Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
This paper presents a novel technique for empty category (EC) detection using distributed word representations. A joint model is learned from the labeled data to map both the distributed representations of the contexts of ECs and EC types to a low dimensional space. In the testing phase, the context of possible EC positions will be projected into the same space for empty category detection. Experiments on Chinese Treebank prove the effectiveness of the proposed method. We improve the precision by about 6 points on a subset of Chinese Treebank, which is a new state-ofthe-art performance on CTB.
News websites give their users the opportunity to participate in discussions about published articles, by writing comments. Typically, these comments are unstructured making it hard to understand the flow of user discussions. Thus, there is a need for organizing comments to help users to (1) gain more insights about news topics, and (2) have an easy access to comments that trigger their interests. In this work, we address the above problem by organizing comments around the entities and the aspects they discuss. More specifically, we propose an approach for entity and aspect extraction from user comments through the following contributions. First, we extend traditional Named-Entity Recognition approaches, using coreference resolution and external knowledge bases, to detect more occurrences of entities in comments. Second, we exploit part-of-speech tag, dependency tag, and lexical databases to extract explicit and implicit aspects around discussed entities. Third, we evaluate our entity and aspect extraction approach, on manually annotated data, showing that it highly increases precision and recall compared to baseline approaches.
We propose and evaluate the use of an affective-semantic model to expand the affective lexica of German, Greek, English, Spanish and Portuguese. Motivated by the assumption that semantic similarity implies affective similarity, we use word level semantic similarity scores as semantic features to estimate their corresponding affective scores. Various context-based semantic similarity metrics are investigated using contextual features that include both words and character n-grams. The model produces continuous affective ratings in three dimensions (valence, arousal and dominance) for all five languages, achieving consistent performance. We achieve classification accuracy (valence polarity task) between 85% and 91% for all five languages. For morphologically rich languages the proposed use of character n-grams is shown to improve performance.
Songs heard between the ages of 15 and 24 should be remembered better and have a stronger relationship to autobiographical memories when compared with music from other phases of life (“reminiscence bump effect”). Additionally, the proportion of music-evoked autobiographical memories (MEAMs) is at a maximum in these years of early adolescence and then declines up to the age of 60. In our study we tried both to replicate these important findings based on a German sample and to further investigate the influence of the affective characteristics of the songs on the frequency of participants’ autobiographical memories. In Experiment 1 a group of adults ( N = 48, M age = 67.1 years) listened to excerpts from 80, number-one, popular music hits from 1930 to 2010 and gave written self-reports on MEAMs. In Experiment 2 the affective characteristics were rated by another group of adults ( N = 22, M age = 66 years) and were used to predict the frequency of MEAMs. As a main result of Experiment 1, we confirmed the reminiscence bump and decline effect with a small effect size for the ratings of feelings evoked by the song and with a medium effect size for the song recognition performance of those songs released during the participants’ age range of 15 to 24 years. The total number of MEAMs was only marginally influenced by a memory bump and decline effect, and participants showed a significant proportion of MEAMs up to the fifth decade. Experiment 2 revealed that the affective ratings of the songs were unequally distributed over the two-dimensional emotion space unlike the average rate of MEAMs which was nearly equally distributed. In contrast to previous research, we therefore conclude that popular songs can be associated with autobiographical memory over five decades of life – independent of the affective character of the music.
The purpose of this paper is to present an approach to create semi-automatically ontology from Arabic texts. The whole process is supervised by a linguistic expert. Our involvement in this project focused on a lexical ontology, taking as model the WordNet ontology, and as input source, the “Arabic verbs” of a contemporary monolingual dictionary () /mζjm Alγny/ in the form a lexical database. The verb, pivot of a sentence, is our goal in creating concepts, by adopting the synset as our meaning representation model. The Markov clustering algorithm of a graph, generated by the defining verbs, obtained from the transitive closure, allowed us to detect similar verbs and to identify as well, for a given verbal entry, all of its synonyms. A tool has been implemented, and experiments have been carried out to evaluate and show efficiency of the proposed approach.
