Showing papers on "Semantic similarity published in 1988"

Cascade processes in picture identification

Abstract: The naming of pictures is typically thought to require sequential access to stored structural knowledge about objects, to semantic knowledge, and to a stored phonological description. Access to these different types of knowledge may constitute discrete processing stages; alternatively, it may be that information is transmitted continuously (in cascade) from one type of description to the next. The discrete stage and the cascade accounts make different predictions about the effects of structural and semantic similarity between objects on picture naming. The discrete stage account maintains that the effects of structural similarity should be confined to the process of accessing an object's structural description, and the effects of semantic similarity should be confined to the process of accessing semantic knowledge. The cascade account predicts that the effect of both variables may be passed on to subsequent processing stages. We present evidence drawn from both normal observers and from a patient...

Using latent semantic analysis to improve access to textual information

Abstract: This paper describes a new approach for dealing with the vocabulary problem in human-computer interaction. Most approaches to retrieving textual materials depend on a lexical match between words in users' requests and those in or assigned to database objects. Because of the tremendous diversity in the words people use to describe the same object, lexical matching methods are necessarily incomplete and imprecise [5]. The latent semantic indexing approach tries to overcome these problems by automatically organizing text objects into a semantic structure more appropriate for matching user requests. This is done by taking advantage of implicit higher-order structure in the association of terms with text objects. The particular technique used is singular-value decomposition, in which a large term by text-object matrix is decomposed into a set of about 50 to 150 orthogonal factors from which the original matrix can be approximated by linear combination. Terms and objects are represented by 50 to 150 dimensional vectors and matched against user queries in this “semantic” space. Initial tests find this completely automatic method widely applicable and a promising way to improve users' access to many kinds of textual materials, or to objects and services for which textual descriptions are available.

Toward a Unified Theory of Similarity and Recognition

Abstract: A new theory of similarity, rooted in the detection and recognition literatures, is developed. The general recognition theory assumes that the perceptual effect of a stimulus is random but that on any single trial it can be represented as a point in a multidimensional space. Similarity is a function of the overlap of perceptual distributions. It is shown that the general recognition theory contains Euclidean distance models of similarity as a special case but that unlike them, it is not constrained by any distance axioms. Three experiments are reported that test the empirical validity of the theory. In these experiments the general recognition theory accounts for similarity data as well as the currently popular similarity theories do, and it accounts for identification data as well as the longstanding "champion" identification model does. The concept of similarity is of fundamental importance in psychology. Not only is there a vast literature concerned directly with the interpretation of subjective similarity judgments (e.g., as in multidimensional scaling) but the concept also plays a crucial but less direct role in the modeling of many psychophysical tasks. This is particularly true in the case of pattern and form recognition. It is frequently assumed that the greater the similarity between a pair of stimuli, the more likely one will be confused with the other in a recognition task (e.g., Luce, 1963; Shepard, 1964; Tversky & Gati, 1982). Yet despite the potentially close relationship between the two, there have been only a few attempts at developing theories that unify the similarity and recognition literatures. Most attempts to link the two have used a distance-based similarity measure to predict the confusions in recognition experiments (Appelman & Mayzner, 1982; Getty, Swets, & Swets, 1980; Getty, Swets, Swets, & Green, 1979; Nakatani, 1972; Nosofsky, 1984, 1985b, 1986; Shepard, 1957, 1958b). It is now widely suspected, however, that standard distance-based similarity measures do not provide an adequate account of perceived similarity (e.g., Krumhansl, 1978; Tversky, 1977). Our approach takes the opposite tack. We begin with a very powerful and general theory of recognition and use it to derive a new similarity measure, which successfully accounts for a wide variety of similarity results in both the recognition and the similarity literatures. The theory, which we call the general recognition theory, is rooted in the detection and recognition literatures and, in fact, is a multivariate generalization of signal-detection

Components of Stroop-like interference in picture naming.

