Instances Of Concepts Research Articles

ANA: AUTOMATIC NATURAL ACQUISITION

Any system for natural language processing must be based on a lexicon. Once a model has been defined, there is the problem of acquiring and inserting words. This task is tedious for a human operator; on the one hand he must not forget any of the words, and on the other the acquisition of a new concept requires the input of a number of parameters. In view of these difficulties, research work has been undertaken in order to integrate pre-existing “paper” dictionaries. Nevertheless, these are not faultless, and are often incomplete when processing a very specialized technical field. We have therefore searched to mitigate these problems by automating the enrichment of an already partially integrated lexicon. We work in a technical field on which we have gathered different sorts of texts: written texts, specialist interviews, technical reports, etc. These documents are stored in an object oriented database, and form part of a wider project, called REX (“Retour d’EXpérience” in French, or “Feedback of Experience” in English). Our system, called ANA, reads the documents, analyses them, and deduces new knowledge, allowing the enrichment of the lexicon. The group of words already integrated into the lexicon form the “Bootstrap” of the discovery process of new words: it collects the instances of the different concepts thought to be interesting, in order to gather the semantic information. A special module makes it possible to avoid an explosion of the size of the database. It is responsible for forgetting certain instances and maintaining the database in such a way that the order in which the texts are introduced bears no influence.

Learning mutually dependent relations

So far, the task of learning relations has been concerned with the acquisition of intensional descriptions of unrelated concepts. However, in many real domains concepts are strictly related to each other and the instances of one of them cannot possibly be recognized without previous recognition of other objects as instances of related concepts. A typical case is the problem of labeling parts of a scene in order to interprete it. This paper extends in several ways the learning relations paradigm; in particular, a new methodology, allowing a recursive theory to be inferred from a set of examples, is presented. Each example may contain instances of several relations at the same time. The learning algorithm works bottom up, creating first an acyclic graph that classifies all the instances in the training set. Afterward, a recursive theory is synthesized from the graph. A prototype implementation of the algorithm has been realized by extending the program ML-SMART. The results obtained for an artificial labeling problem are described and discussed.

The polymorphous concept of money

Three experiments were carried out to investigate people's concept of money. In the first experiment, subjects were asked to rate items for their typicality as money, and subsequently to say, as quickly as possible, whether or not those items were ‘a kind of money’. Subjects were quicker to categorize the more typical instances. In the second experiment, subjects were asked to rate the similarities between pairs of instances of the money concept. The mean ratings were well described by a single-link cluster analysis whose structure was dominated by the factor of typicality. These two experiments were both carried out in England. The third experiment was conducted in both England and the Netherlands, and was preceded by a postal questionnaire aiming to identify items which were widely agreed to be instances of money. It combined the manipulations of the first two experiments, and the same patterns of categorization times and similarity ratings were found. In addition, multidimensional scaling showed that typical value is an important dimension in determining the similarities between kinds of money. There were interpretable differences in the similarity ratings between the English and Dutch samples. The three experiments show that money is a typical ‘polymorphous concept’, that is, a concept whose definition and boundaries cannot be specified precisely, but which nonetheless can be used consistently and efficiently.

Concept recognition: An approximate reasoning framework

This article presents a knowledge based methodology for recognizing concept instances in complex data such as natural scenes or speech signals. the architecture of a prototype system performing this task is also described. In order to obtain the capability of handling noisy patterns, the knowledge representation is based on continuous valued logics, and the inference engine is able to do sophisticated reasoning about the knowledge it uses in order to take into account the possible degradation of cues in the pattern. The knowledge base is subdivided into a bulk knowledge and a degradation theory. the bulk knowledge consists of production rules capturing discriminating cues in the signal as they appear in the normal cases. the degradation theory describes how the cues can be degraded owing to insertion and deletion errors. The whole classification process consists of two phases. First of all, the rules of the bulk knowledge are applied to the pattern, trying to obtain a suitable classification, and the evidence assigned to the rules is combined with a technique based on Dempster-Shafer theory. In the second phase, the classification is refined taking into account also the possible degradations in the pattern. the resulting system is then able to deal with different kinds of uncertainty and errors, such as fluctuation in the values measured for the pattern and insertion-deletion errors. the method is illustrated and evaluated on a simple example of spoken word recognition. (This work has been developed within the pilot project ESPRIT No. 26. © 1989 Wiley Periodicals, Inc.)

