TL;DR: The problems and solutions presented in this paper are related to the more general question of how modelling previous knowledge facilitates instruction in a new skill.
Abstract: This paper describes VP2, a system that has been implemented to tutor nonnative speakers in English. This system differs from many tutoring systems by employing an explicit grammar of its user's native language. This grammar enables VP2 to customize its responses by addressing problems due to interference of the native language. The system focuses on the acquisition of English verb-particle and verb-prepositional phrase constructions. Its correction strategy is based upon comparison of the native language grammar with an English grammar. VP2 is a modular system: its grammar of a user's native language can easily be replaced by a grammar of another language. The problems and solutions presented in this paper are related to the more general question of how modelling previous knowledge facilitates instruction in a new skill.
TL;DR: A shell for expert systems written in Prolog provides a consultation environment and a range of explanation capabilities and there is a new approach for explaining ‘why not,’ when a query to the system fails.
Abstract: We describe a shell for expert systems written in Prolog. The shell provides a consultation environment and a range of explanation capabilities. The design of the shell is modular, making it very easy to extend the shell with extra features required by a particular expert system. The novelty of the shell is twofold. Firstly, it has a uniform design consisting of an integrated collection of meta-interpreters. Secondly, there is a new approach for explaining ‘why not,’ when a query to the system fails.
Nous decrivons un interprete de commandes pour systemes experts Merits en Prolog. L'interprete founit un environnement de consultation et un eventail de capacites ? explications. II est concu de facon modulaire, ce qui rend tres facile ľ 'ajout de particularites requises par un systeme expert particulier. Cet interprete innove de deux facons: d'une part, par sa conception uniforme consistant en un ensemble integre de metainterpretes; ? autre part, par une approche nouyelle pour expliquer le « pourquoi pas »? 'une requete qui echoue.
Mots cles: systemes experts, interprete de commandes pour systemes experts, metainterprete, metaconnaissance, programmation logique, Prolog, explication, « pourquoi pas ».
TL;DR: In this paper, the authors describe an object recognition algorithm both on a sequential machine and on a single instruction multiple data (SIMD) parallel processor such as the MIT connection machine, which is shown to run three to four orders of magnitude faster than the sequential version.
Abstract: This paper describes an object recognition algorithm both on a sequential machine and on a single instruction multiple data (SIMD) parallel processor such as the MIT connection machine. The problem, in the way it is presently formulated on a sequential machine, is essentially a propagation of constraints through a tree of possibilities in an attempt to prune the tree to a small number of leaves. The tree can become excessively large, however, and so implementations on massively parallel machines are sought in order to speed up the problem. Two fast parallel algorithms are described here, a static algorithm and a dynamic algorithm. The static algorithm reformulates the problem by assigning every leaf in the completely expanded unpruned tree to a separate processor in the connection machine. Then pruning is done in nearly constant time by broadcasting constraints to the entire SIMD array. This parallel version is shown to run three to four orders of magnitude faster than the sequential version. For large recognition problems which would exceed the capacity of the machine, a dynamic algorithm is described which performs a series of loading and pruning steps, dynamically allocating and deallocating processors through the use of the connection machine's global router communications mechanism.
Ce rapport decrit un algorithme pour la reconnaissance ? objets qui s'adapte a la fois a un ordinateur sequentiel ou la connection machine du MIT, un multiprocesseur parallele du type SIMD. Sur un ordinateurconventionel, ľ algorithme propage des contraintes a travers un arbre de solutions probables de facon aeliminer le plus de branches que possible. Puisque ľ arbre a tendance a devenir enorme, nous avons considere le transfert de ľ algorithme sur un systeme a multiprocesseurs. Deux algorithmes paralleles sont proposes, un sous forme statique, ľ autre dynamique. Ľ algorithme statique assigne un processeur de la connection machine a chacune des options generees dans ľ arbre. Ľ analyze se fait done en un temps quasi constant en transmettant a chacun des processeurs les contraintes necessaires pour trouver la solution. Ceci accelere ľ algorithme par troi.s ou quatre ordres de grandeur par rapport a la version sequentielle. Pour des problemes plus.complexes qui excederaient la capacitye du multiprocesseur, une autre solution assigne de facon dynamique les differents processeurs lors du traitement en utilisant le systeme ? intercommunication global de la connection machine.