We introduce interpolation of trained MSTParser models as a resource combination method for multi-source delexicalized parser transfer. We present both an unweighted method, as well as a variant in which each source model is weighted by the similarity of the source language to the target language. Evaluation on the HamleDT treebank collection shows that the weighted model interpolation performs comparably to weighted parse tree combination method, while being computationally much less demanding.
We examined the potential cost of practicing suppression of negative thoughts on subsequent performance in an unrelated task. Cues for previously suppressed and unsuppressed (baseline) responses in a think/no-think procedure were displayed as irrelevant flankers for neutral words to be judged for emotional valence. These critical flankers were homographs with one negative meaning denoted by their paired response during learning. Responses to the targets were delayed when suppression cues (compared with baseline cues and new negative homographs) were used as flankers, but only following direct-suppression instructions and not when benign substitutes had been provided to aid suppression. On a final recall test, suppression-induced forgetting following direct suppression and the flanker task was positively correlated with the flanker effect. Experiment 2 replicated these findings. Finally, valence ratings of neutral targets were influenced by the valence of the flankers but not by the prior role of the negative flankers.
With a dependency grammar, this study provides a unified method for calculating the syntactic complexity in linear and hierarchical dimensions. Two metrics, mean dependency distance (MDD) and mean hierarchical distance (MHD), one for each dimension, are adopted. Some results from the Czech-English dependency treebank are revealed: (1) Positive asymmetries in the distributions of the two metrics are observed in English and Czech, which indicates both languages prefer the minimalization of structural complexity in each dimension. (2) There are significantly positive correlations between sentence length (SL), MDD, and MHD. For longer sentences, English prefers to increase the MDD, while Czech tends to enhance the MHD. (3) A trade-off relationship of syntactic complexity in two dimensions is shown between the two languages. English tends to reduce the complexity of production in the hierarchical dimension, whereas Czech prefers to lessen the processing load in the linear dimension. (4) The threshold of the MDD2 and MHD2 in English and
Abstract This work presents the development and evaluation of an extended Urdu parser. It further focuses on issues related to this parser and describes the changes made in the Earley algorithm to get accurate and relevant results from the Urdu parser. The parser makes use of a morphologically rich context free grammar extracted from a linguistically-rich Urdu treebank. This grammar with sufficient encoded information is comparable with the state-of-the-art parsing requirements for the morphologically rich Urdu language. The extended parsing model and the linguistically rich extracted-grammar both provide us better evaluation results in Urdu/Hindi parsing domain. The parser gives 87% of f-score, which outperforms the existing parsing work of Urdu/Hindi based on the tree-banking approach.
This paper proposes neural networks for integrating compositional and non-compositional sentiment in the process of sentiment composition, a type of semantic composition that optimizes a sentiment objective. We enable individual composition operations in a recursive process to possess the capability of choosing and merging information from these two types of sources. We propose our models in neural network frameworks with structures, in which the merging parameters can be learned in a principled way to optimize a well-defined objective. We conduct experiments on the Stanford Sentiment Treebank and show that the proposed models achieve better results over the model that lacks this ability.
Text-based sentiment analysis is a growing research field in affective computing, driven by both commercial applications and academic interest. Continuous dimensional representations, such as valence-arousal (VA) space, can represent the affective state more precisely than discrete effective representations. In building dimensional sentiment applications, affective lexicons with valence-arousal ratings are useful resources but are still very rare. Therefore, recent studies have investigated the automatic development of VA lexicons using linear regression techniques. One of the major limitations of linear regression is the under-fitting problem which can cause a poor fit between the algorithm and the training data. To tackle this problem, this study proposes the use of a locally weighted linear regression (LWLR) model to predict the valence-arousal ratings of affective words. The locally weighted method performs a regression around the point of interest using only training data that are "local" to that point, and thus can reduce the impact of noise from unrelated training data. Experimental results show that the proposed method achieved better performance for VA word prediction.
