SEARCH · PubMed Health

Results for “Natural Language Processing”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,063 records · Page 59Linked to original sources

PubMed1998

The nature of lexical knowledge.

This paper considers the nature of lexical knowledge and its role in language and information processing. The lexicon is the central component of language and plays a pivotal role in current linguistic theory [3, 4] and, increasingly, in natural language processing systems [5-7]. The lexicon embodies information about the lexical items of the language and serves as the foundation for morphologic, syntactic, and semantic processing. The differences as well as commonalities among dictionaries, thesauri, and lexicons are discussed, and distinctions between words, lexical items, and terms are drawn. Next, the scope and content of the SPECIALIST lexicon are presented, followed by a discussion of certain writing conventions that can be troublesome for text processing applications. One approach to handling orthographic and other lexical variation is discussed in a section that reports on the design and implementation of the SPECIALIST lexical programs. The paper concludes with a discussion of controlled terminologies for the medical domain. Throughout the discussion, examples are drawn from the SPECIALIST lexicon and from the other UMLS knowledge sources [8, 9].

Humans↗

PubMedDate not supplied

A natural language user interface for fuzzy scope queries.

This paper presents a two-agent framework to build a natural language query interface for IC information system, focusing more on scope queries in a single English sentence. The first agent, parsing agent, syntactically processes and semantically interprets natural language sentence to construct a fuzzy structured query language (SQL) statement. The second agent, defuzzifying agent, defuzzifies the imprecise part of the fuzzy SQL statement into its equivalent executable precise SQL statement based on fuzzy rules. The first agent can also actively ask the user some necessary questions when it manages to disambiguate the vague retrieval requirements. The adaptive defuzzification approach employed in the defuzzifying agent is discussed in detail. A prototype interface has been implemented to demonstrate the effectiveness.

Algorithms↗

PubMed1986

Studies in artificial aphasia: experiments in processing change.

Computational neurolinguistics, an integrated approach to cognitive modelling of neural processes which may subserve natural language performance, attempts to build computational models that model behavior on two levels: at the neural process level and at the human performance level in its normal state and under pathological conditions. HOPE is one example of such a model. It demonstrates how the design and implementation of such models can provide insights into how a brain-like architecture can produce a behavior as complex as natural language. This paper will briefly describe the neurally motivated or 'natural computation' processes which produce the model's observable and verifiable behavioral results. Experiments with artificially induced aphasia on HOPE will then be described, showing that the results of simulation produce hypothesized patient profiles that are unique. These profiles illustrate the suggested contribution of the computational neurolinguistics research approach as a tool for investigating the breakdown of language performance and its potential contribution to understanding brain function.

Aphasia↗

PubMed1982

Language and the brain: implications from new computer models.

Computer systems for processing and "understanding" natural language have, up to now, used semantic and syntactic procedures that provide meanings of words in terms only of other words or some structural algorithms that ultimately relate to other words. There is a long-standing and widely held belief that words, or their unknown neuronal correlates, are fundamental to verbal meaning, concepts, and thought but over the last decade, evidence has been accumulating in neurology and psychology to indicate that this is not so. We have developed a new computer system for representing the meaning of words and sentences, by which word meanings are described non-verbally in terms of sensorimotor and other modality codes. It is suggested here that the principles of this system might usefully be applied to understanding the cerebral mechanisms of some verbal and cognitive processes and their disorders. In particular the hypothesis is proposed that verbal meaning is essentially non-verbal, and that it may be represented, for both words and sentences in terms of interassociated modal percepts or their neuronal correlates.

Brain↗

PubMedDate not supplied

Fine-grained visualization pipelines and lazy functional languages.

The pipeline model in visualization has evolved from a conceptual model of data processing into a widely used architecture for implementing visualization systems. In the process, a number of capabilities have been introduced, including streaming of data in chunks, distributed pipelines, and demand-driven processing. Visualization systems have invariably built on stateful programming technologies, and these capabilities have had to be implemented explicitly within the lower layers of a complex hierarchy of services. The good news for developers is that applications built on top of this hierarchy can access these capabilities without concern for how they are implemented. The bad news is that by freezing capabilities into low-level services expressive power and flexibility is lost. In this paper we express visualization systems in a programming language that more naturally supports this kind of processing model. Lazy functional languages support fine-grained demand-driven processing, a natural form of streaming, and pipeline-like function composition for assembling applications. The technology thus appears well suited to visualization applications. Using surface extraction algorithms as illustrative examples, and the lazy functional language Haskell, we argue the benefits of clear and concise expression combined with fine-grained, demand-driven computation. Just as visualization provides insight into data, functional abstraction provides new insight into visualization.

