach & Kakas 2000a; Magnani 2001; Aliseda 2003). But the concept has its critics: Some have maintained that it is not a viable mode of inference, especially if abduction is presented as a logic of discovery (Frankfurt 1958; Kapitan 1990, 1992). So it seems that abduction brings forth and conceptualizes issues that are highly important; according to Hintikka, abduction is "the fundamental problem of contemporary epistemology" (Hintikka 1998); but still, the exact nature of abduction is a contentious matter.
35 citations
TL;DR: The results of the study suggest production and causal inference as general nonmonotonic formalisms providing an alternative representation for a significant part of nonMonotonic reasoning.
TL;DR: This paper motivates and describes how the Why2-Atlas tutoring system creates and uses a deeper proof-based representation of student essays in order to provide students with substantive feedback on their explanations.
Abstract: The Why2-Atlas tutoring system presents students with qualitative physics questions and encourages them to explain their answers through natural language. Although there are inexpensive techniques for analyzing explanations, we claim that better understanding is necessary for use within tutoring systems. In this paper we motivate and describe how the system creates and uses a deeper proof-based representation of student essays in order to provide students with substantive feedback on their explanations. We describe in detail the abductive reasoner, Tacitus-lite+, that we use within the tutoring system. We also discuss evaluation results for an early version of the Why2-Atlas system and a subsequent evaluation of the theorem-proving module. We conclude with the discussion of work in progress and additional future work for deriving more benefits from a proof-based approach for tutoring applications.
TL;DR: The rationale for the present volume is simple: the great majority of the everyday reasoning, including that of expert groups engaged in their professions, is informal; by contrast, most of the studies of human inference reported by psychologists in the literature are of formal reasoning.
Abstract: The rationale for the present volume is simple: The great majority of the everyday reasoning, including that of expert groups engaged in their professions, is informal. By contrast, most of the studies of human inference reported by psychologists in the literature are of formal reasoning. This discrepancy provides considerable cause for concern and not only because cognitive psychology should have some practical application. Excessive focus on formal reasoning tasks has also, in our view, inhibited the development of good theories of human reasoning. What Is Informal Reasoning, and Why Do We Need to Study It? Psychological studies of formal reasoning have fallen largely into two domains: deductive reasoning and statistical inference. These two endeavours have much in common and some researchers work in both areas. In both cases, participants are presented with what problem-solving researchers call well-defined problems. A well-defined problem can be solved by use of the information provided and no other; in fact, the correct solution to these problems often requires the reasoner to use only the information provided in the premises, and to avoid adding background information and knowledge to the problem domain. Instead, a correct solution is achieved by applying a normatively appropriate rule of inference. Normative systems are often applied to formal reasoning problems in order to define solutions as right or wrong, such that these problems are then construed as tests of correct and fallacious reasoning. Hence, these problems are designed to measure the extent to which participants bring to the laboratory an understanding and ability to apply - the relative normative principles. In the case of deductive reasoning research, the relevant normative system is formal logic. Participants are given some premises and asked whether a conclusion follows. Under strict deductive reasoning instructions, they are told (a) to assume that the premises are true and (b) to draw or approve only conclusions that necessarily follow. As observed elsewhere (Evans, 2002), this widely used method was developed over 40 years ago when belief in logic as a normative and descriptive system for human reasoning was veiy much higher than it is today. In spite of the method, much evidence has emerged to support the conclusion that pragmatic factors play a large part in human reasoning. We say "in spite of" because standard deductive instructions aim to suppress precisely those factors that dominate informal reasoning: the introduction of prior belief and the expression of uncertainty in premises and conclusions. In research on statistical inference, a similar story is found. People are asked to make statistical inference on the basis of well-defined problems, in which relevant probabilities or frequency distributions are provided, and their answers are assessed for correctness against the norms provided by the probability calculus. Research in this tradition has been mostly conducted by researchers in the "heuristics and biases" tradition inspired by the work of Danny Kahneman and Amos Tversky (Gilovich, Griffin, & Kahneman, 2002; Kahneman, Slovic, & Tversky, 1982). This results in an arguably negative research strategy that is similar to much work on deductive reasoning. That is, researchers show primarily what people cannot do (conform to the principles of logic or probability theory) and only secondarily address what people actually do. Indeed, one of the most common explanations for why intelligent, educated individuals often fail to reason normatively is that they use informal reasoning processes to solve formal reasoning tasks. For example, notwithstanding instructions to the contraiy, reasoners often supplement the information they are provided with background knowledge and beliefs, and make inferences that are consistent with, rather than necessitated by, the premises. If this is the case, it is reasonable to suggest that we study these processes directly, by giving our participants tasks that allow them to express these types of behaviours freely, rather than indirectly, via the observation of poor performance on a formal task. …
Book•
01 Oct 2004
TL;DR: Aristotle on false reasoning, Aristotle on False Reasoning, and Xenophon on false Reasoning as discussed by the authors, and δεσφεραβαβηλαβεβαη βεβε βδεββδαβϵ βαδ βαβββα βε βηγεβ ββεββ βαγα ββαγ βαλα βγαβγαγβα αβγγαα β
Abstract: Aristotle on false reasoning , Aristotle on false reasoning , کتابخانه دیجیتال و فن آوری اطلاعات دانشگاه امام صادق(ع)
TL;DR: This paper presents a general framework for modelling limited reasoning based on approximate reasoning and discusses its properties, starting from Cadoli and Schaerf's approximate entailment and proposing a more general system.
