Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
The recent construction of large linguistic treebanks for spoken and written Dutch (e.g. CGN, LASSY, Alpino) has created new and exciting opportunities for the empirical investigation of Dutch syntax and semantics. However, the exploitation of those treebanks requires knowledge of specific data structures and query languages such as XPath. Linguists who are unfamiliar with formal languages are often reluctant towards learning such a language. In order to make treebank querying more attractive for non-technical users we developed GrETEL (Greedy Extraction of Trees for Empirical Linguistics), a query engine in which linguists can use natural language examples as a starting point for searching the Lassy treebank without knowledge about tree representations nor formal query languages. By allowing linguists to search for similar constructions as the example they provide, we hope to bridge the gap between traditional and computational linguistics. Two case studies are conducted to provide a concrete demonstration of the tool. The architecture of the tool is optimised for searching the LASSY treebank, but the approach can be adapted to other treebank lay-outs.
Mood is an important aspect of music and knowledge of mood can be used as a basic feature in music recommender and retrieval systems. A listening experiment was carried out establishing ratings for various moods and a number of attributes, e.g., valence and arousal. The analysis of these data covers the issues of the number of basic dimensions in music mood, their relation to valence and arousal, the distribution of moods in the valence-arousal plane, distinctiveness of the labels, and appropriate (number of) labels for full coverage of the plane. It is also shown that subject-averaged valence and arousal ratings can be predicted from music features by a linear model.
We also show ensemble dependency parsing and self training approaches applicable to under-resourced languages using our manually annotated dependency structures. We show that for an under-resourced language, the use of tuning data for a meta classifier is more effective than using it as additional training data for individual parsers. This meta-classifier creates an ensemble dependency parser and increases the dependency accuracy by 4.92% on average and 1.99% over the best individual models on average. As the data sizes grow for the the under-resourced language a meta classifier can easily adapt. To the best of our knowledge this is the first full implementation of a dependency parser for Indonesian. Using self-training in combination with our Ensemble SVM Parser we show additional improvement. Using this parsing model we plan on expanding the size of the corpus by using a semi-supervised approach by applying the parser and correcting the errors, reducing the amount of annotation time needed.
Language resources are essential for linguistic research and the development of NLP applications. Low-density languages, such as Irish, therefore lack significant research in this area. This paper describes the early stages in the development of new language resources for Irish – namely the first Irish dependency treebank and the first Irish statistical dependency parser. We present the methodology behind building our new treebank and the steps we take to leverage upon the few existing resources. We discuss language-specific choices made when defining our dependency labelling scheme, and describe interesting Irish language characteristics such as prepositional attachment, copula and clefting. We manually develop a small treebank of 300 sentences based on an existing POS-tagged corpus and report an inter-annotator agreement of 0.7902. We train MaltParser to achieve preliminary parsing results for Irish and describe a bootstrapping approach for further stages of development.
This paper describes a method to convert existing treebanks with syntactic information into banks of meaning representations. The central component is a system of evaluation for a small formal language with respect to an information state. Inputs to the evaluation system are formal language expressions obtained from the conversion of parsed representations conforming to (Penn Treebank Project) guidelines. Outputs from the evaluation system are Davidsonian (higher-order) predicate logic meaning representations. Having a system of evaluation as the basis for generating meaning representations makes possible accepting input with minimal conversion from existing treebanks and from the tools used to construct treebanks. Results of having built corresponding banks of meaning representations from available treebanks are discussed.
This article presents an online dictionary environment, with enhanced sorting and searching functionalities and a text to speech feature, for hearing the pronunciation of the words. The online dictionary environment has been developed as part of the ‘Syntychies’ research program. ‘Syntychies’ online environment is a pioneering webservice for Greek dialectal lexicography and it is the first of its kind for Cypriot Greek.
The Latvian Treebank is being developed since 2010. In this paper we describe the latest developments of this project and the problems currently faced. We examine several gaps in our annotation scheme like determinant, ellipsis and insertion annotation and describe solutions we have chosen.
This paper presents our work for participation in the 2012 CIPS-SIGHAN shared task of Traditional Chinese Parsing. We have adopted two multilingual parsing models – a factored model (Stanford Parser) and an unlexicalized model (Berkeley Parser) for parsing the Sinica Treebank. This paper also proposes a new Chinese unknown word model and integrates it into the Berkeley Parser. Our experiment gives the first result of adapting existing multilingual parsing models to the Sinica Treebank and shows that the parsing accuracy can be improved by our suggested approach.
Treebank is a basic language resource for training and testing syntactic parser which forms a key module in various NLP systems like machine translation system. This paper reports an ongoing research of building dependency treebank for Kashmiri (KashTreeBank) and discusses some main annotation issues. The paper is based on the pilot annotation of 500 sentences.
Morphological segmentation data for the METU-Sabanci Turkish Treebank is provided in this paper. The generalized lexical forms of the morphemes which the treebank previously lacked are added to the treebank. This data maybe used to train POS-taggers that use stemmer outputs to map these lexical forms to morphological tags.