Journal Article↗

PubMed2001

PIES, a protein interaction extraction system.

We consider the problem of extracting, manipulating, and managing biological pathways, especially protein-protein interaction pathways. We discuss here the Protein Interaction Extraction System (PIES). PIES is contructed on top of three main technologies: Kleisli, BioNLP, and Graphviz. Kleisli is a broad-scale data integration system that we use for downloading Medline abstracts and for general manipulation and management of pathway/interaction databases. BioNLP is a natural language-based information extraction module that we use for analysing Medline abstracts and to extract precise protein-protein and other interaction information. Graphviz is a graphical layout package developed for directed graphs that we use for visualization of the extracted pathways. PIES can be augmented with various means for extracting protein interaction information from sequence databases, for example, by using Kleisli's power to integrate sequence comparison tools to detect gene fusion events in sequence databases.

Abstracting and Indexing↗

PubMed1987

Selective word-learning deficits in aphasia.

The purpose of this study was to determine whether patients with focal brain damage can learn anything about a new word, and if so, whether selected aspects of the new word are acquired depending on the nature of the patient's language processing deficit. In the context of drawing pictures with a number of felt pens identical in all respects except color, agrammatic Broca's aphasics and fluent aphasics were exposed to the new word "bice," an adjective referring to the dark green portion of the color spectrum. Our findings revealed that brain-damaged patients can engage in lexical acquisition, but Broca's aphasics and fluent aphasics apparently learn about different aspects of the new word. Specifically, on successive exposures to "bice," both Broca's aphasics and fluent aphasics exhibit progressively more accurate hypotheses for identifying a bice-colored object. During subsequent assessments, agrammatic aphasics reveal on a metalinguistic judgment task their significant difficulty appreciating the grammatical form class of "bice"; on an object classification task, fluent aphasics are significantly impaired in their classification of bice-colored objects as "bice." Taken together, these unique word-learning profiles reinforce earlier observation on the selective nature of language processing deficits after focal left-hemisphere insult, and support the claim that separate language processing devices may be selectively compromised in different groups of aphasics. The relationship between syntactic and lexical semantic processing is discussed.

Adult↗

PubMed2007

Atypical lexical/semantic processing in high-functioning autism spectrum disorders without early language delay.

Although autism is associated with impaired language functions, the nature of semantic processing in high-functioning pervasive developmental disorders (HFPDD) without a history of early language delay has been debated. In this study, we aimed to examine whether the automatic lexical/semantic aspect of language is impaired or intact in these population. Eleven individuals with Asperger's Disorder (AS) or HFPDD-Not Otherwise Specified (NOS) and age-, IQ-, and gender-matched typically developing individuals performed a semantic decision task in four conditions using an indirect priming paradigm. Semantic priming effects were found for near-semantically related word pairs in the controls, whereas this was not the case in the AS or HFPDDNOS participants. This finding suggests similarities in the underlying semantic processing of language across PDD subtypes.

Adolescent↗

PubMed1995

Experience with a mixed semantic/syntactic parser.

The value of the computerized medical record is derived in part from the availability of medical information in a coded form accessible to manipulation by processes designed for automated decision support, medical research, and computer assistance in the management of health care delivery. To meet these needs medical reports captured and stored as natural language documents must be encoded. Below we discuss an ongoing formative process aimed at developing a natural language understanding system for chest x-ray reports. Comparative data showing the progress of this process is presented.

Hospital Information Systems↗

PubMed2006

A neurolinguistic model of grammatical construction processing.

One of the functions of everyday human language is to communicate meaning. Thus, when one hears or reads the sentence, "John gave a book to Mary," some aspect of an event concerning the transfer of possession of a book from John to Mary is (hopefully) transmitted. One theoretical approach to language referred to as construction grammar emphasizes this link between sentence structure and meaning in the form of grammatical constructions. The objective of the current research is to (1) outline a functional description of grammatical construction processing based on principles of psycholinguistics, (2) develop a model of how these functions can be implemented in human neurophysiology, and then (3) demonstrate the feasibility of the resulting model in processing languages of typologically diverse natures, that is, English, French, and Japanese. In this context, particular interest will be directed toward the processing of novel compositional structure of relative phrases. The simulation results are discussed in the context of recent neurophysiological studies of language processing.