Abstract: Real agents (natural or artificial) are limited in their reasoning capabilities. In this paper, we present a general framework for modelling limited reasoning based on approximate reasoning and discuss its properties. We start from Cadoli and Schaerf's approximate entailment. We first extend their system to deal with the full language of propositional logic. A tableau inference system is proposed for the extended system together with a subclassical semantics; it is shown that this new approximate reasoning system is sound and complete with respect to this semantics. We show how this system can be incrementally used to move from one approximation to the next until the reasoning limitation is reached. We also present a sound and complete axiomatization of the extended system. We note that although the extension is more expressive than the original system, it offers less control over the approximation process. We then propose a more general system and show that it keeps the increased expressivity and recovers the control. A sound and complete formulation for this new system is given and its expressivity and control advantages are formally proved.
01 Jan 2004
TL;DR: The goal of this paper is to serve as a survey for the problem of abductive inference (or belief revision) in Bayesian networks by introducing the problem in its two variants: total abduction and partial abduction.
Abstract: The goal of this paper is to serve as a survey for the problem of abductive inference (or belief revision) in Bayesian networks. Thus, the problem is introduced in its two variants: total abduction (or MPE) and partial abduction (or MAP) . Also, the problem is formulated in its general case, that is, looking for the K best explanations. Then, a (non exhaustive) review of exact and approximate algorithms for dealing with both abductive inference problems is carried out. Finally, we collect the main complexity results appeared in the literature for both problems (MPE and MAP).
TL;DR: It will be shown how legal justification with the proposed approach could also take into consideration the credibility of the persons involved in the trial, and the use of ALIAS agents in legal justification allows for plausible explanations for observed pieces of evidence, and to detect collusions or inconsistencies among trial characters.
Abstract: In this paper, we present a new approach to the modeling of the judicial evaluation of criminal evidence, an approach based on an abductive multi-agent system. Legal justification in such a context has, as its main objective, the finding of relatively most plausible explanations for given pieces of evidence, especially (but not necessarily) in a criminal trial. In our approach, the process of justification is carried out through the collaborative abductive reasoning of agents, operating within a logic-based architecture called ALIAS. This enables us the modular composition of the knowledge of the different dramatis personae involved in the trial: the detective, witnesses, suspects, judges, and so forth. Having represented each actor in the trial by a single ALIAS agent, we are able to dynamically group and coordinate them for the explanation of goals (such as, for instance, pieces of evidence or any given observation). We tested our proposed approach on the Peyer case, which was tried in California. It is ...
29 Sep 2004
TL;DR: It is argued that a logic-based planner, defined as the application of general purpose theorem proving techniques to a general purpose action formalism, can be a very solid base for the research on extending the classical planning approach.
Abstract: In this work we show how a planner implemented as an abductive reasoning process can have the same performance and behavior as classical planning algorithms. We demonstrate this result by considering three different versions of an abductive event calculus planner on reproducing some important comparative analyses of planning algorithms found in the literature. We argue that a logic-based planner, defined as the application of general purpose theorem proving techniques to a general purpose action formalism, can be a very solid base for the research on extending the classical planning approach.