The goal of this work is to bring semantics into the tasks of text recognition and retrieval in natural images. Although text recognition and retrieval have received a lot of attention in recent years, previous works have focused on recognizing or retrieving exactly the same word used as a query, without taking the semantics into consideration. In this paper, we ask the following question: can we predict semantic concepts directly from a word image, without explicitly trying to transcribe the word image or its characters at any point? For this goal we propose a convolutional neural network (CNN) with a weighted ranking loss objective that ensures that the concepts relevant to the query image are ranked ahead of those that are not relevant. This can also be interpreted as learning a Euclidean space where word images and concepts are jointly embedded. This model is learned in an end-to-end manner, from image pixels to semantic concepts, using a dataset of synthetically generated word images and concepts mined from a lexical database (WordNet). Our results show that, despite the complexity of the task, word images and concepts can indeed be associated with a high degree of accuracy.
This paper describes the submitted discourse parsing system of the natural language group of Soochow University (SoNLP-DP) to the CoNLL 2015 shared task. Our System classifies discourse relations into explicit and non-explicit relations and uses a pipeline platform to conduct every subtask to form an end-toend shallow discourse parser in the Penn Discourse Treebank (PDTB). Our system is evaluated on the CoNLL-2015 Shared Task closed track and achieves the 18.51% in F1-measure on the official blind test set.
We show that Combinatory Categorial Grammar (CCG) supertags can improve Telugu dependency parsing. In this process, we first extract a CCG lexicon from the dependency treebank. Using both the CCG lexicon and the dependency treebank, we create a CCG treebank using a chart parser. Exploring different morphological features of Telugu, we develop a supertagger using maximum entropy models. We provide CCG supertags as features to the Telugu dependency parser (MST parser). We get an improvement of 1.8% in the unlabelled attachment score and 2.2% in the labelled attachment score. Our results show that CCG supertags improve the MST parser, especially on verbal arguments for which it has weak rates of recovery.
Abstract: We present the multiplicative recurrent neural network as a general model for compositional meaning in language, and evaluate it on the task of fine-grained sentiment analysis. We establish a connection to the previously investigated matrix-space models for compositionality, and show they are special cases of the multiplicative recurrent net. Our experiments show that these models perform comparably or better than Elman-type additive recurrent neural networks and outperform matrix-space models on a standard fine-grained sentiment analysis corpus. Furthermore, they yield comparable results to structural deep models on the recently published Stanford Sentiment Treebank without the need for generating parse trees.
We introduce an approach to train lexicalized parsers using bilingual corpora\nobtained by merging harmonized treebanks of different languages, producing\nparsers that can analyze sentences in either of the learned languages, or even\nsentences that mix both. We test the approach on the Universal Dependency\nTreebanks, training with MaltParser and MaltOptimizer. The results show that\nthese bilingual parsers are more than competitive, as most combinations not\nonly preserve accuracy, but some even achieve significant improvements over the\ncorresponding monolingual parsers. Preliminary experiments also show the\napproach to be promising on texts with code-switching and when more languages\nare added.\n
Context: Sentences are rich in redundancy, and therefore, their identification is often facilitated by the context. The use of phrases introduces limited contextual cues into the process of identification and facilitates the evocation of words. Thus, there is a need to develop phrase recognition test to assess identification abilities. Aims: To develop and validate phrase recognition test in Kannada language for assessing speech recognition in noise. Settings and Design: Normative research design was utilized. Subjects and Methods: A total of 70 phrases in Kannada language were constructed and 67 of them were selected based on familiarity rating. Ten participants each in two groups were involved for the list equivalency and validation. Statistical Analysis Used: Repeated measure of analysis of variance was utilized for the lists equivalency and standardization. Results: Sixty-seven phrases were shortlisted from 70 phrases through familiarity rating. These phrases were embedded in different 5 signal to noise ratios (SNRs) (−9 dB SNR to −1 dB SNR in steps of 2 dB). Analysis of results showed 50% recognition score at ~−5 dB SNR. In addition, the phrases that were too easy and too difficult were eliminated. From the remaining phrases, five lists of 10 phrases each were constructed and compared for their equal intelligibility in noise. The results revealed no significant differences across the phrase lists. Conclusions: The homogenous five lists of the Kannada phrase recognition test will be useful to assess identification ability of the listeners and hearing aid benefit.