Brain↗

PubMed1998

Towards a conceptualization of language and cognitive impairment in closed-head injury: use of clinical measures.

Language deficits following closed-head injury (CHI) are widely considered as impairments secondary to the cognitive disruptions common to closed-head injury. In addition, the use of standardized measures to assess the functioning of the language system in closed-head injury has mostly been limited to aphasia test batteries which examine primary language functions only. This has resulted in generalizations as to the integrity of the language system following closed-head injury, and as a consequence, consideration of the contribution of the language system to the achievement of communicative competence in this population has been minimized. This paper presents a framework in which the functional language system is identified as a hierarchical system containing primary and higher-order language processes involved in reciprocal relationships with cognitive functions at each level. A group of 25 closed-head injury subjects and 23 demographically matched control subjects were examined for linguistic proficiency using a battery of standardized tests which investigated the language system across a hierarchy of complexity, structure and predictability. In addition, 23 of the closed-head subjects were administered a comprehensive neuropsychological test battery in order to determine the nature and extent of concomitant impairments in cognitive processes and their relationship with impairments in the language system. The language battery was found to consistently discriminate between the control and clinical groups indicating that the linguistic system is significantly impaired following CHI, with the lexical-semantic system being the most vulnerable to disruption. A strong influential relationship between language and cognitive processes was statistically confirmed and the nature of the relationship between aspects of language and cognition further delineated.

Adolescent↗

PubMed1988

Computer-assisted assessment of patient care in the hospital.

Using his natural language medical text analyzing system, the author has computer-processed the discharge summary segment of the patient record. The output of the text analyzer is a list of medical facts. The low-cost, high productivity process is eminently suited for screening the quality of clinical care provided in the hospital.

Artificial Intelligence↗

PubMed1994

Connectionist networks and language disorders.

Although neuropsychological localization constitutes the principal approach to the study of language disorders, there is reason to think it may not be entirely correct. Recent anatomical studies suggest that cognitive brain functions do not localize to precise anatomical locations. Connectionist (parallel distributed processing) approaches to the study of language, which emphasize the distributed nature of computational processes, may help explain the variability found in these anatomical studies, and provide a new way to approach the neurological study of language. This combined computational and empirical method focuses on the interplay between a computational model and the appropriate neurological, neuropsychological, and speech and language data, the whole couched in connectionist mechanisms that map naturally to what is known of the neurophysiological structure of the brain. This paper introduces the concepts of connectionist modeling, with emphasis on their use in understanding language disorders.

Brain↗

PubMedDate not supplied

On the role of perception in shaping phonological assimilation rules.

Assimilation of nasals to the place of articulation of following consonants is a common and natural process among the world's languages. Recent phonological theory attributes this naturalness to the postulated geometry of articulatory features and the notion of spreading (McCarthy, 1988). Others view assimilation as a result of perception (Ohala, 1990), or as perceptually tolerated articulatory simplification (Kohler, 1990). Kohler notes that certain consonant classes (such as nasals and stops) are more likely than other classes (such as fricatives) to undergo place assimilation to a following consonant. To explain this pattern, he proposes that assimilation tends not to occur when the members of a consonant class are relatively distinctive perceptually, such that their articulatory reduction would be particularly salient. This explanation, of course, presupposes that the stops and nasals which undergo place assimilation are less distinctive than fricatives, which tend not to assimilate. We report experimental results that confirm Kohler's perceptual assumption: In the context of a following word initial stop, fricatives were less confusable than nasals or unreleased stops. We conclude, in agreement with Ohala and Kohler, that perceptual factors are likely to shape phonological assimilation rules.

Female↗

PubMed1999

Effects of right- and left-hemisphere damage on understanding conversational implicatures.