Abstract: The semantic interference effect observed in Stroop tasks and picture-word interference tasks might be due to the previous confounding of semantic similarity with task relevance (in the Stroop task) and with perceptual similarity (in the picture-word interference task). A picture-word variant of the Stroop task was devised in which the factors of task relevance and perceptual similarity were controlled. The distractor conditions allowed for the examination of four types of context effects. The results show that the overall Stroop-like interference effect can be decomposed into interference effects due to (1) a semantic relation between distractor and target, (2) the semantic relevance of the distractor word in the task at hand, (3) the presence of the distractor word in the response set, and (4) the mere presence of a word. Implications of these findings for the locus or loci of Stroop and picture-word interference effects are discussed. It is concluded that distractor words in Stroop-like naming tasks interfere mainly in the process of name retrieval.

Semantic facilitation with pictures and words.

Abstract: The present experiments explored the role of processing level and strategic factors in cross-form (word-picture and picture-word) and within-form (picture-picture and word-word) semantic facilitation. Previous studies have produced mixed results. The findings presented in this article indicate that semantic facilitation depends on the task and on the subjects' strategies. When the task required semantic processing of both picture and word targets (e.g., category verification), equivalent facilitation was obtained across all modality combinations. When the task required name processing (e.g., name verification, naming), facilitation was obtained for the picture targets. In contrast, with word targets, facilitation was obtained only when the situation emphasized semantic processing. The results are consistent with models that propose a common semantic representation for both picture and words but that also include assumptions regarding differential order of access to semantic and phonemic features for these stimulus modalities.

A Logic for Semantic Interpretation

Abstract: We propose that logic (enhanced to encode probability information) is a good way of characterizing semantic interpretation. In support of this we give a fragment of an axiomatization for word-sense disambiguation, nounphrase (and verb) reference, and case disambiguation. We describe an inference engine (Frail3) which actually takes this axiomatization and uses it to drive the semantic interpretation process. We claim three benefits from this scheme. First, the interface between semantic interpretation and pragmatics has always been problematic, since all of the above tasks in general require pragmatic inference. Now the interface is trival, since both semantic interpretation and pragmatics use the same vocabulary and inference engine. The second benefit, related to the first, is that semantic guidance of syntax is a side effect of the interpretation. The third benefit is the elegance of the semantic interpretation theory. A few simple rules capture a remarkable diversity of semantic phenomena.

Method and apparatus to identify the relation of meaning between words in text expressions

Abstract: A text comprehension and retrieval method and apparatus that uses letter-semantic analysis of the micro-syntax, or the syntax between the letters, in two words to measure how much two words are related as to their meanings or the human language concepts they present. Letter-semantic analysis involves assigning numerical values to the letters of a first word and a second word based on the dual characteristics of orientation and category inherent in the letters, and then analyzing those numerical values to identify the semantic relatedness of the letters of the first word to the letters of the second word. A letter semantic-matrix assigns weights to the meaningful letters to allow the application of letter semantic rules to convert the concepts represented by the letters of the words to numeric values. The numeric values represent the amount of relatedness of the first word to the second word and are used to retrieve text from documents having concepts related to a user supplied query expression.

An expert/expert-locating system based on automatic representation of semantic structure

Abstract: The expert/expert-locator (EEL) pairs requests for technical information with appropriate technical organizations in a large research and development company. The system automatically constructs a semantic space of organizations and terms, using a statistical matrix decomposition technique (singular value decomposition) to represent semantic similarity present in large text sources. In EEL, organizations are characterized by their documents. Using these documents as input, the analysis simultaneously fits organizations and the terms they use into the same 100-dimensional space. Similarity among organizations is determined by their overall pattern of term usage. Users' requests are processed and also fit into the high-dimensional space. The similarities between the request all organizational objects in the space are computed, and the most similar organizations are returned to the user. It is shown that this technique is superior to keyword matching. >

Semantic Priming in the Intact Brain: Separate Roles for the Right and Left Hemispheres?