The main subject of this article is the linguistic norm and its (in)variants with regard to evaluation of lingual expressions. The article tries to show various German and Polish conceptions of linguistic norms and to point to their reference to lingual errors/mistakes and interference. Functional norm, usus, feel for language and their importance for language standardization are also discussed in the article, together with the relativity of linguistic norm.
Dutch is a pluricentric language: in Europe, it is spoken in two different countries (the Netherlands in the north and Flanders, Belgium, in the south) with differing linguistic norms. Vismans investigates what happens when the northern and southern Dutch address systems meet. His data come from in-depth radio interviews between Dutch journalists and Flemish academics. In a qualitative analysis, he tracks the development of the relationship between the two speakers and their use of address forms, as well as other markers of (in)formality. The analysis also takes into account other possible factors affecting the interaction (age, gender, residence in the other country) and pays special attention to speakers’ commentary on the variation between familiar and formal second-person pronouns.
Eating slowly is associated with a lower body mass index. However, the underlying mechanism is poorly understood. Here, our objective was to determine whether eating a meal at a slow rate improves episodic memory for the meal and promotes satiety. Participants (N=40) consumed a 400ml portion of tomato soup at either a fast (1.97ml/s) or a slow (0.50ml/s) rate. Appetite ratings were elicited at baseline and at the end of the meal (satiation). Satiety was assessed using; i) an ad libitum biscuit 'taste test' (3h after the meal) and ii) appetite ratings (collected 2h after the meal and after the ad libitum snack). Finally, to evaluate episodic memory for the meal, participants self-served the volume of soup that they believed they had consumed earlier (portion size memory) and completed a rating of memory 'vividness'. Participants who consumed the soup slowly reported a greater increase in fullness, both at the end of the meal and during the inter-meal interval. However, we found little effect of eating rate on subsequent ad libitum snack intake. Importantly, after 3h, participants who ate the soup slowly remembered eating a larger portion. These findings show that eating slowly promotes self-reported satiation and satiety. For the first time, they also suggest that eating rate influences portion size memory. However, eating slowly did not affect ratings of memory vividness and we found little evidence for a relationship between episodic memory and satiety. Therefore, we are unable to conclude that episodic memory mediates effects of eating rate on satiety.
Studies demonstrating a mnemonic benefit for encoding words in a survival scenario have revived interest in how human memory is shaped by evolutionary pressures. Prior work on the survival-processing advantage has largely examined cognitive factors as potential proximate mechanisms. The current study, by contrast, focused on the role of perceived threat. Guided by the idea that a survival scenario implies threat, we combined measures of heart rate (HR) with affective ratings to probe the potential presence of fear bradycardia as a marker of freezing--a parasympathetically dominated HR deceleration that reflects the initial stage of the defensive engagement. We replicated the mnemonic advantage in behavior and found that the survival scenario was rated higher in perceived negative arousal than a commonly used control scenario. Critically, words encountered in the survival scenario were associated with more extensive HR deceleration, and this effect was directly related to subsequent recall performance. Our findings point to a role for the involvement of neurobiological fear responses in producing the survival processing advantage, as well as potential links between autonomic changes and cognitive processing in adaptive memory.