TL;DR: This paper proposes a framework for information exchange among abductive agents whose local knowledge bases are enlarged with a set of abduced hypotheses, and integrates the aspects of information exchange and abductive reasoning.
Abstract: In this paper, we propose a framework for information exchange among abductive agents whose local knowledge bases are enlarged with a set of abduced hypotheses. We integrate the aspects of information exchange and abductive reasoning, and show theoretically the information inferred by the single abductive agent as a product of joint reasoning activity. We show examples, like dining philosophers, resource exchange and speculative computation, and give an implementation of the space of interactions based on CLP(SET).
07 Jun 2004
TL;DR: This paper introduces an ID-Logic based framework for database schema integration that allows to uniformly represent and reason with independent source databases that contain information about a common domain, but may have different schemas.
Abstract: ID-Logic is a knowledge representation language that extends first-order logic with non-monotone inductive definitions. This paper introduces an ID-Logic based framework for database schema integration. It allows us to to uniformly represent and reason with independent source databases that contain information about a common domain, but may have different schemas. The ID-Logic theories that are obtained are called mediator-based systems. We show that these theories properly capture the common methods for data integration (i.e., global-as view and local-as-view with either exact or partial definitions), and apply on them a robust abductive inference technique for query answering.
Book Chapter•
01 Sep 2004
TL;DR: Kankkunen et al. as discussed by the authors found that concept mapping provided a means for students to discover tentative meanings for the concepts taught, and Peirce's semiotic paradigm provided a pragmatic framework for tracking the process of "updating meanings" which is intrinsic to learning.
Abstract: his paper grapples with the problem of how to track a student’s real progress in learning, which cannot be absolutely quantified at any given point as a result of a particular intervention. Here, only some results are presented for a long-term qualitative and quantitative classroom study, during which the method of concept mapping was applied and interpreted in light of the semiotic paradigm developed by Charles Sanders Peirce (1931–1958). Peirce’s semiotic paradigm was thought to have sufficient rigour and flexibility to give new access to the multiplicity of processes at work in the classroom.
A natural learning environment was built over a four-year period in a Finnish primary school. The students, ranging in age from 9 to 12 years, were encouraged to use qualitative judgement (intuition, tacit knowledge) to give them greater intellectual access to the meanings of the concepts taught. The goal was to bring them to Vygotsky’s stage of ‘conceptual learning’, and to evaluate the effectiveness of concept mapping as an ‘advance organiser’ used in conjunction with Peirce’s semiotic paradigm.
This paper concentrates on the theory building of concept mapping. However, the four-year longitudinal study results (Kankkunen 1999; 2001) show that concept mapping provided a means for students to discover tentative meanings for the concepts taught. In parallel, Peirce’s semiotic paradigm provided a pragmatic framework for tracking the process of ‘updating meanings’ (the habit of changing habits) which is intrinsic to learning. The marriage of Peircean theory and concept mapping is lasting.
Keywords: abductive reasoning, concept mapping, connection-making, establishing meaning, learning environment for understanding, longitudinal study, students as evaluators
01 Jan 2004
TL;DR: The CADRE system (Continuous Analysis and Discovery from Relational Evidence) addresses this deficiency by automating the link analysis process by combining an expressive knowledge representation of threat patterns with efficient, constraint-based abductive reasoning algorithms to automatically infer links and construct coherent threat hypotheses from structured data.
Abstract: Intelligence agencies are under increasing pressure to “connect the dots” between fragments of evidence from disparate sources to enable preemption of potential threats such as terrorist attacks. Most systems for threat detection in use today provide only data visualization tools for manual “link analysis,” leading to methods that do not scale to massive data sets. The CADRE system (Continuous Analysis and Discovery from Relational Evidence) addresses this deficiency by automating the link analysis process. CADRE combines an expressive knowledge representation of threat patterns with efficient, constraint-based abductive reasoning algorithms to automatically infer links and construct coherent threat hypotheses from structured data. A compact, factored representation of multiple hypotheses avoids redundant storage and enables scaling to large data sets. CADRE efficiently manages the growth of the hypotheses using probabilistic evaluation models and a consistency checking algorithm to prune unlikely hypotheses.