An important goal of addiction research and treatment is to predict behavioural responses to drug-related stimuli. This goal is especially important for patients with impaired insight, which can interfere with therapeutic interventions and potentially invalidate self-report questionnaires. This research tested (i) whether event-related potentials, specifically the late positive potential, predict choice to view cocaine images in cocaine addiction; and (ii) whether such behaviour prediction differs by insight (operationalized in this study as self-awareness of image choice). Fifty-nine cocaine abusers and 32 healthy controls provided data for the following laboratory components that were completed in a fixed-sequence (to establish prediction): (i) event-related potential recordings while passively viewing pleasant, unpleasant, neutral and cocaine images, during which early (400-1000 ms) and late (1000-2000 ms) window late positive potentials were collected; (ii) self-reported arousal ratings for each picture; and (iii) two previously validated tasks: one to assess choice for viewing these same images, and the other to group cocaine abusers by insight. Results showed that pleasant-related late positive potentials and arousal ratings predicted pleasant choice (the choice to view pleasant pictures) in all subjects, validating the method. In the cocaine abusers, the predictive ability of the late positive potentials and arousal ratings depended on insight. Cocaine-related late positive potentials better predicted cocaine image choice in cocaine abusers with impaired insight. Another emotion-relevant event-related potential component (the early posterior negativity) did not show these results, indicating specificity of the late positive potential. In contrast, arousal ratings better predicted respective cocaine image choice (and actual cocaine use severity) in cocaine abusers with intact insight. Taken together, the late positive potential could serve as a biomarker to help predict drug-related choice--and possibly associated behaviours (e.g. drug seeking in natural settings, relapse after treatment)--when insight (and self-report) is compromised.
As little is known about the effectiveness of different types of implementation intentions on the regulation of emotions, the present experiments focused on the differential effectiveness of various implementation intentions on the down-regulation of disgust responses. In Experiment 1, an antecedent-focused implementation intention based on cognitive reappraisal allowed participants to rate disgusting pictures as being less unpleasant than participants in the control condition or the goal intention condition, while the reported intensity (arousal) ratings stayed unaffected. In Experiment 2, participants with a response-focused implementation intention, devised to regulate the intensity of the emotional experience, reported a lower evoked arousal after seeing the disgusting slides, while the valence ratings remained unchanged. Thus, implementation intentions were shown to exert differential effects depending on whether they targeted one or another emotional dimension (i.e., valence vs. arousal).
The annotations of explicit and implicit discourse connectives in the Penn Discourse Treebank make it possible to investigate on a large scale how different types of discourse relations are expressed. Assuming an account of the Uniform Information Density hypothesis, we expect that discourse relations should be expressed explicitly with a discourse connector when they are unexpected, but may be implicit when the discourse relation can be anticipated. We investigate whether discourse relations which have been argued to be expected by the comprehender exhibit a higher ratio of implicit connectors. We find support for two hypotheses put forth in previous research which suggest that continuous and causal relations are presupposed by language users when processing consecutive sentences in a text. We then proceed to analyze the effect of Implicit Causality (IC) verbs (which have been argued to raise an expectation for an explanation) as a local cue for an upcoming causal relation.
The Chinese comma signals the boundary of discourse units and also anchors discourse relations between adjacent text spans. In this work, we propose a discourse structureoriented classification of the comma that can be automatically extracted from the Chinese Treebank based on syntactic patterns. We then experimented with two supervised learning methods that automatically disambiguate the Chinese comma based on this classification. The first method integrates comma classification into parsing, and the second method adopts a “post-processing ” approach that extracts features from automatic parses to train a classifier. The experimental results show that the second approach compares favorably against the first approach. 1
We propose Symbol-Refined Tree Substitution Grammars (SR-TSGs) for syntactic parsing. An SR-TSG is an extension of the conventional TSG model where each nonterminal symbol can be refined (subcategorized) to fit the training data. We aim to provide a unified model where TSG rules and symbol refinement are learned from training data in a fully automatic and consistent fashion. We present a novel probabilistic SR-TSG model based on the hierarchical Pitman-Yor Process to encode backoff smoothing from a fine-grained SR-TSG to simpler CFG rules, and develop an efficient training method based on Markov Chain Monte Carlo (MCMC) sampling. Our SR-TSG parser achieves an F1 score of 92.4% in the Wall Street Journal (WSJ) English Penn Treebank parsing task, which is a 7.7 point improvement over a conventional Bayesian TSG parser, and better than state-of-the-art discriminative reranking parsers. 1
Noun with genitive marker in Indo-Aryan language can variously be a child of a noun, a verb or a complex predicate, thus making it an important parsing issue. In this paper, we examine genitive data of Hindi and aim to automatically determine the attachment and relational label of the same in a dependency framework. We implement two approaches: a rule based approach and a statistical approach. The rule based approach produces promising results but fails to handle certain constructions because of its greedy selection. The statistical approach overcomes this by using a single candidate approach that considers all the possible candidates for the head and chooses the most probable candidate among them. Both approaches are applied on controlled and open environment data. A Controlled environment refers to the situation when the relational labels are attested to the input data except for the genitive data; while open environment refers to cases in which the input is only POS tagged and chunked. The rule based and statistical systems produce a high accuracy of 95 % and 97 % respectively for attachment and perform considerably well for labeling in controlled environment but poorly in open environment.