Processing of implicatures was examined in 27 right-brain-damaged (RBD) and 31 left-brain-damaged (LBD) stroke patients with focal lesions using a new implicatures battery (IB) as part of an exploration of the neural basis and modularity of natural language pragmatics. Following Grice, we sampled implicatures of Quantity, Quality, Relation, and Manner. Verbal implicatures consisted of two-sentence conversational vignettes which are literally problematic. Nonverbal implicatures consisted mostly of famous paintings that are literally problematic (e.g., Magritte's "Le Domain d'Arnheim"). The patient has to identify and solve the problem. To compare with performance on the IB, patients also received a Hebrew adaptation of Gardner and Brownell's Right Hemisphere Communication Battery, a new test of basic speech acts (verbal and nonverbal assertions, questions, requests, and commands), a Hebrew version of the Western Aphasia Battery, and standardized neuropsychological tests. Both LBD and RBD patients were significantly impaired in implicature processing relative to age-matched normal controls. In general, both patient groups showed weak correlations of implicatures with extents of lesions in left perisylvian language area or its right-hemisphere (RH) homolog. However, performance of LBD and RBD patients on the IB revealed different patterns of correlations with other pragmatic, language, and nonlanguage tests. In LBD patients, there was a greater association between performance on verbal and nonverbal implicatures and between performance on implicatures and basic speech acts than in RBD patients. Given the different modes in which right-and left-hemisphere (LH) damage affect the processing of conversational implicatures, it remains to be discovered how the two hemispheres interact to process natural language pragmatics in the normal brain in real time.

Adult↗

PubMedDate not supplied

A general natural-language text processor for clinical radiology.

OBJECTIVE: Development of a general natural-language processor that identifies clinical information in narrative reports and maps that information into a structured representation containing clinical terms. DESIGN: The natural-language processor provides three phases of processing, all of which are driven by different knowledge sources. The first phase performs the parsing. It identifies the structure of the text through use of a grammar that defines semantic patterns and a target form. The second phase, regularization, standardizes the terms in the initial target structure via a compositional mapping of multi-word phrases. The third phase, encoding, maps the terms to a controlled vocabulary. Radiology is the test domain for the processor and the target structure is a formal model for representing clinical information in that domain. MEASUREMENTS: The impression sections of 230 radiology reports were encoded by the processor. Results of an automated query of the resultant database for the occurrences of four diseases were compared with the analysis of a panel of three physicians to determine recall and precision. RESULTS: Without training specific to the four diseases, recall and precision of the system (combined effect of the processor and query generator) were 70% and 87%. Training of the query component increased recall to 85% without changing precision.

Diagnosis, Computer-Assisted↗

PubMed1996

Whole language and deaf bilingual-bicultural education--naturally!

This position paper discusses how the tenets of Whole Language and Deaf Bilingual-Bicultural Education complement each other. It stresses that Whole Language is based on natural processes through which children can translate their constructs of personal experiences, observations, and perspectives into modes of communication that include written language and, in the present case, American Sign Language. The paper is based on two emphases: (a) Whole Language emphasizes a two-way teaching/learning process, teachers learning from children, and vice versa; and (b) Deaf Bilingual-Bicultural Education emphasizes American Sign Language as a language of instruction and builds on mutual respect for the similarities and differences in the sociocultural and socioeducational experiences and values of Deaf and hearing people. Both Whole Language and Deaf Bilingual-Bicultural Education attempt to authenticate curriculum by integrating Deaf persons' worldviews as part of educational experience.

Canada↗

PubMed2006

Organization of the state space of a simple recurrent network before and after training on recursive linguistic structures.

Recurrent neural networks are often employed in the cognitive science community to process symbol sequences that represent various natural language structures. The aim is to study possible neural mechanisms of language processing and aid in development of artificial language processing systems. We used data sets containing recursive linguistic structures and trained the Elman simple recurrent network (SRN) for the next-symbol prediction task. Concentrating on neuron activation clusters in the recurrent layer of SRN we investigate the network state space organization before and after training. Given a SRN and a training stream, we construct predictive models, called neural prediction machines, that directly employ the state space dynamics of the network. We demonstrate two important properties of representations of recursive symbol series in the SRN. First, the clusters of recurrent activations emerging before training are meaningful and correspond to Markov prediction contexts. We show that prediction states that naturally arise in the SRN initialized with small random weights approximately correspond to states of Variable Memory Length Markov Models (VLMM) based on individual symbols (i.e. words). Second, we demonstrate that during training, the SRN reorganizes its state space according to word categories and their grammatical subcategories, and the next-symbol prediction is again based on the VLMM strategy. However, after training, the prediction is based on word categories and their grammatical subcategories rather than individual words. Our conclusion holds for small depths of recursions that are comparable to human performances. The methods of SRN training and analysis of its state space introduced in this paper are of a general nature and can be used for investigation of processing of any other symbol time series by means of SRN.

Artificial Intelligence↗