Mots cles: traitement parallele, reconnaissance ? objets, connection machine, ordinateur adressable parcontenu, recherche dans des structures a base ? arbres.
Summary
A well-established algorithm for object recognition on a sequential machine was explained and fast connection machine algorithms were devised. First, a static parallel algorithm was discussed, in which the entire tree of possible interpretations is loaded into the connection machine before run time. The search is then done in one step and therefore in constant time.
For problems that are too large to fit in the array, a dynamic algorithm was devised. In this algorithm, the connection machine is loaded with as many levels of the tree as can fit. Then a pruning is done in parallel as in the static algorithm. However, processors that survive the initial pruning step and still represent consistent pairings must find unused processors which can be assigned to the new branches of the tree. This is similar to the way the sequential algorithm first described prunes the search space. Processors that are pruned are deallocated and reused on the next iteration. The router mechanism of the connection machine which is not used in the static algorithm is necessary here to support dynamically allocating and deallocating processors.
TL;DR: The difference in procedures and outcome between formative and summative evaluation practices is identified and these practices are related to the development and validation of expert systems in education.
Abstract: Researchers developing and validating educational products often expect the same field-test activities to provide information on product improvement and product effectiveness. For effective and economical use of resources, these two goals, product improvement and product validation, must be stressed at different times and with different tools and strategies. This article identifies the difference in procedures and outcome between formative and summative evaluation practices and relates these practices to the development and validation of expert systems in education.
Les chercheurs qui developpent et valident des produits educatifs attendent souvent des měmes activity ? experimentation sur le terrain qu'elles procurent des informations sur ľ amelioration du produit et ľ efficacite du produit. Pour ľ utilisation efficace et economique des ressources, ces deux buts, amelioration et validation du produit, doivent ětre mis en avant a differents moments et avec des outils et des strategies differents. Cet article identifie la difference au niveau des procedures et du resultat entre des activites ?evaluation constructive ou descriptive et relie ces pratiques au developpement et a la validation de systemes experts en education.
Mots cles: systemes experts, developpement de produit, validation de produit, modeles de recherche.
TL;DR: The SCENT (Student Computing ENvironmenT) project is concerned with building an intelligent tutoring system to help student programmers debug their LISP programs and necessitates a blackboard‐style scheme to coordinate information dissemination and control flow.
Abstract: The SCENT (Student Computing ENvironmenT) project is concerned with building an intelligent tutoring system to help student programmers debug their LISP programs. The major thrust of current SCENT investigations is into the design of the SCENT advisor which is meant to provide debugging assistance to novice students. Six conceptual levels constitute the advisor. At the lowest level is the “raw data,” consisting of the student's (possibly buggy) program. This can be interpreted by a “program behaviour” level which can produce traces, cross-reference charts, etc. from the student's program. These traces, etc. can be analyzed by “observers” for interesting patterns. At the next level are “strategy judges” and “diagnosticians” which determine which strategy the student has used in his/her program and bugs in this strategy. A “task expert” provides task-specific input into the process of analyzing the student's solution, and a “student-knowledge component” provides student-specific input into this process. Information from the six levels interacts in a variety of ways and control is similarly hierarchical. This necessitates a blackboard-style scheme to coordinate information dissemination and control flow.
This paper discusses the objectives of SCENT and focusses on organizing the process of debugging student programs. A complete example is given to illustrate how entities at the six levels interact and to indicate the kinds of information sharing that occur in the SCENT advisor. The paper concludes with an evaluation of the strengths and weaknesses of this approach to automated debugging, and suggestions about directions for further exploration.
Le projet SCENT (Environnement Informatique pour Etudiant) vise a construire un systeme tutoriel intelligent pour aider les etudiants en programmation a corriger leurs programmes LISP. Le principal effort des etudes actuelles sur SCENT porte sur la structuration du consultant SCENT qui est cense foumir aux etudiants debutants une assistance a la correction de programme. Six niveaux conceptuels forment le consultant. Au plus bas niveau se trouve la « donnee brute », constitutee du programme (eventuellement errone) de ľetudiant. II peut ětre interprete par un niveau de « comportement de programme », qui peut produire des traces, des tableaux de renvoi, etc. a partir du programme de ľetudiant. Ces traces, etc., peuvent ětre analyses par des « observateurs » de schemas pertinents. Au niveau suivant se trouvent des «evaluateurs de strategie » et des « diagnosticiens » qui determinent quelle strategic a utilisee ľetudiant dans son programme et trouvent les erreurs dans cette strategie. Un « expert de tache » foumit au processus ? analyse de la solution de ľetudiant une entree specifique a la tache, et un « composant de savoir de ľetudiant » foumit a ce processus une entree specifique aľetudiant. Ľ information provenant de ces six niveaux interagit de multiples facons et le controle est aussi non hierarchique. Ceci necessite un systeme du genre tableau noir pour coordonner la propagation de ľ information et controler le courant ? information.