Various websites are dedicated to rating physicians. The goals of this study were to: (1) evaluate the prevalence of orthopedic surgeon ratings on physician rating websites in the United States and (2) evaluate factors that may affect ratings, such as sex, practice sector (academic or private), years of practice, and geographic location. A total of 557 orthopedic surgeons selected from the 30 most populated US cities were enrolled. The study period was June 1 to July 31, 2013. Practice type (academic vs private), sex, geographic location, and years since completion of training were evaluated. For each orthopedic surgeon, numeric ratings from 7 physician rating websites were collected. The ratings were standardized on a scale of 0 to 100. Written reviews were also collected and categorized as positive or negative. Of the 557 orthopedic surgeons, 525 (94.3%) were rated at least once on 1 of the physician rating websites. The average rating was 71.4. The study included 39 female physicians (7.4%) and 486 male physicians (92.6%). There were 204 (38.9%) physicians in academic practice and 321 (61.1%) in private practice. The greatest number of physicians, 281 (50.4%), practiced in the South and Southeast, whereas 276 (49.6%) practiced in the West, Midwest, and Northeast. Those in academic practice had significantly higher ratings (74.4 vs 71.1; P<.007). No significant difference based on sex (72.5 male physicians vs 70.2 female physicians; P=.17) or geographic location (P=.11) were noted. Most comments (64.6%) were positive or extremely positive. Physicians who were in practice for 6 to 10 years had significantly higher ratings (76.9, P<.01) than those in practice for 0 to 5 years (70.5) or for 21 or more years (70.7).
Because of their superior ability to preserve sequence information over time,\nLong Short-Term Memory (LSTM) networks, a type of recurrent neural network with\na more complex computational unit, have obtained strong results on a variety of\nsequence modeling tasks. The only underlying LSTM structure that has been\nexplored so far is a linear chain. However, natural language exhibits syntactic\nproperties that would naturally combine words to phrases. We introduce the\nTree-LSTM, a generalization of LSTMs to tree-structured network topologies.\nTree-LSTMs outperform all existing systems and strong LSTM baselines on two\ntasks: predicting the semantic relatedness of two sentences (SemEval 2014, Task\n1) and sentiment classification (Stanford Sentiment Treebank).\n
Recursive neural network models have achieved promising results in many natural language processing tasks. The main difference among these models lies in the composition function, i.e., how to obtain the vector representation for a phrase or sentence using the representations of words it contains. This paper introduces a novel Adaptive Multi-Compositionality (AdaMC) layer to recursive neural network models. The basic idea is to use more than one composition function and adaptively select them depending on input vectors. We develop a general framework to model the semantic composition as a distribution of these composition functions. The composition functions and parameters used for adaptive selection are jointly learnt from the supervision of specific tasks. We integrate AdaMC into existing recursive neural network models and conduct extensive experiments on the Stanford Sentiment Treebank and semantic relation classification task. The experimental results demonstrate that AdaMC improves the performance of recursive neural network models and outperforms the baseline methods.
The aim of this study was to investigate if and how temporal context influences subjective affective responses to emotional images. To do so, we examined whether the subjective evaluation of a target image is influenced by the valence of its preceding image, and/or its overall position in a sequence of images. Furthermore, we assessed if these potentially confounding contextual effects can be moderated by a common procedural control: randomized stimulus presentation. Four groups of participants evaluated the same set of 120 pictures from the International Affective System (IAPS) presented in four different sequences. Our data reveal strong effects of both aspects of temporal context in all presentation sequences, modified only slightly in their nature and magnitude. Furthermore, this was true for both valence and arousal ratings. Subjective ratings of negative target images were influenced by temporal context most strongly across all sequences. We also observed important gender differences: females expressed greater sensitivity to temporal-context effects and design manipulations relative to males, especially for negative images. Our results have important implications for future emotion research that employs normative picture stimuli, and contributes to our understanding of context effects in general.