Development of interpersonal relationships is a fundamental human motivation, and behaviors facilitating social bonding are prized. Some individuals experience enhanced reward from alcohol in social contexts and may be at heightened risk for developing and maintaining problematic drinking. We employed a 3 (group beverage condition) ×2 (genotype) design (N = 422) to test the moderating influence of the dopamine D4 receptor gene (DRD4 VNTR) polymorphism on the effects of alcohol on social bonding. A significant gene x environment interaction showed that carriers of at least one copy of the 7-repeat allele reported higher social bonding in the alcohol, relative to placebo or control conditions, whereas alcohol did not affect ratings of 7-absent allele carriers. Carriers of the 7-repeat allele were especially sensitive to alcohol's effects on social bonding. These data converge with other recent gene-environment interaction findings implicating the DRD4 polymorphism in the development of alcohol use disorders, and results suggest a specific pathway by which social factors may increase risk for problematic drinking among 7-repeat carriers. More generally, our findings highlight the potential utility of employing transdisciplinary methods that integrate genetic methodologies, social psychology, and addiction theory to improve theories of alcohol use and abuse.
Domain adaptation is an important topic for natural language processing. There has been extensive research on the topic and various methods have been explored, including training data selection, model combination, semi-supervised learning. In this study, we propose to use a goodness measure, namely, description length gain (DLG), for domain adaptation for Chinese word segmentation. We demonstrate that DLG can help domain adaptation in two ways: as additional features for supervised segmenters to improve system performance, and also as a similarity measure for selecting training data to better match a test set. We evaluated our systems on the Chinese Penn Treebank version 7.0, which has 1.2 million words from five different genres, and the Chinese Word Segmentation Bakeoff-3 data.
In this thesis, an implementation of the MAT algorithm for query learning is improved and tested, specifically by attempted simulation of the MAT oracle using large corpora. In order to arrive at suitable test corpora, algorithms for generating positive and negative examples in relation to regular tree languages are developed.
Among various neural network language models (NNLMs), recurrent neural network-based language models (RNNLMs) are very competitive in many cases. Most current RNNLMs only use one single feature stream, i.e., surface words. However, previous studies proved that language models with additional linguistic information achieve better performance. In this study, we extend RNNLM by explicitly integrating additional linguistic information, including morphological, syntactic, or semantic factors. Our proposed RNNLM is called a factored RNNLM that is expected to enhance RNNLMs. A number of experiments are carried out that show the factored RNNLM improves the performance for all considered tasks: consistent perplexity and word error rate (WER) reductions. In the Penn Treebank corpus, the relative improvements over n-gram LM and RNNLM are 29.0 % and 13.0%, respectively. In the IWSLT-2011 TED ASR test set, absolute WER reductions over RNNLM and n-gram LM reach 0.63 and 0.73 points. Title and Abstract in another language (Chinese) ddddddddddddddd ddddddddddddNNLMddddddddddddddRNNLMddd ddddddddddddddddddddRNNLMddddddddddddddd ddddddddddddddddddddddddddddddddddddddd ddddddddddddddddddRNNLMddddddddddddddddd dddddddddfRNNLMddddddddddddddddfRNNLMdddddd dddddddRNNLMdddddddddddddddddddddWERddddd ddddfRNNLMdddddddddnddddddRNNLMddddd29.0%d13.0%d dIWSLT-2011 TEDdddddddddfRNNLMddddddddddd0.63ddd dnddddddd0.73ddddRNNLMdd
We present a detailed error analysis of a transition-based dependency parser trained on a Hindi dependency treebank. Parser error analysis has not been systematically examined from the point of view of treebanking before and this work intends to contribute in this area. We address two main questions in this paper: Can the parsing of certain structures be made easier by using alternative analyses for these structures? Are there certain linguistic cues implicit (or missing) in the current treebank that can be made explicit (or added) in order to make the parsing of complex constructions easier? These questions will guide us in examining the potential benefits of parser error analysis during treebanking. Through our experiments and analysis we were able to shed light on the causes of errors and subsequently have been able to improve the performance of the parser.
Dependency parsing has attracted considerable interest from researchers and developers in natural language processing. However, to obtain a high‐accuracy dependency parser, supervised techniques require a large volume of hand‐annotated data, which are extremely expensive. This paper presents a simple and effective approach for improving dependency parsing with subtrees derived from unannotated data, which are easy to obtain. First, we use a baseline parser to parse large‐scale unannotated data. Then, we extract subtrees from dependency parse trees in the auto‐parsed data. Next, the extracted subtrees are classified into several sets according to their frequency. Finally, we design new features based on the subtree sets for parsing algorithms. To demonstrate the effectiveness of our proposed approach, we conduct experiments on the English Penn Treebank and Chinese Penn Treebank. The results show that our approach significantly outperforms baseline systems. It also achieves the best accuracy for the Chinese data and an accuracy competitive with the best known systems for the English data.