Cet article discute des objectifs de SCENT et s'attache aľ organisation du processus de correction des programmes ?etudiant. On donnera un exemple complet pour illustrer comment interagissent les entites aux six niveaux et pour indiquer les types de repartition de ľ information que ľ on retrouve dans le consultant SCENT. Ľ article conclut par une evaluation des avantages et des faiblesses de cette approche de la correction automatique, et par des suggestions pour les orientations que doit prendre une prochaine exploration.
Mots cles: tutorat intelligent, correction automatique, diagnostic des erreurs logiques et conceptuelles, evaluation de strategie, controle sur tableau noir, schematisation de ľetudiant.
TL;DR: A student model based on the student's cognitive processes is described that includes, for each student, his/her domain knowledge and the specific heuristics and it uses meta‐knowledge of problem solving.
Abstract: ELECTRE is a project to build an intelligent tutoring system for learning basic electricity. This paper describes a student model based on the student's cognitive processes. This model includes, for each student, his/her domain knowledge and the specific heuristics. Moreover, it uses meta-knowledge of problem solving. This model is simulated by a knowledge-based system that controls the solving processes by meta-rules. A case study is presented.
ELECTRE est un projet de systeme tutoriel intelligent ? apprentissage de ľelectriciteelementaire. Ce papier decrit une modelisation de ľetudiant basee sur ses processus cognitifs. Ce modele comprend, pour chaque etudiant, les connaissances du domaine ainsi que les heuristiques specifiques. De plus, il utilise la metaconnaissance de resolution de probleme. Ce modele est simule par un systeme expert qui controle les processus de resolution par des metaregles. Une etude de cas est presentee.
Mots cles: systeme tutoriel intelligent, modele cognitif, systeme expert, metaconnaissance, science cognitive, enseignement de lcaron; electricite.
TL;DR: The stratified model which uses a Λ‐categorial language for meaning representation incorporates valuable features of Fregean‐type semantics along with features of situation semantics developed by Barwise and Perry.
Abstract: We are concerned with developing a computational method for selecting possible antecedents of referring expressions over sentence boundaries. Our stratified model which uses a Λ-categorial language for meaning representation incorporates valuable features of Fregean-type semantics (a la Lewis, Montague, Partee, and others) along with features of situation semantics developed by Barwise and Perry. We consider a series of selected two-sentence stories which we use to illustrate referential interdependencies between sentences. We explain the conditions under which such dependencies arise, explain the conditions under which various translations can be performed, and formalize a set of rules which specify how to compute the reference. We restrict our discussion to two-sentence stories to avoid most of the problems inherent in where to look for the reference, that is, how to determine the proper antecedent. We restrict our considerations in this paper to situations where a reference, if it can be computed at all, has a unique antecedent. Thus we consider examples such as John wants to catch a fish. He (John) wants to eat it. and John interviewed a man. The man killed him (John). We then summarize the transformation which encompasses these rules and relate it to the stratified model. We discuss three aspects of this transformation that merit special attention from the computational viewpoint and summarize the contributions we have made. We also discuss the computational characteristics of the stratified model in general and present our ideas for a computer realization; there is no implementation of the t“ratified model at this time.