Proceedings Article•
01 Jan 2004
TL;DR: This paper presents a computational model for reasoning with causal explanations of observations within the framework of Abductive Event Calculus (AEC), based on abductive reasoning based on the notions of ”deserts” and ”oases” on the time line.
Abstract: This paper presents a computational model for reasoning with causal explanations of observations within the framework of Abductive Event Calculus (AEC). The model is based on abductive reasoning based on the notions of ”deserts” and ”oases” on the time line. Our work is motivated from the need to recover from the inconsistency that can arise when observations of fluents are added to the narrative of a domain description. We study how such observations can be assimilated via abductive explanations in order to render the domain frame consistent. Typically, such explanations would involve non-ground events whose time of occurrence can only be constraint within some interval. We present some notions of minimal commitment for such explanations and study how we can reason and compute with these explanations once they have been chosen and added to the theory. The computational model proposed can be readily implemented by exploiting, in a modular way, any of the different computational models for Abductive Logic Programming or for Answer Set Programming, augmented, again in a modular way, by suitable forms of temporal constraint solving.
01 Jan 2004
TL;DR: A new proof procedure for abductive logic programming is introduced by integrating abductive reasoning with constraint solving and by relaxing the restrictions on allowed inputs for which the procedure can operate correctly.
Abstract: We introduce a new proof procedure for abductive logic programming and prove two soundness results. Our procedure extends that of Fung and Kowalski by integrating abductive reasoning with constraint solving and by relaxing the restrictions on allowed inputs for which the procedure can operate correctly. An implementation of our proof procedure is available and has been applied successfully in the context of
TL;DR: This work considers the problem of embedding abduction of surprising and anomalous observations in defeasible (nonmonotonic) theories and explores the use of partial structures approach as a semantic foundation for the system.
01 Jan 2004
TL;DR: A hybrid symbolic-connectionist learning architecture for multicausal abduction is proposed, which tightly integrates a symbolic Soar model for generating and modifying hypotheses with Echo, a connectionist model for evaluating hypotheses.
Abstract: Multicausal abductive tasks appear to have deliberate and implicit components: people generate and modify explanations using a series of recognizable steps, but these steps appear to be guided by an implicit hypothesis evaluation process. This paper proposes a hybrid symbolic-connectionist learning architecture for multicausal abduction. The architecture tightly integrates a symbolic Soar model for generating and modifying hypotheses with Echo, a connectionist model for evaluating hypotheses. The symbolic component uses knowledge compilation to quickly acquire general rules for generating and modifying hypotheses, and for making decisions based on the current best explanation. The connectionist component learns to provide better hypothesis evaluation by implicitly acquiring explanatory strengths based on the frequencies of events during problem solving.
01 Jan 2004
TL;DR: For instance, the CogSci 2004 Symposium on Abduction and Creative Inferences in Science as discussed by the authors explores abduction (inference to explanatory hypotheses), an important but neglected topic in scientific reasoning, and develops important ideas about aspects of abductive reasoning that have been relatively neglected in philosophy of science.
Abstract: CogSci2004 Symposium Abduction and Creative Inferences in Science Lorenzo Magnani (lorenzo.magnani@unipv.it) - Organizer, University of Pavia, Italy Atocha Aliseda (atocha@filosoficas.unam.mx), UNAM. Mexico City, Mexico Thomas Addis (tom.addis@port.ac.uk) and David Gooding (hssdcg@bath.ac.uk), University of Portsmouth, Portsmouth, UK and University of Bath, Bath, UK John Woods (jhwoods@interchange.ubc.ca) and Dov Gabbay (dg@dcs.kcl.ac.uk), University of British Columbia, CA and King’s College London, UK Joke Meheus (joke.meheus@rug.ac.be), Ghent University, Ghent, Belgium Matti Sintonen and Sami Paavola (matti.sintonen@helsinki.fi,sami.paavola@helsinki.fi, - Discussants, University of Helsinki, Finland The symposium aims to explore abduction (inference to explanatory hypotheses), an important but neglected topic in scientific reasoning. The aim is to integrate philosophi- cal, cognitive, and computational issues. The main thesis is that abduction is a significant kind of scientific reasoning, helpful in delineating the first principles of a new theory of science. The status of abduction is very controversial. When dealing with abductive reasoning misinterpretations and equivocations are common. What are the differences between abduction and induction? What are the differ- ences between abduction and the well-known hypothetico- deductive method? What did Peirce mean when he consid- ered abduction