After a period when the focus was essentially on mental architecture, the cognitive sciences are increasingly integrating the social dimension. The rise of a cognitive sociolinguistics is part of this trend. The article argues that this process requires a re-evaluation of some entrenched positions in linguistics: those that see linguistic norms as antithetical to a descriptive and variational linguistics. Once such a re-evaluation has taken place, however, the social recontextualization of cognition will enable linguistics (including sociolinguistics as an integral part), to eliminate the cracks in the foundations that were the result of suppressing the sociocultural underpinnings of linguistic facts. Structuralism, cognitivism and social constructionism introduced new and necessary distinctions, but in their strong forms they all turned into unnecessary divides. The article tries to show that an evolutionary account can reintegrate the opposed fragments into a whole picture that puts each of them in their ‘ecological position’ with respect to each other. Empirical usage facts should be seen in the context of operational norms in relation to which actual linguistic choices represent adaptations. Variational patterns should be seen in the context of structural categories without which there would be only ‘differences’ rather than variation. And emergence, individual choice, and flux should be seen in the context of the individual’s dependence on lineages of community practice sustained by collective norms.
Lexical substitutes have found use in areas such as paraphrasing, text simplification, machine translation, word sense disambiguation, and part of speech induction. However the computational complexity of accurately identifying the most likely substitutes for a word has made large scale experiments difficult. In this letter we introduce a new search algorithm, FASTSUBS, that is guaranteed to find the K most likely lexical substitutes for a given word in a sentence based on an n-gram language model. The computation is sublinear in both K and the vocabulary size V. An implementation of the algorithm and a dataset with the top 100 substitutes of each token in the WSJ section of the Penn Treebank are available at http://goo.gl/jzKH0.
It is necessary to have a large annotated Treebank to build a statistical dependency parser.Acquisition of such a Treebank is time consuming,tedious and expensive.This paper presents a method to reduce this demand via active learning,which selects the most uncertain samples for annotation instead of the whole training corpus.Experiments are carried out on the HIT-CIR-CDT,our results show that the parsing accuracy rises about 0.8 percent by active learning when using the same amount of training samples.In other words,for about the same parsing accuracy,we only need to annotate 70% of the samples as compared to the usual random selection method.
This paper presents a new method to evaluate machine translation (MT) systems against a parallel treebank. This approach examines specific linguistic phenomena rather than the overall performance of the system. We show that the evaluation accuracy can be increased by using word alignments extracted from a parallel treebank. We compare the performance of our statistical MT system with two other competitive systems with respect to a set of problematic linguistic structures for translation between German and French.
An investigation of the relationship between background knowledge and reading comprehension performance on standardized reading tests (the California STAR Test) was conducted with sixth, seventh, and eighth-grade ethnic minority children from low-income backgrounds (N = 68). Predictor variables examined included perceived background knowledge (overall and topic-specific), GPA, basic literacy skills, reading self-concept, race and ethnicity, language background, gender, and grade level. Research questions addressed participants' familiarity with topics discussed in STAR test reading passages and about the predictive nature of participant rankings and ratings of passages, as measured by the Topic Familiarity Ranking Measure and the Topic Familiarity Rating Scale. Results indicated that background knowledge of passage topics had a significant positive association (p <.05) with reading comprehension performance for 30% of the CST passages, for seventh and eighth-grade participants. Hierarchical regression analyses conducted on three of the passages showed that between 7% and 16% of the variance in reading comprehension performance was accounted for by background knowledge, as measured by the Topic Familiarity Rating Scale.
Training data selection is a common method for domain adaptation, the goal of which is to choose a subset of training data that works well for a given test set. It has been shown to be effective for tasks such as machine translation and parsing. In this paper, we propose several entropy-based measures for training data selection and test their effectiveness on two tasks: Chinese word segmentation and part-of-speech tagging. The experimental results on the Chinese Penn Treebank indicate that some of the measures provide a statistically significant improvement over random selection for both tasks.
In this article, we investigate ambiguity in syntactic annotation. The ambiguity in question is inherent in a way that even human annotators interpret the meaning differently. In our experiment, we detect potential structurally ambiguous sentences with Constraint Grammar rules. In the linguistic phenomena we investigate, structural ambiguity is primarily caused by word order. The potentially ambiguous particle or adverbial is located between the main verb and the (participial) NP. After detecting the structures, we analyze how many of the potentially ambiguous cases are actually ambiguous using the double-blind method. We rank the sentences captured by the rules on a 1 to 5 scale to indicate which reading the annotator regards as the primary one. The results indicate that 67% of the sentences are ambiguous. Introducing ambiguity in the treebank/parsebank increases the informativeness of the representation since both correct analyses are presented.
We investigate the stacking of dependency and phrase structure parsers, i.e. we define features from the output of a phrase structure parser for a dependency parser and vice versa. Our features are based on the original form of the external parses and we also compare this approach to converting phrase structures to dependencies then applying standard stacking on the converted output. The proposed method provides high accuracy gains for both phrase structure and dependency parsing. With the features derived from the phrase structures, we achieved a gain of 0.89 percentage points over a state-of-the-art parser and reach 93.95 UAS, which is the highest reported accuracy score on dependency parsing of the Penn Treebank. The phrase structure parser obtains 91.72 F-score with the features derived from the dependency trees, and this is also competitive with the best reported PARSEVAL scores for the Penn Treebank.