Nous nous interessons au developpement de methodes computationnelles pour choisir les antecedents possibles de references dans une autre phrase. Notre modele stratifye qui utilise un langage Λ-categoriel pour la representation de sens incorpore des caracteristiques tres utiles des semantiques du type Frege (a la Lewis Montague, Partee et autres) ainsi que des traits des semantiques situationnelles de Barwise et Perry. Nous considerons une serie d' histoires en deux phrases que nous utilisons pour montrer les interdependances referentielles entre phrases. Nous expliquons les conditions sous lesquelles de telles dependances se presentent ainsi que les conditions sous lesquelles diverses traductions peuvent ětre obtenues et formalisons un ensemble de regies qui specifient comment calculer la reference. Nous restreignons notre discussion a des histoires en deux phrases pour eviter autant que possible le probleme de determiner ou il faut regarder pour la reference, c'est-a-dire de determiner le bon antecedent. Dans cet article, nous nous restreignons aux situations ou la reference a, dans le cas ou on peut le calculer, un antecedent unique. Ainsi, nous considerons des exemples tels John wants to catch a fish. He (John) wants to eat it et John interviewed a man. The man killed him (John). Nous resumons ensuite la transformation qui tient compte de ces regies et la mettons en relation avec le modele stratifye. Nous discutons trois aspects de cette transformation qui meritent une attention particuliere du point de vue computationnel et resumons notre contribution. Nous discutons aussi des caracteristiques computationnelles du modele stratifye en general et presentons nos idees concernant leur implantation sur ordinateur; il n'existe jusqu'a present aucune implementation du modele stratifye.
Mots cles: traitement du langage naturel, representation du sens, analyse du discours, anaphore.
TL;DR: Unrestricted gapping Grammars, as implemented in FIGG, can be used to describe grammars (or metagrammars) that utilize the gap concept, such as Gazdar's generalized phrase structure grammarmars.
Abstract: Since Colmerauer's introduction of metamorphosis grammars (MGs), with their associated type-O-like grammar rules, there has been a desire to allow more general rule formats in logic grammars. Gap symbols were added to the MG rule by Pereira, resulting in extraposition grammars (XGs). Gaps, which are referenced by gap symbols, are sequences of zero or more unspecified symbols which may be present anywhere in a sentence or in a sentential form. However, XGs imposed restrictions on the position of gap symbols and on the contents of gaps. With the introduction of gapping grammars (GGs) by Dahl, these restrictions were removed but the rule was still required to possess a nonterminal symbol as the first symbol on the left-hand side. This restriction is removed with the introduction of unrestricted gapping grammars. FIGG, a flexible implementation of gapping grammars, possesses a bottom-up parser which can process a large subset of unrestricted gapping grammars. It can be used to examine the usefulness of unrestricted GGs for describing phenomena of natural languages such as free word order and partially free word/constituent order. Unrestricted gapping grammars, as implemented in FIGG, can also be used to describe grammars (or metagrammars) that utilize the gap concept, such as Gazdar's generalized phrase structure grammars.
Depuis ľ introduction par Colmerauer des grammaires de metamorphose (GM), et des regies de grammaire correspondantes de type-0, on a ressenti le besoin de disposer, pour les grammaires logiques, de structures de regie plus generales. Les symboles vides ont ete ajoutes a la regie GM par Pereira, dormant les grammaires ? extraposition (XGs). Les vides, auxquels on refere par des symboles vides, sont des sequences (possiblement vides) de symboles non specifies qui peuvent ětre presents n'importe ou dans une phrase ou une forme phrastique. Neanmoins, les XGs imposent des restrictions sur la position des symboles vides et sur les contenus de ces suppressions. Avec ľ introduction des gapping grammaires (GGs) par Dahl, ces restrictions ont ete enlevdes mais la regie devait encore posseder un symbole non terminal comme premier symbole de la partie gauche. Cette restriction est supprimee avec ľ introduction des gapping grammaires non-restreintes. FIGG, une implantation flexible des gapping grammaires, possede un analyseur ascendant qui peut traiter un large sous-ensemble des GGs non-restreintes. II peut etre utilise afin ? examiner ľ utilitye des GGs non-restreintes pour decrire les phenomenes du langage naturel tels que ľ ordre libre de mots, et ľ ordre partiellement libre des mots/constituants. Les GGs non-restreintes, telles qu'implantees dans FIGG, peuvent aussi ětre utilisees pour decrire des grammaires (ou des metagrammaires) qui utilisent le concept de vide, comme les grammaires syntagmatiques generalisees de Gazdar.
Mots cles; grammaires logiques, gapping grammaires, analyse, grammaires syntagmatiques generalisees, grammaires de dominance immediate/precedence lineaire, controle procedural.
TL;DR: This illustration examines the search spaces for a bottom‐up parse of a sentence with a series of four strongly equivalent grammars stated in logical languages of increasing expressiveness, each restatement resulting in a more concise grammar and a smaller search space.