a kind of inference? Does abduction in- volve only the generation of hypotheses or their evaluation too? Are the criteria for the best explanation in abductive reasoning epistemic, or pragmatic, or both? How many kinds of abduction are there? The symposium aims to in- crease knowledge about creative and expert inferences. The study of these high-level methods of abductive rea- soning is situated at the crossroads of philosophy, episte- mology, artificial intelligence, cognitive psychology, and logic; that is at the heart of cognitive science. More than a hundred years ago, the great American philosopher Charles Sanders Peirce coined the term “ab- duction” to refer to inference that involves the generation and evaluation of explanatory hypotheses. The study of abductive inference was slow to develop, as logicians con- centrated on deductive logic and on inductive logic based on formal calculi such as probability theory. In recent dec- ades, however, there has been renewed interest in abduc- tive inference from two primary sources. Philosophers of science have recognized the importance of abduction in the discovery and evaluation of scientific theories, and researchers in artificial intelligence have realized that ab- duction is a key part of medical diagnosis and other tasks that require finding explanations. Psychologists have been slow to adopt the terms “abduction” and “abductive infer- ence” but have been showing increasing interest in causal and explanatory reasoning. Thus abduction is now a key topic of research in phi- losophy of science. First, this symposium ties together the concerns of philosophers of science and logicians, show- ing, for example, the connections between formal models and abduction (Meheus, Woods and Gabbay). Second, it lays out a useful general framework for discussion of vari- ous kinds of abduction (Magnani), such as model-based and manipulative abductions. Third, it develops important ideas about aspects of abductive reasoning that have been relatively neglected in philosophy of science, including the role of testing in abductive inference (Aliseda), and the interrogative model of inquiry and the role of different kinds of why-questions and strategic principles employed in attempts to find and construct answers also at the com- putational level (Sintonen and Paavola, Addis and Good- ing). The clarification of these topics aims to increase knowledge about some aspects of explanatory reasoning and hypothesis formation very relevant in many epistemic tasks. 1. If we stress the concept of model-based and manipu- lative abduction (Magnani), creative inferences in science can be seen as formed by the application of heuristic (strate- gic) procedures that involve all kinds of good and bad infer- ential actions and both internal and external representations, and not only the mechanical application of rules. 2. Recent logical models can illustrate in a rigorous way how these (strategic) abductive steps are combined with deductive steps (Meheus, Woods and Gabbay). 3. Common to all abduction problems is a cognitive tar- get that cannot be hit on the basis of what the abducer presently knows. Abductive hypotheses do not enhance a reasoner’s knowledge. Abduction, accordingly, is igno- rance-preserving inference. These abductive processes are dynamical (Woods and Gabbay). 4. The “abductive steps” are also analyzable in terms of responses to surprising singular or general facts, showing a connection to explanation-seeking why-questions (Sinto- nen and Paavola). 5. The importance of experimental verification for hy- potheses evaluation in science is stressed by the relation- ship between abduction and pragmatism in Peirce (Al- iseda). 6. Abduction cannot be thought of in isolation from the two other type of inference (deduction and induc- tion/validation) identified by Peirce. Computer models of scientific behaviour and music conversation suggest that in simulation of abduction requires the use of mixed strate- gies using random actions as suggested by game theory (Addis and Gooding).
01 Jan 2004
TL;DR: This work introduces a new proof procedure for abductive logic programming which it calls CIFF, which extends the IFF procedure of Fung and Kowalski by integrating abductive reasoning with constraint solving.
Abstract: Abduction has found broad application as a powerful tool for hypothetical reasoning with incomplete knowledge, which can be handled by labelling some pieces of information as abducibles, i.e. as possible hypotheses which can be assumed to hold, provided that they are consistent with the given knowledge base. Abductive Logic Programming (ALP) [4] combines abduction with logic programming enriched with integrity constraints to further restrict the range of possible hypotheses. We introduce a new proof procedure for abductive logic programming which we call CIFF. Our procedure extends the IFF procedure of Fung and Kowalski [3] by integrating abductive reasoning with constraint solving. Another feature of our approach is that we do not attempt to provide a static characterisation of the class of allowed inputs on which the procedure can operate correctly, but rather check allowedness dynamically during a derivation. This allows us to cover a larger class of inputs.