We describe a transformation-based learning method for learning a sequence of mono-lingual tree transformations that improve the agreement between constituent trees and word alignments in bilingual corpora. Using the manually annotated English Chinese Transla-tion Treebank, we show how our method au-tomatically discovers transformations that ac-commodate differences in English and Chi-nese syntax. Furthermore, when transforma-tions are learned on automatically generated trees and alignments from the same domain as the training data for a syntactic MT system, the transformed trees achieve a 0.9 BLEU im-provement over baseline trees. 1
In this paper we describe a two-stage dependency parser for Bangla. In the first stage, we build a model using a Bangla dependency Treebank and subsequently this model is used to build a data driven Bangla parser. In the second stage, constraint based parsing has been used to modify the output of the data driven parser. This second stage module implements the Bangla specific constraints with the help of demand frames of Bangla verbs. The features of the words used in both these stages include morphological features like gender, number, person, etc., parts-of-speech tags, chunk tags and named entity tags. The evaluation results show that this two stage parser performs better than one stage parsers. 1
This contribution presents an online lexical database and shows how some of its data categories were converted to make them more accessible to users. The database is called DiCoInfo and contains English, French, and Spanish terms related to the fields of computing and the Internet. Entries are compiled according to the principles of Explanatory Combinatorial Lexicology, ECL (Mel’čuk et al. 1995). First, we present the basic structure of the entry, focussing on the encoding of collocations (based on lexical functions). Then, we show how the meaning of collocations can be described with natural language explanations, how actantial structures can be better reflected in these explanations, how users can browse collocations in order to find collocates that express specific meanings, and finally how they can search for translations of collocations. Our work tends to demonstrate that, even though the encoding performed by lexicographers is semi-formal and proves necessary for the new functionalities described, it can still lend itself to adaptations defined according to specific user needs, and ultimately to meet those needs. This seems to be confirmed by the preliminary results of a pilot study we conducted on the browsing of collocations.
Decoding pain in others is of high individual and social benefit in terms of harm avoidance and demands for accurate care and protection. The processing of facial expressions includes both specific neural activation and automatic congruent facial muscle reactions. While a considerable number of studies investigated the processing of emotional faces, few studies specifically focused on facial expressions of pain. Analyses of brain activity and facial responses elicited by the perception of facial pain expressions in contrast to other emotional expressions may unravel the processing specificities of pain-related information in healthy individuals and may contribute to explaining attentional biases in chronic pain patients. In the present study, 23 participants viewed short video clips of neutral, emotional (joy, fear), and painful facial expressions while affective ratings, event-related brain responses, and facial electromyography (Musculus corrugator supercilii, M. orbicularis oculi, M. zygomaticus major, M. levator labii) were recorded. An emotion recognition task indicated that participants accurately decoded all presented facial expressions. Electromyography analysis suggests a distinct pattern of facial response detected in response to happy faces only. However, emotion-modulated late positive potentials revealed a differential processing of pain expressions compared to the other facial expressions, including fear. Moreover, pain faces were rated as most negative and highly arousing. Results suggest a general processing bias in favor of pain expressions. Findings are discussed in light of attentional demands of pain-related information and communicative aspects of pain expressions.
Title: Valency of verbs in the Prague Dependency Treebank Author: PhDr. Zdeňka Urešová Department: Institute of Formal and Applied Linguistics MFF UK Supervisor: Prof. PhDr. Eva Hajičová, DrSc. Abstract: This dissertation describes PDT-Vallex, a valency lexicon of Czech verbs, and its relation to the annotation of the Prague Dependency Treebank (PDT). The PDT-Vallex lexicon was created during the an- notation of the PDT and it is a valuable source of verbal valency information available both for linguistic research and for computer- ized natural language processing. In this thesis, we describe not only the structure and design of the lexicon (which is closely related to the notion of valency as developed in the Functional Generative De- scription of language) but also the relation between the PDT-Vallex and the PDT. The explicit and full-coverage linking of the lexicon to the treebank prompted us to pay special attention to diatheses; we propose formal transformation rules for diatheses to handle their surface realization even when the canonical forms of verb arguments as captured in the lexicon do not correspond to the forms of these arguments actually appearing in the corpus.
The paper presents and evaluates an efficient algorithm for measuring semantic similarity of texts. Calculating the level of semantic similarity of texts is a very difficult task and the proposed up to now methods suffer from computational complexity. This substantially limits their application area. The proposed algorithm tries to reduce the problem by merging a computationally efficient statistical approach to text analysis with a semantic component. The semantic properties of text words are extracted from the WordNet lexical database. The approach was tested using WordNets for two languages: English and Polish. The basic properties of this approach are also studied. The paper concludes with an analysis of the performance of the proposed method on a sample database and suggests some possible application areas.