Abstract: The primary goal of this paper is to illustrate how smaller deductive search spaces can be obtained by extending a logical language with restricted quantification and tailoring an inference system to this extension. The illustration examines the search spaces for a bottom-up parse of a sentence with a series of four strongly equivalent grammars. The grammars are stated in logical languages of increasing expressiveness, each restatement resulting in a more concise grammar and a smaller search space.
A secondary goal is to point out an area where further research could yield results useful to the design of efficient parsers, particularly for grammatical formalisms that rely heavily on feature systems.
Ľ objectif principal de cet article est de montrer comment on peut obtenir de plus petits espaces de recherche deductive en etendant un langage logique avec une quantification restreinte et en adaptant un systeme ? inference a cette extension. On examine les espaces de recherche crees par un analyse ascendante ? une phrase faite aľ aide de quatre grammaires fortement equivalentes. Chacune des grammaires est definie avec un langage logique plus expressif que celui de la grammaire preedente. Chaque redefinition permet ? obtenir une grammaire plus concise et un espace de recherche plus restreint.
Un des objectifs secondaires de cet article est de pointer les endroits ou des recherches permettraient ? obtenir des resultats utiles pour la conception ? analyseurs efficaces, plus particulierement pour des formalismes grammaticaux qui reposent principalement sur des systemes de traits.
Mots cles: analyse, quantification restreinte, logique avec especes, grammaires logiques, deduction automatique, traitement automatique du langage naturel, linguistique informatique, raisonnement automatise, strategic de recherche a implication minimale, systemes de traits, unification avec especes, report de contraintes.
TL;DR: A problem solver is designed that accumulates knowledge about the pros and cons of alternative selections at choice points during heuristic search, which it updates in light of an evolving problem situation whenever untenable situations arise.
Abstract: Problem solvers that use heuristics to guide choices often run into untenable situations that can be characterized as overconstrained. When this happens, the problem must be able to identify the right culprit from among its heuristic choices in order to avoid a potentially explosive search. In this paper, we present a solution to this for a certain class of problems where the justifications associated with choice points involve an explicit assessment of the pros and cons of choosing each alternative relative to its competitors. We have designed a problem solver that accumulates such knowledge about the pros and cons of alternative selections at choice points during heuristic search, which it updates in light of an evolving problem situation. Whenever untenable situations arise, this preserved knowledge is used to determine the most appropriate backtracking point. By endowing the backtracker with access to this domain-specific knowledge, a highly contextual approach to reasoning in backtracking situations can be achieved.
Les resolveurs automatiques de problemes qui utilisent des heuristiques pour guider leurs choix sont souvent pris dans des situations intenables caracteiisees par un exces de contraintes. Quand cela se produit, le programme doit ětre capable de trouver le vrai coupable parmi ses choix heuristiques de facon aeviter une recherche pouvant mener a une explosion combinatoire. Dans cet article, nous montrons comment cela peut se faire pour une classe particuliere de problemes ou les justifications associees aux points de choix font intervenir une estimation explicite du pour et du contre de chacune des possibilites par rapport a ses concurrentes. Nous avons congu un resolveur de probleme qui emmagasine une telle connaissance du pour et du contre associe aux differentes possibilityes en un point de choix durant une recherche heuristique et qui met cette connaissance a jour tout au long de la solution du probleme. Chaque fois que surgit une situation intenable, cette connaissance emmagasinee est utilisee pour determiner le point de retour arriere le plus adequat. En munissant le mecanisme de retour arriere de cette connaissance qui depend du domaine, on peut arriver a un mode de raisonnement en situation de retour arriere qui soit hautement contextuel.
TL;DR: This paper reevaluates some of the contributions of Montague grammar in view of the increasing importance of computational considerations in linguistic theory and the demand for linguistic theories that can provide support in the design of natural language systems.
Abstract: This paper reevaluates some of the contributions of Montague grammar in view of the increasing importance of computational considerations in linguistic theory and the demand for linguistic theories that can provide support in the design of natural language systems. It also considers Montague grammar in relation to work on lexical semantics and semantic nets. In this perspective the techniques of Montague grammar for systematically linking syntactic form and a model-theoretic semantics emerge as the most significant feature of the theory, while a number of the specific semantic assumptions recede in importance. Yet, with different ways of thinking about the structure mapping between levels of linguistic form and interpretations (e.g., constraint systems), we can also implement this connection using different techniques from what Montague had at his disposal.