Abstract: In this chapter evidence is considered about lateralized lexical processing in the intact brain obtained when the visual pathways leading to one cerebral hemisphere are selectively stimulated [the visual half-field (VHF) method — see Beaumont 1982] Such studies have yielded considerable information about hemisphere differences in word recognition processes (see Chiarello, in press, for review) In the realm of semantics, most investigations have attempted to delineate the kinds of words (ie, concrete, emotional) that the right hemisphere can recognize The research I describe here takes a different tack Our focus is on the processes available to each hemisphere for accessing word meanings and semantic relationships between words Accessing knowledge about semantic relatedness, the linking of word meanings via association or semantic similarity, is an important aspect of language comprehension If the hemispheres differ in how lexical-semantic relations are processed, this would also have; consequences for “higher order” semantic operations which depend on the availability of this lexical knowledge

An Account of Coherence, Semantic Relations, Metonymy, and Lexical Ambiguity Resolution

Abstract: Publisher Summary This chapter defines the relationship between the resolution of lexical ambiguity, semantic relations, and metonymy within an account of coherence. Coherence is the synergism of knowledge, where synergism is the interaction of two or more discrete agencies to achieve an effect of which none is individually capable. Semantic relations and metonymy are instances of coherence and coherence is also used in resolving lexical ambiguity. This account of coherence, semantic relations, metonymy and lexical ambiguity resolution is embodied in Collative Semantics (CS). The four components of CS are sense-frames, collation, semantic vectors, and screening. Sense-frames are the knowledge representation scheme and represent individual word-senses. Collation matches the sense-frames of two word-senses, finds any metonymies between the sense-frames of the word-senses, and also discriminates the semantic relations between the word-senses as a complex system of mappings between their sense-frames. Semantic vectors represent systems of mappings produced by collation and therefore the semantic relations are encoded in those mappings.

Semantic distance as a predictor of metaphor selection

Abstract: Empirical research on metaphor has focused on the interpretation and comprehension of figurative language, while ignoring the production or encoding of metaphors. This research presents a basic model that attempts to explain the encoding of metaphors in expressive communication. A basic premise of the model is that similarity in connotative meaning, measured as proximity in semantic space, leads to metaphor selection. Two experiments that tested the premise are described here. The first utilized a semantic differential based on Osgood's work, while the second utilized a specially developed instrument. Results of both experiments supported the hypothesis.

Pictorial knowledge representation using pictorial semantic networks

Abstract: Classifications of pictures and pictorial knowledge are presented. Pictorial knowledge is divided into three classes – angular pictorial knowledge, side pictorial knowledge, and angular and side pictorial knowledge. A block diagram of these three pictorial knowledge classes and a pictorial knowledge transformation module is also presented with illustrative examples. Pictorial semantic networks which in terms of pictorial nodes, property nodes, “is a” links, “has property” links, and “if and only if” links are introduced. Transitivity, generalization, specialization, inheritance hierarchy, and knowledge transformation properties are stated and illustrated by examples. Triangular, quadrangular, and polygonal knowledge representation using pictorial semantic networks are presented. The concepts of deducible property nodes are also presented with illustrative examples. Additional facts can be established from pictorial semantic networks. Thus, pictorial semantic networks are a useful way to represent pictorial knowledge in domains that use well-established taxonomies to simplify problem solving in pictorial information systems. Pictorial semantic networks offer what appears to be a fertile field for future study. The results may have useful applications in knowledge representation, expert systems, artificial intelligence, knowledge - based systems, pictorial information systems and related areas.

Effects of relation similarity on part-whole decisions.

Abstract: Subjects were presented with word pairs (e.g., bed-mattress) and timed as they decided whether one word named part of the item named by the other word. Yes responses were facilitated, and no responses were impeded, by relation similarity (i.e., the similarity of the relation between the two stimulus items to the part-whole relation). Item similarity, the similarity of the two stimulus items to each other, had no effect. The results support the view that similarity effects in semantic decision tasks are due to the comparison of elements of the relation between the two stimulus words against relation elements that serve as criteria for the target relation that the subject has been asked to identify.