A novel method for hybrid graph-based dependency parsing of natural language text is proposed. It is based on k-best maximum spanning tree dependency parsing and evaluation of the spanning trees by using a verb valency lexicon for a given language as a reranking knowledge base. The approach is compared with existing state-of-the-art transition-based and graph-based approaches to dependency parsing. As the proposed generic method was developed specifically for improving the accuracy of Croatian dependency parsing, Croatian Dependency Treebank and CROVALLEX verb valency lexicon are used in the experiment. The suggested approach scored approximately 77.21 % LAS, outperforming the tested state-of-the-art approaches by at least 2.68 % LAS. TITLE AND ABSTRACT IN CROATIAN Ovisnosno parsanje pomoću k najboljih razapinjućih stabala i ponovnoga vrjednovanja valencijskim rječnikom glagola Predlaže se novi pristup hibridnom ovisnosnom parsanju tekstova prirodnoga jezika temeljenom na teoriji grafova. Pristup je zasnovan na ovisnosnom parsanju pomoću k najboljih razapinjućih
Unsupervised dependency parsing is one of the most challenging tasks in natural languages processing. The task involves finding the best possible dependency trees from raw sentences without getting any aid from annotated data. In this paper, we illustrate that by applying a supervised incremental parsing model to unsupervised parsing; parsing with a linear time complexity will be faster than the other methods. With only 15 training iterations with linear time complexity, we gain results comparable to those of other state of the art methods. By employing two simple universal linguistic rules inspired from the classical dependency grammar, we improve the results in some languages and get the state of the art results. We also test our model on a part of the ongoing Persian dependency treebank. This work is the first work done on the Persian language. 1
State of the art parsers are currently trained on converted versions of Penn Treebank into dependency representations which however don’t include null elements. This is done to facilitate structural learning and prevent the probabilistic engine to postulate the existence of deprecated null elements everywhere (see [15]). However it is a fact that in this way, the semantics of the representation used and produced on runtime is inconsistent and will reduce dramatically its usefulness in real life applications like Information Extraction, Q/A and other semantically driven fields by hampering the mapping of a complete logical form. What systems have come up with are “Quasi”-logical forms or partial logical forms mapped directly from the surface representation in dependency structure. We show the most common problems derived from the conversion and then describe an algorithm that we have implemented to apply to our converted Italian Treebank, that can be used on any CONLL-style treebank or representation to produce an “almost complete” semantically consistent dependency treebank.
A major computational burden, while performing document clustering, is the calculation of similarity measure between a pair of documents. Similarity measure is a function that assign a real number between 0 and 1 to a pair of documents, depending upon the degree of similarity between them. A value of zero means that the documents are completely dissimilar whereas a value of one indicates that the documents are practically identical. Traditionally, vector-based models have been used for computing the document similarity. The vector-based models represent several features present in documents. These approaches to similarity measures, in general, cannot account for the semantics of the document. Documents written in human languages contain contexts and the words used to describe these contexts are generally semantically related. Motivated by this fact, many researchers have proposed semantic-based similarity measures by utilizing text annotation through external thesauruses like WordNet (a lexical database). In this paper, we define a semantic similarity measure based on documents represented in topic maps. Topic maps are rapidly becoming an industrial standard for knowledge representation with a focus for later search and extraction. The documents are transformed into a topic map based coded knowledge and the similarity between a pair of documents is represented as a correlation between the common patterns. The experimental studies on the text mining datasets reveal that this new similarity measure is more effective as compared to commonly used similarity measures in text clustering.
This chapter explores the potential and challenges of basing pedagogical, bilingual electronic dictionaries on WordNet – a comprehensive machine-readable lexical database of the English language. It introduces the specifics of WordNet’s complex lexical architecture; explores the interfaces between WordNet and humanistic, pedagogical and bilingual lexicography; and highlights the Transpoetika Dictionary – an innovative, bilingualised Serbian-English pedagogical dictionary. The chapter shows that the potential of relational lexical databases by far exceeds their exclusive application in language engineering and automatic text processing. The author argues that WordNet can not only serve as a general lexicographic framework for building humanistic eDictionaries, but also pave the way for other collaborative projects in Natural Language Processing (NLP) and pedagogic lexicography. The author also discusses strategies for expanding the scope of WordNet-based humanistic dictionaries through web services and social media platforms such as Flickr and Twitter.
