No description available.

No description available.

Markov Logic (ML) is a novel approach to Natural Language Processing tasks [Richardson and Domingos, 2006; Riedel, 2008]. It is a Statistical Relational Learning language based on First Order Logic (FOL) and Markov Networks (MN). It allows one to treat a task as structured classification. In this work, we investigate ML for the semantic processing tasks of Spoken Language Understanding (SLU) and Semantic Role Labelling (SRL). Both tasks consist of identifying a semantic representation for the meaning of a given utterance/sentence. However, they differ in nature: SLU is in the field of dialogue systems where the domain is closed and language is spoken [He and Young, 2005], while SRL is for open domains and traditionally for written text [M´arquez et al., 2008]. Robust SLU is a key component of spoken dialogue systems. This component consists of identifying the meaning of the user utterances addressed to the system. Recent statistical approaches to SLU depend on additional resources (e.g., gazetteers, grammars, syntactic treebanks) which are expensive and time-consuming to produce and maintain. On the other hand, simple datasets annotated only with slot-values are commonly used in dialogue system development, and are easy to collect, automatically annotate, and update. However, slot-values leave out some of the fine-grained long distance dependencies present in other semantic representations. In this work we investigate the development of SLU modules with minimum resources with slot-values as their semantic representation. We propose to use the ML to capture long distance dependencies which are not explicitly available in the slot-value semantic representation. We test the adequacy of the ML framework by comparing against a set of baselines using state of the art approaches to semantic processing. The results of this research have been published in Meza-Ruiz et al. [2008a,b]. Furthermore, we address the question of scalability of the ML approach for other NLP tasks involving the identification of semantic representations. In particular, we focus on SRL: the task of identifying predicates and arguments within sentences, together with their semantic roles. The semantic representation built during SRL is more complex than the slot-values used in dialogue systems, in the sense that they include the notion of predicate/argument scope. SRL is defined in the context of open domains under the premises that there are several levels of extra resources (lemmas, POS tags, constituent or dependency parses). In this work, we propose a ML model of SRL and experiment with the different architectures we can describe for the model which gives us an insight into the types of correlations that the ML model can express [Riedel and Meza-Ruiz, 2008; Meza-Ruiz and Riedel, 2009]. Additionally, we tested our minimal resources setup in a state of the art dialogue system: the TownInfo system. In this case, we were given a small dataset of gold standard semantic representations which were system dependent, and we rapidly developed a SLU module used in the functioning dialogue system. No extra resources were necessary in order to reach state of the art results.

Temporal relation classification is one of the most challenging areas of natural language processing. Advances in this area have direct relevance to improving practical applications, such as question-answering and summarization systems, as well as informing theoretical understanding of temporal meaning realization in language. With the development of annotated textual materials, this domain is now accessible to empirical machine-learning oriented approaches, where systems treat temporal relation processing as a classification problem: i.e. a decision as per which label (before, after, identity, etc) to assign to a pair (i, j) of event indices in a text. Most reported systems in this new research domain utilize classifiers that make decisions effectively in isolation, without explicitly utilizing the decisions made about other indices in a document. In this work, we present a new strategy for temporal relation classification that utilizes global models of temporal relations in a document, choosing the optimal classification for all pairs of indices in a document subject to global constraints which may be linguistically motivated. We propose and evaluate two applications of global models to temporal semantic processing: joint prediction of situation entities with temporal relations, and temporal relations prediction guided by global coherence constraints. / text

Une question centrale des recherches sur la lecture concerne la nature séquentielle ou parallèle de l’identification des mots pendant la lecture de phrases. L’hypothèse dominante postule que l’attention spatiale est allouée à un seul mot à la fois, et qu’avec cette contrainte, l’identification des mots doit forcément s’opérer de manière séquentielle. Cependant, un certain nombre de résultats suggèrent, au contraire, que l’attention spatiale peut être allouée à plusieurs mots à la fois, de manière distribuée. Cette attention disbribuée pourrait permettre l’identification en parallèle de plusieurs mots de manière simultanée, et les travaux présentés dans cette thèse cherchent à déterminer la viabilité de cette hypothèse. Notamment, nos travaux visent à préciser le niveau de traitement (visuel, orthographique, lexical, sémantique ou syntaxique) permis par cette attention distribuée. / This thesis addresses one of the most hotly debated issues in reading research: Are words processed serially or in parallel during reading? One could argue that this is primarily a question of visuo-spatial attention: is attention distributed across multiple words during reading? The research presented here suggests that attention can indeed be allocated to multiple words at once. It is further established that attention is a key factor driving (sub-lexical) orthographic processing. The next question, then, is whether multiple lexical representations can be activated in parallel. This thesis comprises a wealth of evidence for parallel lexical activation: firstly we have found that readers activate embedded words (e.g., ‘use’ in ‘houses’) alongside the word that is to be recognized, indicating that parallel lexical processing would occur even if readers could effectively focus their attention on single words. Moreover, we have found that semantic and syntactic categorization decisions about foveal target words are influenced by the semantic and syntactic aspects of surrounding words, even when all these words are presented for a duration shorter than the average time needed to recognize a single word. Hence, given that readers’ attention is spread across multiple words and that multiple lexical representations can be activated in parallel, it seems reasonable to claim that the reading system is in principle a parallel processing system.