OBJECTIVES: Risky behaviour seriously impacts the life of adult patients with attention deficit hyperactivity disorder (ADHD). Such behaviours have often been attributed to their exaggerated reward seeking, but dysfunctional anticipation of negative outcomes might also play a role. METHODS: The present study compared adult patients with ADHD (n = 28) with matched healthy controls (n = 28) during anticipation of monetary losses versus gains while undergoing functional magnetic resonance imaging (fMRI) and skin conductance recording. RESULTS: Skin conductance was higher during anticipation of losses compared to gains in both groups. Affective ratings of predictive cues did not differ between groups. ADHD patients showed increased activity in bilateral amygdalae, left anterior insula (region of interest analysis) and left temporal pole (whole brain analysis) compared to healthy controls during loss versus gain anticipation. In the ADHD group higher insula and temporal pole activations went along with more negative affective ratings. CONCLUSIONS: Neural correlates of loss anticipation are not blunted but rather increased in ADHD, possibly due to a life history of repeated failures and the respective environmental sanctions. Behavioural adaptations to such losses, however, might differentiate them from controls: future research should study whether negative affect might drive more risk seeking than risk avoidance.
This paper focuses on a relatively new and much discussed phenomenon on Flemish television: The practice of intralingual subtitling of Dutch, i.e. Dutch subtitling of native speakers of (varieties of) Dutch. Our study investigates the linguistic determinants of intralingual subtitling and subsequently confronts actual subtitling practice with viewer needs. The analyses reveal a striking inconsistency between intralingual subtitling practice in fiction versus non-fiction programs. This appears to be symptomatic of a tension between the official language policy in Flanders and present day linguistic reality. As such, subtitling practice subtly reflects the existence of shifting linguistic norms in Flanders.
This paper describes a parsing model that combines the exact dynamic programming of CRF parsing with the rich nonlinear featurization of neural net approaches. Our model is structurally a CRF that factors over anchored rule productions, but instead of linear potential functions based on sparse features, we use nonlinear potentials computed via a feedforward neural network. Because potentials are still local to anchored rules, structured inference (CKY) is unchanged from the sparse case. Computing gradients during learning involves backpropagating an error signal formed from standard CRF sufficient statistics (expected rule counts). Using only dense features, our neural CRF already exceeds a strong baseline CRF model (Hall et al., 2014). In combination with sparse features, our system achieves 91.1 F1 on section 23 of the Penn Treebank, and more generally outperforms the best prior single parser results on a range of languages.
Penn Discourse Treebank style discourse parsing is a composite task of identifying discourse relations (explicit or nonexplicit), their connective and argument spans, and assigning a sense to these relations from the hierarchy of senses. In this paper we describe University of Trento parser submitted to CoNLL 2015 Shared Task on Shallow Discourse Parsing. The span detection tasks for explicit relations are cast as token-level sequence labeling. The argument span decisions are conditioned on relations' being intra-or intersentential. Non-explicit relation detection and sense assignment tasks are cast as classification. In the end-to-end closedtrack evaluation, the parser ranked second with a global F-measure of 0.2184
In the past seven years, Language Research Institute of Inner Mongolia University has constructed a 500,000-word scale Mongolian dependency treebank. The syntactic treebank provides a favorable data platform for language research and information processing. In order to effectively use the treebank, we have designed and implemented a graphical syntactic information retrieval system based on the Mongolian dependency treebank. As an application system, this retrieval system offers search and statistical analysis on word, phrase, syntactic fragment and syntactic structure level.
About half of the discourse relations annotated in Penn Discourse Treebank (Prasad et al., 2008) are not explicitly marked using a discourse connective. But we do not have extensive theories of when or why a discourse relation is marked explicitly or when the connective is omitted. Asr and Demberg (2012a) have suggested an information-theoretic perspective according to which discourse connectives are more likely to be omitted when they are marking a relation that is expected or predictable. This account is based on the Uniform Information Density theory (Levy and Jaeger, 2007), which suggests that speakers choose among alternative formulations that are allowed in their language the ones that achieve a roughly uniform rate of information transmission. Optional discourse markers should thus be omitted if they would lead to a trough in information density, and be inserted in order to avoid peaks in information density. We here test this hypothesis by observing how far a specific cue, negation in any form, affects the discourse relations that can be predicted to hold in a text, and how the presence of this cue in turn affects the use of explicit discourse connectives.