You have accessThe ASHA LeaderFeature1 Oct 2012Coaching Via Telepractice Cheryl Broekelmann, andMS Jeanne FlowersMS, CCC-SLP Cheryl Broekelmann Google Scholar More articles by this author, MS and Jeanne Flowers Google Scholar More articles by this author, MS, CCC-SLP https://doi.org/10.1044/leader.FTR6.17122012.np SectionsAbout ToolsAdd to favorites ShareFacebookTwitterLinked In Molly was born 10 weeks premature. As her parents prepared to take Molly home from the neonatal intensive care unit, they received news that she had failed her hearing screening. Follow-up tests indicated bilateral, profound sensorineural hearing loss, caused by the antibiotics used to keep her alive. She received bilateral cochlear implants when she was 13 months old. A mom and her young daughter work with an ihear therapists using a variety of toys and incorporating language into her typical play routines. Today, Molly lives on a farm in a rural community with her parents and older brother. Her grandparents run the neighboring farm and numerous aunts, uncles, and cousins live nearby. The family is close-knit, serving as one another’s support system. To help develop Molly’s spoken language, her mom contacts the state early intervention program’s staff. They, in turn, refer her to a speech-language pathologist who visits her town once a month. The SLP, however, has never provided intervention to a child with cochlear implants nor received any specialized training in using auditory skills to develop language. The SLP thinks Molly’s family should work with a more experienced clinician, but no one in the area has the right training. Molly’s story is all too familiar. Even as advances in hearing technology help more children develop spoken language through listening, there is a shortage of SLPs trained to work with families to achieve these goals (ASHA, 2011). As a result, families living far from metropolitan areas might drive two or more hours every week to visit a professional, leaving children unable to learn in their natural environments. Teacher of the Deaf, Barb Meyers, works with a child and her mother on responding to the onset of a sound by putting a peg into a peg board. This is a task that mom will work on throughout the week to help her daughter learn to respond to sounds. The ihear Curriculum Recognizing these limitations for families—as well as innovations in telepractice—St. Joseph Institute for the Deaf in St. Louis conducted a pilot study to deliver intervention via the Internet in 2008. The study examined a number of questions: Is service delivery over the Internet effective? How would St. Joseph staff know that the quality of the auditory signal—so critical for children developing auditory skills—was adequate? Could they build relationships with families and children online? How would they keep a 2-year-old’s attention online, or monitor progress? The result was the ihear™ Internet Therapy Program (ihear) and associated curriculum, which allows professionals trained in the development of listening and spoken language to deliver treatment to children with hearing loss, regardless of where the child lives, by coaching a parent or professional who is on-site with the child. The ihear program includes computerized, interactive lesson plans that embed individual objectives within each session. The sessions are outcome-driven and include new strategies that the parent or professional (SLP, teacher of the deaf, or others) can use to improve a specific skill. The ihear program clinician—who may be an SLP or teacher of the deaf (TOD) who has specialized in developing spoken language through listening—coaches the parent or on-site professional to implement strategies, and empowers him or her to understand the needs of the child. Each lesson focuses on the child’s affinities and strengths to teach to individual objectives. Although digitized lesson plans are often used to target specific skills, the program allows therapists to, for example, access websites or utilize a white board to capture a child’s interest. The length and frequency of sessions are based on the child’s individual needs. Sessions occur as often as five times a week for 30 minutes for those children who are working on specific individualized education plan goals, and as little as every other week for 30 minutes for those children who are simply maintaining their language and auditory skills. The ihear curriculum is based on three main components: coaching a clinician, parent, or caregiver in strategies to help children use their newly learned skills; maintaining the client’s privacy under applicable federal laws; and gathering data to determine each lesson’s efficacy. Empowering by Coaching Professionals at St. Joseph Institute for the Deaf use a coaching model in their family-centered early intervention. Therefore, the ihear program considered coaching as a means to serve the school-age population, either at home or in school. Coaching allows a child’s local school professional—teacher of the deaf, SLP, or others—to help develop a child’s auditory skills so the child can learn language and develop articulation. Coaching builds a cooperative and collaborative relationship between the coach—in this case, the ihear clinician—and the on-site parent or professional through empowerment. The partnership between a school professional and ihear clinician makes them equals in the therapeutic process (Turnbull & Turnbull, 2001). Together they determine a specific, evidence-based coaching strategy to be practiced in each session. The ihear clinician demonstrates the coaching strategy and applies it during the intervention session, and the parent or professional then practices the strategy and receives feedback from the ihear clinician. For example, the child may delete the final /s/ from words, so the ihear clinician may suggest checking the child’s auditory detection of the /s/ sound. Professionals depend on each other to help the child succeed. This interdependence is critical for telepractice success, because the professional or parent working directly with the child keeps the ihear clinician apprised of quality indicators, and incorporates cultural and linguistic norms appropriate for the child’s natural environment. For example, nuances of English can vary from region to region. One child recently talked about “silage,” and the ihear clinician didn’t know whether it was a noun or a verb, or what it meant. The child’s mother explained that silage is the food their farm animals eat. The parent’s support prevented a communication breakdown between the clinician and the child. Maintaining Privacy Clinicians providing coaching via telepractice—like all clinicians providing