Examining the role of individual differences, especially variations in human motivation, in vigilance tasks will result in a better understanding of sustained semantic attention and processing, which has, to date, received limited study in the literature (see Fraulini, Hancock, Neigel, Claypoole, & Szalma, 2017; Epling, Russell, & Helton, 2016; Thomson et al., 2016). This present study seeks to understand how individual differences in intrinsic motivation affect performance in a short semantic vigilance task. Performance across two conditions (lure vs. standard condition) were compared in the present study of 79 undergraduate students at the University of Central Florida. The results indicated significant main effects of intrinsic motivation on pre- and post-task stress factors, workload, and performance measures, which included correct detections, false alarms, and response time. Sensitivity and response bias, which are indices of signal detection theory, were also examined in the present study. Intrinsic motivation influenced sensitivity, but not response bias, which was affected by period on watch. The theoretical and practical implications of this research are also discussed.

In this study, the effect of semantic information processing was investigated by the oddball paradigm, by presenting consecutive Turkish words or word-like non-words while EEG signals are recorded. In an oddball paradigm, a series of events are presented of which one class is rarer than the other. Subjects are asked to respond to the infrequent stimuli (e.g. press a button, or count the number). The event related potential (ERP) component P300 obtained from EEG is considered as the marker of this attention capturing paradigm. P300 is obtained consistently for both visual and verbal stimulus. On the other hand, the ERP component N400 is consistently associated with semantic processing in neurolinguistics. Additionally, Late Positive Component (LPC) is a marker for the top-down attention mechanism during word comprehension. Moreover, there are other components, called early ERPs, which occur between 100-200 ms after the stimulus onset. These components orthographically and phonologically reflect low-level features of words. The target words chosen for our study are strictly limited to belong to a neutral category and not consist of any emotional content, to rule out emotional interference in semantic processing. Based on the ERP components that were obtained from this study, the LPC potential exhibited for words had higher amplitude than that of non-words consistently and statistically significantly. However, our study was confounded with the heterogeneity of non-words because some of the non-words were non-sense letter sequences while others were pseudowords. Due to this, although we observed the P300 and N400 ERPs consistently for all stimuli, we did not find significant differences for these potentials between words and non-words. To the best of our knowledge, our investigation is one of the few studies conducted with EEG recordings in a task that involved lexical decision making in Turkish.

Investigations of the anticipated effects of alcohol indicate that cognitive frameworks are highly correlated with drinking and other variables associated with alcohol use, explaining up to 50% of the variance in drinking outcomes (Goldman, Darkes, & Del Boca, 1999; Goldman, 2002; Goldman et al., 2006; Goldman, Reich, & Darkes, 2006). Furthermore, alcohol expectancies appear to mediate the relationship between a variety of risk factors, such as sensation seeking, and alcohol outcomes (Darkes, Greenbaum, & Goldman, 2004). The current study examined the relationship of these cognitive networks with a physiological index of expectancy violation Participants were presented with statements reflecting a wide range of alcohol outcome effects, which either violated or confirmed the participant’s own set of alcohol expectancies, while the ERPs evoked by these stimuli were recorded. As predicted, the P300 amplitude elicited by negative alcohol expectancy stimuli was positively correlated with the degree of endorsement of positive/arousing expectancies on the self-report measure. That is, the higher the individual’s positive/arousing expectancies, the larger the P300 elicited by stimuli asserting the negative effects of alcohol. There was no significant correlation, however, between P300 amplitude elicited by positive alcohol expectancy stimuli and the degree of endorsement of negative/sedating expectancies on the selfreport measure. In addition, individual differences relating to alcohol expectancies were examined as well. These results were able to identify specific stimuli that violated expectancies for each individual, as well as those that tended to violate expectancies in systematic ways across subjects. These findings provide a way forward for more precise assessment and prediction based on the well developed cognitive model of Alcohol Expectancies. In sum, variations in the amplitude of the P300 were consistent with the model of Alcohol Expectancies. Words imputing negative/sedating effects of alcohol elicited a large P300 in individuals with higher positive alcohol expectancies. By indexing the brain’s electrophysiological response sensitive to expectancy violations, these findings demonstrate concordance between verbal measures of alcohol expectancies, which by their very nature are introspective, and a psychophysiological index of expectancy thought to operate automatically and to be independent of overt responding.