Cette communication reevalue quelques-uns des elements de la grammaire de Montague afin de tenir compte de ľ importance grandissante des elements informatiques en theorie linguistique et de la necessite?elaborer des theories linguistiques pouvant faciliter la conception de systemes de langage naturel. Elle examine egalement la grammaire de Montague en rapport avec ľetude de la semantique lexicale et des reseaux semantiques. Dans cette perspective, les techniques de la grammaire de Montague pour etablir une relation systematique entre une forme syntaxique et une semantique de modele theorique apparaissent ětre la caracteristique la plus importante de la theorie, alors qu'un nombre d'hypotheses ? ordre semantique sont releguees au second plan. En envisageant de facons differentes le mappage de la structure entre des niveaux de forme linguistique et ? interpretation (ex. Systemes de contraintes), nous pouvons egalement etablir cette relation en utilisant des techniques autres que celles dont disposait Montague.
Mots cles: grammaire de Montague, semantique du langage naturel, comprehension du langage naturel, reseaux semantiques, representation des connaissances.
TL;DR: It is claimed that the representational approach is the only currently practical one for the design of large intelligent systems, but it is argued that the representations used should be efficient in order to account for the system's embedding in its environment.
Abstract: Situation semantics proposes novel and attractive treatments for several problem areas of natural language semantics, such as efficiency (context sensitivity) and prepositional attitude reports. Its focus on the information carried by utterances makes the approach very promising for accounting for pragmatic phenomena. However, situation semantics seems to oppose several basic assumptions underlying current approaches to natural language processing and the design of intelligent systems in general. It claims that efficiency undermines the standard notions of logical form, entailment, and proof theory, and objects to the view that mental processes necessarily involve internal representations. The paper attempts to clarify these issues and discusses the impact of situation semantics’ criticisms for natural language processing, knowledge representation, and reasoning. I claim that the representational approach is the only currently practical one for the design of large intelligent systems, but argue that the representations used should be efficient in order to account for the system's embedding in its environment. The paper concludes by stating some constraints that a computational interpretation of situation semantics should obey and discussing remaining problems.
La semantique situationnelle (situation semantics) propose des traitements originaux et seduisants de plusieurs domaines problematiques de la semantique des langues naturelles, tels que ľ efficacite (dependance du contexte) et les rapports ? attitudes propositionnelles. Son emphase sur ľ information contenue dans une enonciation rend ľ approche tres prometteuse pour rendre compte des phenomenes pragmatiques. Cependant, la semantique situationnelle semble remettre en question plusieurs suppositions sous-jacentes aux approches presentes de traitement des langues naturelles et de conception de systemes intelligents en general. Selon cette theorie, ľ efficacite des langues naturelles nous force a reconsiderer les notions traditionnelles de forme logique, de consequence logique et de theorie des preuves. La theorie s'oppose aussi aľ idee que des representations internes sont necessairement impliquees dans les processus mentaux. Ľ article tente de clarifier ces questions et discute ľ impact des critiques apportees par la semantique situationnelle sur le traitement des langues naturelles, la representation des connaissances et le raisonnement. Je soutiens que ľ approche representationelle est presentement la seule qui soit suffisamment developee pour supporter le design de systemes intelligents de grande envergure, mais egalement que les representations utilisees devraient ětre efficaces en vue de rendre compte de ľ enchassement du systeme dans son environment. Je conctus en enoncant certaines restrictions qu'une interpretation computationnelle de la semantique situationnelle devrait respecter et en discutant les problemes qui demeurent.
Mots cles: interpretation semantique, semantique situationnelle, efficacite, dependance du contexte, forme logique, representation, intentionalite, attitudes propositionnelles, inference.
TL;DR: The Coach system, a computer simulation of a human tutor, was constructed with the goal of obtaining a better understanding of how a tutor interprets the student's behavior, diagnoses difficulties, and gives advice.
Abstract: The Coach system, a computer simulation of a human tutor, was constructed with the goal of obtaining a better understanding of how a tutor interprets the student's behavior, diagnoses difficulties, and gives advice. Coach gives advice to a student who is learning a simple computer programming language. Its intelligence is based on a hierarchy of active schemas that represent the tutor's general concepts and on more specific information represented in a semantic network. The coordination of conceptually guided and data-driven processing enables the Coach system to interpret student behavior, recognize errors, and give advice to the student.