Motion Verb Acquisition: Development of Definitions

This experiment explored as a function of development the changes in the content of subjects' concepts of motion verbs. 60 subjects (3-yr.-olds, 4-yr.-olds, 5-yr.-olds and adults) took part in interviews in which questions designed to assess their understanding of the meanings of 10 verbs were posed. Analyses of the response-types produced, the level of generality of concept instances discussed, and the appropriateness of the word definitions supplied all indicated that children's concepts are initially quite particular and limited mainly to knowledge of the situations and entities to which the verbs can apply. Development of verb meanings appears to consist of generalization of the concepts underlying use of the words, replacing particular specific information about the referents to which the words may apply.

Interpretation of scientific or mathematical concepts: Cognitive issues and instructional implications

Scientific and mathematical concepts are significantly different from everyday concepts and are notoriously difficult to learn. It is shown that particular instances of such concepts can be identified or generated by different possible modes of concept interpretation. Some of these modes use formally explicit knowledge and thought processes; others rely on less formal case-based knowledge and more automatic recognition processes. The various modes differ in attainable precision, likely errors, and ease of use. A combination of such modes can be used to formulate an “ideal” model for interpreting scientific concepts both reliably and efficiently. Comparisons are made with the actual concept interpretations of expert scientists and novice students. The discussion elucidates some cognitive and metacognitive reasons why the learning of scientific or mathematical concepts is difficult. It also suggests instructional guidelines for teaching such concepts more effectively.

Interpretation of Scientific or Mathematical Concepts: Cognitive Issues and Instructional Implications

Scientific and mathematical concepts are significantly different from everyday concepts and are notoriously difficult to learn. It is shown that particular instances of such concepts can be identified or generated by different possible modes of concept interpretation. Some of these modes use formally explicit knowledge and thought processes; others rely on less formal case‐based knowledge and more automatic recognition processes. The various modes differ in attainable precision, likely errors, and ease of use. A combination of such modes can be used to formulate an “ideal” model for interpreting scientific concepts both reliably and efficiently. Comparisons are made with the actual concept interpretations of expert scientists and novice students. The discussion elucidates some cognitive and metacognitive reasons why the learning of scientific or mathematical concepts is difficult. It also suggests instructional guidelines for teaching such concepts more effectively.

Hemispheric asymmetry in long-term semantic relationships

Abstract Latencies in a category membership decision task were used to investigate hemispheric organisation of concepts in long-term semantic memory (LTSM). The experiment used the hemi-field technique to lateralise pictures representing high versus low typicality instances of pre-designated concepts to the right and left hemispheres. The results demonstrated hemispheric asymmetries in storage/retrieval of category members. In the right hemisphere, latencies were shorter for high than for low typicality members, consistent with previously reported prototypicality effects in free vision (Rosch, 1975), whereas for the left hemisphere, no dissociation was found. The results are interpreted in terms of two distinct conceptual organisations in the brain, one specialised in the right hemisphere and based on family resemblances/prototypicality, and one specialised in the left hemisphere and possibly based on necessary and sufficient/logical definitions. The view adopted is that typicality effects are not obligatory in all classification systems.

ETUDE LEXICOGRAPHIQUE DE “ZAI” (A NOUVEAU)

Lexical studies in Chinese provide us with an opportunity to illustrate how general linguistic operators are realized in language without conjugation. Based on a detailed description of the item ꞌzaiꞌ (again), this paper gives an example of a very special instance of the concept of reiteration.

Acquisition and generalization of novel object concepts by young language learners.