We here present the development and validation of the Verbal Affective Memory Test-24 (VAMT-24). First, we ensured face validity by selecting 24 words reliably perceived as positive, negative or neutral, respectively, according to healthy Danish adults' valence ratings of 210 common and non-taboo words. Second, we studied the test's psychometric properties in healthy adults. Finally, we investigated whether individuals diagnosed with Seasonal Affective Disorder (SAD) differed from healthy controls on seasonal changes in affective recall. Recall rates were internally consistent and reliable and converged satisfactorily with established non-affective verbal tests. Immediate recall (IMR) for positive words exceeded IMR for negative words in the healthy sample. Relatedly, individuals with SAD showed a significantly larger decrease in positive recall from summer to winter than healthy controls. Furthermore, larger seasonal decreases in positive recall significantly predicted larger increases in depressive symptoms. Retest reliability was satisfactory, rs ≥.77. In conclusion, VAMT-24 is more thoroughly developed and validated than existing verbal affective memory tests and showed satisfactory psychometric properties. VAMT-24 seems especially sensitive to measuring positive verbal recall bias, perhaps due to the application of common, non-taboo words. Based on the psychometric and clinical results, we recommend VAMT-24 for international translations and studies of affective memory.
OBJECTIVE: To examine the impact of plain packaging of cigarettes with enhanced graphic health warnings on adolescents' perceptions of pack image and perceived brand differences. METHODS: Cross-sectional school-based surveys conducted in 2011 (prior to introduction of new cigarette packaging) and in 2013 (7-12 months afterwards). Students aged 12-17 years (2011 n=6338; 2013 n=5915) indicated whether they had seen a cigarette pack in previous 6 months. Students rated the character of four popular cigarette brands, indicated level of agreement regarding differences between brands in ease of smoking, quitting, addictiveness, harmfulness and look of pack; and indicated positive and negative perceptions of pack image. Changes in responses of students seeing cigarette packs in the previous 6 months (2011: 60%; 2013: 65%) were examined. RESULTS: Positive character ratings for each brand reduced significantly between 2011 and 2013. Changes were found for four of five statements reflecting brand differences. Significantly fewer students in 2013 than 2011 agreed that 'some brands have better looking packs than others' (2011: 43%; 2013: 25%, p<0.001), with larger decreases found among smokers (interaction p<0.001). Packs were rated less positively and more negatively in 2013 than in 2011 (p<0.001). The decrease in positive image ratings was greater among smokers. CONCLUSIONS: The introduction of standardised packaging has reduced the appeal of cigarette packs. Further research could determine if continued exposure to standardised packs creates more uncertainty or disagreement regarding brand differences in ease of smoking and quitting, perceived addictiveness and harms.
Subjective self-assessment is a frequently used method when it comes to Quality of Experience assessment in general. If correctly used a variety of scales can be applied to assess the quality of cloud gaming for different constructs such as experienced quality or flow. Besides self-assessment, physiological correlates are a promising method to measure the influence of cloud gaming quality on the player without the interruption introduced by the rating task. We present subjective and physiological results and lessons learned from a laboratory study in which 32 participants played a first person shooter game in a cloud gaming setup with varying levels of video quality caused by different video compression bitrates. We found that the video quality influenced the perceived quality, player experience, the valence rating, and the alpha frequency band power. It is shown that 1) subjective methods assess the quality variation and 2) physiological measures capture the influence on the player in terms of a reduced cognitive state. Taken together, test set-ups will benefit from a mixed methods approach in cloud gaming QoE testing.