Le systeme Coach, une simulation informatique ? un tuteur humain, a ete construit dans le but ? obtenir une meilleure comprehension de la facon dont un tuteur interprete le comportement de ľetudiant, decele les difficultes, et conseille. Coach donne des conseils a un etudiant qui apprend un langage de programmation simple. Son intelligence est basee sur des schemas actifs hierarchises qui representent les concepts generaux du tuteur, et sur une information plus specifique representee dans un reseau semantique. La coordination du traitement guide par les concepts et du traitement guide par les donnees permet au systeme Coach ? interpreter le comportement de ľetudiant, de reconnaitre les erreurs, et de conseiller ľetudiant.
Mots cles: conseil, Coach, langage de programmation, EIAO (ICAI), enseignement intelligent assiste par ordinateur, schema, simulation, tuteur.
TL;DR: A formalism to represent syntactic and semantic features in logic‐based grammars and a set of tools that will be useful to grammar writers to describe features and their relations in grammar rules are introduced.
Abstract: In this paper, we introduce and initiate a formalism to represent syntactic and semantic features in logic-based grammars. We also introduce technical devices to express feature-checking and feature-inheritance mechanisms. This leads us to propose some extensions to the basic unification mechanism of PROLOG. Finally, we consider the problem of long-distance dependency relations between constituents in gapping grammars rules from the point of view of morphosyntactic features that may change depending on the position occupied by the moved constituents. What we propose is not a new linguistic theory about features, but rather a formalism and a set of tools that we think will be useful to grammar writers to describe features and their relations in grammar rules.
Dans cet article, nous introduisons et justifions un formalisme pour representer les traits syntaxiques et semantiques dans les grammaires logiques. Nous proposons, en complement, des outils pour traiter du controle et de ľ heritage des traits. Ceci nous conduit a proposer quelques extensions au mecanisme de base ? unification de PROLOG. Enrin, nous considerons le probleme des relations de dlpendances distantes entre constituants dans les regies des grammaires discontinues du point de vue des traits morpho-syntaxiques qui peuvent changer en fonction de la position occupee par les constituants deplaces. Ce que nous proposons n'est pas une nouvelle theorie linguistique, mais plutot un formalisme et un ensemble ? outils que nous pensons ětre utiles pour les concepteurs de grammaires lorsqu'ils veulent decrire des traits et leurs relations dans les regies de grammaire.
Mots cles: comprehension du langage naturel, programmation en logique, systemes de traits, grammaire discontinues, dependances lointaines.
TL;DR: During the past two decades, there have been tremendous advances in artificial intelligence, along with a recent increase in interest, support, and funding, but what effect will the advance of A1 research have on education?
Abstract: During the past two decades, there have been tremendous advances in artificial intelligence (AI), along with a recent increase in interest, support, and funding. A1 research has already had an industrial impact; for example, consider the current proliferation of expert systems and associated authoring tools. However, what effect will the advance of A1 research have on education? In particular, what effect will A1 research have in the classroom? One is tempted to answer with a promise to produce wonderful intelligent computer-assisted instruction (ICAI) systems, exciting learning environments, systems to diagnose as well as tutor and teach, and computer-based games that teach as well as captivate. But before we get caught up in a wave of enthusiasm, let us play the role of cynic for a moment. During the 60’s and ~ O ’ S , many articles promised that the computer would revolutionize education.
TL;DR: A model of problem solving flexible enough to account for a variety of problem‐solving behaviors and general enough to allow new problem domains to be defined easily is presented.
Abstract: The focus of this paper is on the underlying knowledge base for an intelligent tutorial system for high-school algebra problems. We present a model of problem solving flexible enough to account for a variety of problem-solving behaviors and general enough to allow new problem domains to be defined easily. The model is based upon the analysis of protocols between students and expert tutors. We show how student errors can be monitored and remediated using the model, and we provide an approach to understanding problem difficulty that can be used to generate challenging problems and also provides a mechanism for planning their solution.
Cet article traite de la base de connaissances implicite ? un systeme tutoriel intelligent pour les problemes ? algebre ?ecole secondaire. Nous presentons un modele de resolution de probleme suffisamment flexible pour rendre compte ? un nombre important de comportements en resolution de probleme et suffisamment general pour permettre de definir aisement de nouveaux domaines de travail. Le modele se base sur ľ analyse de protocoles entre etudiants et tuteurs experts. Nous montrons comment les erreurs ?eudiant peuvent etre controlees et comment on peut y remedier en utilisant le modele; nous poposons aussi une approche pour comprendre la difficulte? un probleme qui peut ětre utilisee pour generer des problemes stimulants ainsi que pour obtenir un mecanisme permettant de construire leur solution.