services—must adhere to the ASHA Code of Ethics and federal and state laws if applicable (ASHA, 2005), including the Health Insurance Portability and Accountability Act (HIPAA) and Family Educational Rights and Privacy Act (FERPA). For more information about how HIPAA relates to the use of telepractice, visit ASHA’s frequently asked questions. For Molly’s family, and all the families ihear works with, we wanted to ensure their internet sessions and personal information were secure. The ihear program uses proprietary, noncommercial software that is certified HIPAA- and FERPA-compliant. Other considerations that maintain privacy are the security of treatment rooms during sessions. Identify all people at both sites prior to each session, or whenever someone new enters the session. When a child enrolls in the program, the family or school district works with an ihear technical specialist to ensure that the on-site computer and Internet connection function properly. The ihear technician also installs necessary software on the client’s computer. Measuring Efficacy Consistent with ASHA’s position on telepractice (ASHA, 2005), the ihear program ensures that services provided to the children are of the same quality as those occurring within the classrooms at St. Joseph Institute for the Deaf. To measure the quality of services, ihear used the Internet Therapy Outcome Tracking System (iTOTS™) to monitor and assess intervention efficacy. iTOTS monitors all aspects of the ihear program, with two goals: to aggregate data about the program’s overall efficacy (including Internet service quality, ihear clinician quality via Internet, and digital lesson plan effectiveness) and to collect data assessing children’s progress toward their objectives. In many instances these goals overlap, but analysis occurs individually. Data analysis demonstrates the child’s ability to increase a skill, the lesson’s ability to engage the child, and the parent or professional’s ability to use the strategy in which they’ve been coached. Molly’s Experience At 10 months old, before she received her implants, Molly began ihear telepractice services twice a week for 30 minutes per session. Molly used a few single-word utterances and approximately 50 “baby signs.” The ihear clinician coached Molly’s parents in strategies to develop auditory skills and increase her expressive language. For example, when the phone rang during an ihear session, the ihear clinician taught mom to alert Molly to the sound by pointing at her ear and saying, “Molly, I hear something...do you?” Reinforcing teachable moments in the natural environment makes listening a part of everyday routines. The ihear clinician embedded a strategy, and modeled a technique for skill development. When she began the program, Molly’s standardized assessment score for spoken language was 68, well below the average range of 85-115. Six months later—with her parents and grandparents incorporating the coaching strategies into Molly’s daily activities and routines—her score increased to 82. She talked in three- to four-word sentences and her expressive vocabulary increased almost daily. Molly’s language growth continued as her family learned more strategies over the next two years. In addition to the ability to communicate with their daughter, Molly’s family also gained less tangible—but equally important—benefits. They are learning how to meet her needs and are comfortable helping teach her to listen and talk. And they are able to stay in their close-knit community with appropriate support systems, despite their remote location. According to Molly’s mom, “ihear benefited our family by teaching us how to communicate with our daughter. We are learning so much from our sessions about meeting her needs...knowing what she wants and giving her language to express those wants. Now, we’re more comfortable with the process of teaching her how to listen and talk. We no longer feel ‘out there,’ on our own without any support.” Sources American Speech-Language-Hearing Association. (2005). Speech-language pathologists providing clinical services via telepractice: position statement [Position Statement]. Available from www.asha.org/policy. Google Scholar American Speech-Language-Hearing Association. (2011). Supply and demand resource list for school-based SLPs. Retrieved August 28, 2012 from www.asha.org/uploadedFiles/2011-School-SLP-Supply-Demand.pdf [PDF]. Google Scholar Rush D. D., & Shelden M. L. (2005). Evidence-based definition of coaching practices.CASEinPoint, 1(6), 1–6. Available at:http://www.fippcase.org/caseinpoint/caseinpoint_vol1_no6.pdf Google Scholar Turnbull A., & Turnbull H. (2001). Families, professionals, and exceptionality: Collaborating for empowerment.4th ed.). Columbus, Ohio: Merrill. Google Scholar Author Notes Cheryl Broekelmann, MS, is director of early intervention and ihear at St. Joseph Institute for the Deaf in St. Louis, Missouri. She teaches language in deaf education in the Department of Communication Disorders and Deaf Education at Fontbonne University. Contact her at[email protected]. Jeanne Flowers, MS, CCC-SLP, is an early intervention and ihear clinician at St. Joseph Institute for the Deaf in St. Louis, Missouri. She is an affiliate of Special Interest Group 18, Telepractice. Contact her at[email protected] Advertising Disclaimer | Advertise With Us Advertising Disclaimer | Advertise With Us Additional Resources FiguresSourcesRelatedDetails Volume 17Issue 12October 2012 Get Permissions Add to your Mendeley library History Published in print: Oct 1, 2012 Metrics Current downloads: 2,154 Topicsasha-topicsleader_do_tagasha-article-typesCopyright & Permissions© 2012 American Speech-Language-Hearing AssociationLoading...
The study is based on an experiment to measure the affective states of computer users via their use of mouse and keyboard. The experiment was replicated from a previous study by Khan et al., [5] resulting in significant correlations between the computer users pattern of interactions and their valence, arousal ratings. This study utilized the same data set from [5] and re-confirmed its validity by training Artificial Neural Networks (ANN). The data was divided into two portions for each individual. A portion to train ANN on his/her patterns of interaction and other portion to test the ANN. The study resulted in an average recognition rate of 64.72 % for valence and 61.02 % for arousal ratings. The highest recognition rates for individual participants' valence and arousal were 100% and 87% respectively. These figures suggest that ANN is a bright prospect for the measurement of affective states of individual computer users via their interaction with keyboard and mouse.