This chapter uses data extracted from two recently compiled historical corpora of Australian English, the Corpus of Oz Early English (19th century) and AusCorp (20th century) to examine developments in ten morphosyntactic variables over the past two centuries. At the same time comparisons are drawn with British and American English, using data from ARCHER (A Representative Corpus of Historical English Registers). While Australian usage is found to have diverged from that of its British colonial parent, reflecting increasing independence from British linguistic norms, it has shifted towards that of American English – the new centre of gravity of grammatical change in English world-wide – which emerges as the most advanced variety on eight of the ten variables. Keywords: grammar; Australian English; diachronic; corpus; variation
Abstract This paper surveys methods for annotation error detection and correction. Methods can broadly be characterized as to whether they detect inconsistencies with respect to some statistical model based only on the corpus data or whether they detect inconsistencies with respect to a grammatical model, in general, some external information source. Two extended examples are presented, illustrating these different techniques: (1) the variation n ‐gram method, which searches for inconsistences in annotation for identical strings; and (2) a method of ad hoc rule detection, for syntactic annotation, which compares treebank rules to a grammar to determine which are anomalous. Methods for detecting annotation errors have developed much over the last decade, and thus corpus practitioners can benefit greatly from them, while at the same time NLP researchers can learn more about the nuances of the annotation they use and see how error correction methods intersect with NLP techniques.
BACKGROUND: Migraine is a disorder of periodic disabling headache. Facilitated cortical responsivity has been suggested as one predisposing factor. Although the underlying mechanisms of migraine attack onsets are not fully understood, facilitated cortical responsivity has been suggested as one predisposing factor. Here, we investigate if enhanced cortical responsivity is reflected in altered event-related potentials during processing of complex pictures. METHOD: Altogether, 16 migraine patients and 16 healthy volunteers participated in this study. Each patient had a diagnosed migraine and was headache- and medication-free for the study. Participants watched positive, negative and neutral pictures from the international affective picture system. An electroencephalogram was recorded during picture presentation. Afterwards, participants were asked to rate the pictures for valence and arousal. RESULTS: Migraine patients showed significantly more negative-going early event-related potential components from 100 ms to 180 ms to all picture categories over occipital regions as well as more positive-going late potentials over central regions. Patients and controls did not differ in valence and arousal ratings for the international-affective picture system. DISCUSSION: Patients with migraine seem to react cortically more intensely to all kinds of pictorial stimuli, regardless of emotional content. This facilitated processing may be related to the high cortical responsivity shown in various other event-related potential studies and might contribute to the recurring intense headache attacks.
This paper focuses on improving a specific opinion spam detection task, deceptive spam. In addition to traditional word form and other shallow syntactic features, we introduce two types of deep level linguistic features. The first type of features are derived from a shallow discourse parser trained on Penn Discourse Treebank (PDTB), which can capture inter-sentence information. The second type is based on the relationship between sentiment analysis and spam detection. The experimental results over the benchmark dataset demonstrate that both of the proposed deep features achieve improved performance over the baseline.
The task of recognizing events from video has attracted a lot of attention in recent years. However, due to the complex nature of user-defined events, the use of purely audio- visual content analysis without domain knowledge has been found to be grossly inadequate. In this paper, we propose to construct a semantic-visual knowledge base to encode the rich event-centric concepts and their relationships from the well- established lexical databases, including FrameNet, as well as the concept-specific visual knowledge from ImageNet. Based on this semantic-visual knowledge bases, we design an effective system for video event recognition. Specifically, in order to narrow the semantic gap between the high-level complex events and low-level visual representations, we utilize the event-centric semantic concepts encoded in the knowledge base as the intermediate-level event representation, which offers both human-perceivable and machine-interpretable semantic clues for event recognition. In addition, in order to leverage the abundant ImageNet images, we propose a robust transfer learning model to learn the noise- resistant concept classifiers for videos. Extensive experiments on various real-world video datasets demonstrate the superiority of our proposed system as compared to the state-of-the-art approaches.
We present a new treebank of English and French technical forum content which has been annotated for grammatical errors and phrase structure. This double annotation allows us to empirically measure the effect of errors on parsing performance. While it is slightly easier to parse the corrected versions of the forum sentences, the errors are not the main factor in making this kind of text hard to parse.