Mots cles: enseignement intelligent assiste par ordinateur, EIAO (ICAI), systeme expert, resolution de probleme, systemes tutoriels.
TL;DR: This paper introduces another candidate language for learning environments, Nial, the nested interactive array language.
Abstract: Most of the recent research on programming languages for education has been centered around the language Logo. In this paper we introduce another candidate language for learning environments, Nial, the nested interactive array language.
Nial is a general-purpose programming language based on a formal theory of mathematics called array theory. This paper introduces Nial as a language for learning programming and developing and using computer-aided instruction tools. A comparison with Logo is provided to evaluate these two languages in terms of their strengths and weaknesses as programming environments for novice programmers. We also demonstrate that a programming environment can be both simple to leam at the novice level and extendible to a powerful and sophisticated language.
L'ensemble de la recherche recente sur les langages de programmation pour ľ 'education a ete centre sur le langage Logo. Dans cet article, nous presentons un autre candidat comme langage pour apprendre les environnements, Nial, le langage de rang interactif hiearchise.
Nial est un langage de programmation a utilisations multiples base’ sur une theorie formelle des mathdmatiques appelee la theorie des rangs. Cet article presente Nial comme un langage pour apprendre la programmation et developper et utiliser des outils destruction appuyes par l'informatique. On effectue une comparaison avec Logo pour Ivaluer ces deux langages en termes de leurs forces et faiblesses comme environnements de programmation pour les programmeurs novices. Nous demontrons aussi qu'un environnement de programmation peut etre a la fois simple a apprendre pour le debutant et extensible a un langage puissant et sophistique.
Mots cles: langages de programmation, Nial, Logo, education, micromondes, logique, intelligence artificielle.
TL;DR: If any feature which lacks the property “ordinary” is discovered in the starting graph and only one example has been given, then there is feedback to the user including a request for more examples to ascertain whether the extraordinary property is a necessary part of the situation.
Abstract: The input to the described program, in learning mode, consists of examples of starting graph and result graph pairs. The starting graph is transformable into the result graph by adding or deleting certain edges and vertices. The essential common features of the starting graphs are stored together with specifications of the edges and vertices to be deleted or added. This latter information is obtained by mapping each starting graph onto the corresponding result graph. On subsequent input of similar starting graphs without a result graph, the program, in performance mode, recognizes the characterizing set of features in the starting graph and can perform the proper transformation on the starting graph to obtain the corresponding result graph. The program also adds the production to its source code so that after recompilation it is permanently endowed with the new production. If any feature which lacks the property “ordinary” is discovered in the starting graph and only one example has been given, then there is feedback to the user including a request for more examples to ascertain whether the extraordinary property is a necessary part of the situation.
En mode ? apprentissage, le programme que nous decrivons prend en entree des exemples de paires « graphe de depart -graphe ? arrivee » Le graphe ? arrivee peut s'obtenir du graphe de depart par ajout ou elimination ? un certain nombre ? arětes et de sommets. Les caracteristiques communes essentielles des graphes de depart sont mis en memoire avec une specification des arětes et sommets a ajouter ou eliminer. On obtient cette demiere information aľ aide ? une application de chaque graphe de depart sur le graphe ? arrivee correspondent. Par la suite, sur ľ entree de graphes de depart non accompagnes de leur graphe ? arrivee, le programme, en mode de travail, reconnait ľ ensemble caracteistique de traits dans le graphe de depart et sait effectuer sur ce dernier les transformations pertinentes pour produire le graphe ? arrivee correspondant. Le programme ajoute aussi cette production dans son code source de sorte qu'apres recompilation il garde en permanence la nouvelle production. Si le programme trouve dans le graphe de depart un trait qui n'a pas la propriete« ordinaire » et qui ne figure que dans un seul exemple, alors il retourne un message aľ usager et lui demande des exemples additionnels pour determiner si le trait est un constituant essentiel de la situation.
Mots cles: apprentissage, induction, transformation de graphes, acquisition de connaissances, generalisation instance-classc.