ABSTRACTTwenty-month-old children learned to recognize nonsense labels for five novel object concepts and were tested on generalization to variants of these concepts. Children were presented with either one or three examples of each object type during learning sessions. Results showed that receptive learning of names for object concepts was significantly related to a number of possible manipulations specific to each object type and to labelling by children. Children's generalization choices were consistent with adults' ranking of similarity of variants to concept prototypes. Children who learned less well were more likely to generalize to new instances of an object concept and to a greater number of variants if they had been exposed to three rather than one example during training sessions. Results also support the hypothesis that differentiation of objects in interaction is important to the formation of an object concept at this age.

Semantic analysis of inductive reasoning

Inductive learning is analysed from the semantic point of view. Processes of forming generalisations of concepts determined by examples of expert decisions are discussed. It is claimed that since concepts are relative to background knowledge, their inductive generalisations can be determined only approximately. Induction rules are defined providing a method of forming generalisations preserving positive and/or negative instances of concepts.

Acquisition of inductive biconditional reasoning skills: Training of simultaneous and sequential processing

One hundred forty-four 10th-grade students received training on one of three processing methods: coding-mapping (simultaneous), coding only, or decision tree (sequential). Then they learned a biconditional rule under one of eight transfer test conditions based on a 2 (paradigm: rule vs complete learning) × 2 (memory aids: 0 vs 4) × 2 (focus instance: presence vs absence) design. Although the coding-mapping students processed concept instances in much the same way as the coding-only students, they acquired the target rule more frequently, and they processed instances more quickly and more consistently than the decision-tree students. The observed ordinality of the responses of four truth-table classes was found to be more consistent with the simultaneity than with the sequentiality hypothesis. As expected, training interacted with paradigm and also with memory aids and focus instance. The induced simultaneous processing strategy apparently works optimally under rule learning, while the sequential strategy is difficult to induce and/or not optimal for rule-learning operations.

Eight-Month Retention of Instance and Definition Recognition for Educational Psychology Conceptsa

Thirty-two students took a retention test eight months after completing an educational psychology course. Their ability to recognize definitions of concepts was compared to their ability to recognize new instances of the same concepts. No differences were found and retention of both abilities was about 60%. Two subgroups showed different patterns of results. One subgroup appeared to use their memory for instances in recognizing definitions and new instances. The significance of the procedure of equating for content while varying the outcome measured is discussed.

Simple Feature Frequency versus Feature Validity Models of Formation of Prototypes

The typicality of a concept instance is related to the frequency with which its constituent features occur among all instances of the focal concept and possibly of other contrasting concepts. According to a simple frequency model, the prototype consists of the features that have occurred most often among instances of the focal concept. But according to a validity model, the prototype has the features that have occurred frequently among focal instances and, at the same time, infrequently among instances of contrast categories. By this account, the extent to which features overlap among focal and contrast instances is important in prototype formation. Here the feature frequencies among focal instances and the degree to which these features overlapped with contrast instances were manipulated. Twenty undergraduate college students first studied instances of the contrast category and then instances of the focal category. Next, they provided typicality ratings, frequency estimates of whole stimuli, and frequency estimates of features. All three measures supported the prediction of a simple frequency model of prototype formation.

Feature frequency and hypothesis testing in the acquisition of rule-governed concepts.

Alternative explanations of how people learn rule-governed concepts, frequency theory and hypothesis theory, were contrasted in a feature-identification task. On each trial of a problem, subjects received multiple instances of a biconditional concept. These were arranged so that some features occurred with high frequency and others with low frequency. Subjects provided a hypothesis about the pair .of relevant features and, on some trials, attempted to recall the hypothesis given on the preceding trial. At the end of each problem, subjects described the prototype, the instance that best represented all exemplars. The results were as follows: (1) High-frequency features were sampled as hypotheses more often than would be expected on the basis of random sampling, (2) prototype descriptions included more high-frequency features than would be expected by chance, and (3) previous hypotheses were accurately recalled. Whereas the last finding supports hypothesis theory, the other results confirm predictions of frequency theory. It was concluded that a satisfactory account of concept learning may need to incorporate elements of both theories.