No description available.

Le traitement sémantique est le processus mental par lequel nous accédons au sens. Il occupe donc une place centrale dans la compréhension et la production du langage, mais également dans le fonctionnement humain en général, puisqu'il permet de conceptualiser le monde qui nous entoure et de lui donner un sens, en le confrontant en pleine conscience aux connaissances que nous emmagasinons au fil de nos expériences. Si les bases neurales corticales du traitement sémantique sont bien documentées par de nombreuses études basées sur les données de l'imagerie fonctionnelle notamment, l'analyse de la connectivité sous-corticale impliquée dans ce traitement a jusqu'ici reçu moins d'attention. Les auteurs s'accordent néanmoins sur l'existence d'une voie ventrale sémantique, parallèle à une voie dorsale dédiée au traitement phonologique. Le présent ouvrage se propose d'apporter un nouvel éclairage à la connaissance des bases neurales du traitement sémantique du mot isolé, en lien avec le cadre plus large du traitement sémantique non-verbal, par l'étude des habiletés sémantiques de patients présentant un gliome de grade 2 OMS et pour lequel ils bénéficient d'une prise en charge chirurgicale en condition éveillée, avec cartographie cortico-sous-corticale peropératoire. Il met ainsi en évidence l'importance cruciale du faisceau fronto-occipital inférieur gauche dans cette voie ventrale sémantique, au sein d'une organisation cérébrale fonctionnelle en réseaux parallèles et distribués de zones corticales interconnectées par des faisceaux d'association de substance blanche. Il souligne également le caractère interactif du fonctionnement cognitif, ainsi que l'importance des mécanismes de contrôle dans le traitement du langage, et de la mesure de la chronométrie mentale lors son évaluation. Ces différentes considérations nous amènent à proposer un modèle hodotopique général d'organisation anatomo-fonctionnelle du langage. Les résultats présentés dans cet ouvrage peuvent donc avoir des implications cliniques et scientifiques majeures, quant à la compréhension de l'organisation cérébrale fonctionnelle du langage, de ses dysfonctionnements, des mécanismes de réorganisation fonctionnelle en cas de lésion et à l'élaboration de programmes de réhabilitation. / Semantic processing is the mental process by which we access to meaning. Therefore, it takes a central place in language comprehension and production, but also in the whole human functioning, since it allows conceptualizing and giving a meaning to the world, by confronting it consciously with the knowledge we store over our experiences. If the neural bases of semantic processing are well known at the cortical level, thanks to numerous studies based particularly on functional neuroimaging data, the analysis of the subcortical connectivity underlying this processing received so far less attention. Nevertheless, the authors agree on the existence of a semantic ventral stream, parallel to a phonological dorsal stream.The present work mean to bring a new highlight on the knowledge of the neural bases of semantic processing at the level of the single word, in connection with the wider setting of non verbal semantic processing, by the study of semantic skills in patients presenting with WHO grade 2 glioma, and for which they undergo a surgery in awaken conditions, with cortico-subcortical intraoperative mapping. Thus, this work highlights the crucial role of the inferior fronto-occipital fascicle, in this ventral semantic route, within a functional brain organization in parallel and distributed networks of cortical areas interconnected by white matter association fibers.it underlines also the interactive feature of cognitive functioning, and the significance of control mechanisms in language processing, as well as the measuring of mental chronometry when assessing it. These considerations lead us to propose a general hodotopical model of language anatomo-functional organization.The results presented in this work may thus have important clinical and scientific implications, regarding the comprehension of language brain functional organization, of its dysfunctioning, of functional reorganization mechanisms in case of brain lesion, and the elaboration of rehabilitation programs.