Lexical semantics and preference semantics analysis

Abstract: This dissertation describes a computational system for the analysis of English prose under the Preference Semantics theory of language understanding The two main areas of investigation are these: (1) the design and implementation of a series of programs for extracting semantic information from a machine-readable dictionary, with this semantic information in a form suitable for use by a subsequent semantic analysis program, and (2) the design and implementation of a semantic analysis program that accepts short English texts and creates a corresponding representation from them The resulting representation is in a suitable form, such that other Artificial Intelligence programs can use it as a knowledge source

Semantic properties of the root-pattern array

Abstract: This paper describes a program (and a grammar) for carrying out the semantic phase of ‘root-pattern’ word formation in Hebrew. This is achieved by first automatically extracting semantic features of roots from a Hebrew thesaurus. Once the roots are reduced to feature-value sets, a grammar is used to combine a specific root with a specific grammatical pattern. The result of the grammar operation yields the final set of semantic features and values for the word. Thus, the root-pattern words are not the minimal units of grammatical investigation. They are products of the grammar. In addition to the ability to automatically generate word meanings, the semantic information derived in this way can serve also for syntactic ambiguity resolution and automatic compilation of machine-oriented dictionaries, thesauri etc. The method described here can also be used for treating affixation and thus be useful for a wide scope of languages, including English, Hebrew and Finnish.

A new efficient incremental evaluator for attribute grammars allowing conditional semantic equations

Abstract: The evaluator presented performs a depth-first search of the (static) reverse dependency graph associated with a parse tree, interleaved with the execution of semantic rules. The full compound dependency graph is not constructed. Instead, it is implicitly represented by the semantic tree and the dependency graph of the productions. The semantic rules are precompiled as programs written in intermediate code and called semantic modules. Evaluation is a call-by-need evaluation and it is optimal in the number of attribute instances evaluated. >

Semantic mediation or spurious correlation? A reply to Semin and Krahé's critique of Epstein and Teraspulsky

Abstract: Semin and Krahe draw inappropriate conclusions from their two studies. The correlation between semantic similarity and assumed relatedness may well be spurious and should not be interpreted causally. Their finding that semantic similarity estimates are not influenced by the period of observation whereas, according to Epstein and Teraspulsky, the assumed relatedness of behaviours is influenced, points to the inappropriateness of Semin and Krahe's claims. Epstein and Teraspulsky's psychometric hypothesis accounts for the available findings more easily than Semin and Krahe's semantic mediation hypothesis.

Beyond semantic ambiguity

Abstract: An advice giving system, such as an expert system gathers information from a user in order to provide advice. In this type of dialogue a single user statement or question may map into several facts of an underlying system, while several non consecutive statements may derive only one such fact. To support this type of interaction, a truly flexible natural language interface must be able to handle an extended notion of semantic ambiguity; it must avoid failure on producing partial semantic interpretations and be able to gather additional information for the interpretations from subsequent input. In this paper we describe a semantic mechanism that is able to handle this type of semantic ambiguity, while retaining other desirable properties of a general semantic interpreter.

Universal quantification in machine translation

Abstract: This approach has been developed in the context of the EUROTRA machine translation (MT) project and thus has been designed with respect to a syntax based stratificational translation process. We assume that in a semantic representation determiners are deleted and that their semantic function which is represented by semantic features is percolated into the mothernode. The semantic functions of determiners are explicated. The interaction between grammatical and lexical quantification is outlined. Ensemble theory is applied to the "count"/"mass" noun distinction. Transfer of quantification between German, English, and French is illustrated with respect to the "count"/"mass" distinction. The article closes with an outlook on the relevance of generalized quantifiers for Machine Translation.

Comparison between co-ordination and other semantic relations in Setswana

Abstract: A preliminary description of co-ordination is presented and it is pointed out that further investigation is necessary. Thereafter co-ordination as a semantic relation is compared with other types of semantic relations. It was found that co-ordination overlaps with intra-sentential conjunctive binding. With regard to a comparison between the range or scope of semantic relations in sentences and texts, it was found that co-ordination occupies an intermediate position: It does not only appear in the internal structure of sentences but also intra-sententially where it overlaps with intra- sentential conjunctive binding. Because of these properties co-ordination has a wider range of accidence than other semantic relations. Anaphoric semantic connectivity is the most mobile because it is not bound by structure.