Transfer of stimulus differentiation and perceptual selection rules.

Three experiments examined the conditions under which transfer of stimulus differentiation occurs. Two experiments involved transfer in paired associate lists having identical stimuli and re-paired response terms compared with lists having different stimuli. In one experiment, transfer was determined for lists of three-word stimuli in which various elements were shared among the stimuli. The pattern of redundancy was such that the spatial position of a discriminative element was either consistent or not consistent in different items, or there was no discriminative element. Transfer of differentiation was obtained only when position was constant. In the second experiment, stimuli were single words and were either unrelated, rhymed, or were instances of common concepts. No evidence of transfer of differentiation was obtained with either rhymes or concepts. In the third experiment, subjects learned pairs in which the stimuli were single letters; then subjects transferred to a list in which either rhyming or unrelated stimuli began with the same letters. Transfer was not differential.

SAME/DIFFERENT CONCEPT LEARNING IN THE PIGEON: THE EFFECT OF NEGATIVE INSTANCES AND PRIOR ADAPTATION TO TRANSFER STIMULI

Pigeons were trained on a matching-to-sample or oddity-from-sample task with shapes (circle and plus). Half of each group was exposed to "negative instance" trials i.e., for matching birds, neither comparison key matched the sample, and for oddity birds both comparison keys matched the sample. When all birds were transferred to a new task involving colors (red and green), nonshifted birds (transferred from matching to matching, or oddity to oddity) performed significantly better than shifted birds (transferred from matching to oddity, or oddity to matching), but only if they had experienced negative instances of the training concept. When all birds were exposed to negative instances of the transfer task and then transferred to a new color task (yellow and blue), dramatic transfer effects were observed. The effect of pre-exposure to the yellow and blue colors, in order to reduce transfer-stimulus novelty, had a minor effect on transfer.

The Effect of Multiple Discrimination Training on Pitch-Matching Behaviors of Uncertain Singers

This study investigated the effect of multiple discrimination training on vocal pitch-matching and instrumental pitch-matching behaviors, and on the aural music achievement test scores of uncertain singers. Eighty fourth- and fifth-grade students served as the subjects. Multiple discrimination training is a paradigm for concept teaching, during which care is taken to insure that the responses to a concept are controlled by the essential characteristics of the concept rather than by the irrelevant characteristics. The specific multiple discrimination training given to two of the four treatment groups consisted of presenting concept instances in the form of five pitches used as criteria in the pretest while varying the irrelevant characteristics of duration, timbre, and intensity. The procedure for shaping remained constant for all four treatment groups. Each student was reinforced for successive approximations to the correct pitch. Results of this study show that multiple discrimination training, added to successive approximation techniques, is appropriate as a standard part of training for uncertain singers.

Retroactive inhibition as a function of the conceptual structure of original and interpolated lists

Subjects learned either a hierarchically structured serial list of concept instances or a list made up of the same words but without structure. These two original lists were followed by one of four interpolated lists varying in number of hierarchical levels and consisting of new instances of the concepts which occurred in the original lists. Recall of the original structured list showed heavy retroactive inhibition, with the magnitude being independent of the number of conceptual levels in the interpolated lists. For the unstructured original list, retroactive inhibition decreased as the number of conceptual levels in the interpolated list increased. Some mechanisms believed involved in producing these outcomes were discussed.

Effects of a definition and a varying number of examples and nonexamples on concept attainment.

Fourth grade children read experimental lessons, each of which presented one of the following: (1) a concept definition and placebo material; (2) a rational set of examples and nonexamples of the concept and placebo material; (3) the definition, a rational set, and placebo material; or (4) the definition and three different rational sets. Control children read the placebo material only. Each experimental group performed significantly better than the controls; and children reading a lesson with a definition and three rational sets performed significantly better than those who received only a definition. The use of rational sets of concept instances and of a concept definition is validated as a powerful controllable variable in instructional material. (Author) iffiklg N'i.:\`';'14 474:4*WT*44:;4 '44,4ve41,1r,