Báo cáo khoa học: "Domain Adaptation with Active Learning for Word Sense Disambiguation" pdf

... touse active learning for domain adaptation for WSD.A similar work is the recent research by Chen et al.(2006), where active learning was used successfullyto reduce the annotation effort for ... 49–56,Prague, Czech Republic, June 2007.c2007 Association for Computational LinguisticsDomain Adaptation with Active Learning for Word Sense DisambiguationYee Seng Chan and Hwee Tou NgDepartment ... importance of domain adaptation for word sense disambiguation. In this pa-per, we ﬁrst show that an active learning ap-proach can be successfully used to performdomain adaptation of WSD systems....

Báo cáo khoa học: "SenseRelate::TargetWord – A Generalized Framework for Word Sense Disambiguation" doc

... the word senses.In our system, this module ﬁrst decides the base(uninﬂected) form of each of the n words. It thenretrieves all the senses for each word from the sense inventory. We use WordNet ... MeasureContextTarget Sense PreprocessingFormat Filter Sense InventoryContext SelectionPostprocessingPick Sense Figure 1: A generalized framework for Word Sense Disambiguation.modules to perform any ... byselecting that sense of the target word which is most related to the context words.Relatedness between word senses is mea-sured using the WordNet::Similarity Perlmodules.1 IntroductionMany words...

Báo cáo khoa học: "An Empirical Study on Class-based Word Sense Disambiguation" pdf

... semanticclasses deﬁned for a word. In the sense approach,one classiﬁer is generated for each word sense, andthe classiﬁers choose between the possible senses for the word. The examples to train ... for groupingsenses of the same word, thus producing coarser word sense groupings for better disambiguation.Wikipedia3has been also recently used to over-come some problems of automatic learning ... clas-siﬁer for a concrete word are all the examples ofthis word sense. In the semantic–class approach,one classiﬁer is generated for each semantic class.So, when we want to label a word, our...

Báo cáo khoa học: "Multi-Criteria-based Active Learning for Named Entity Recognition" ppt

... local con-text of the target word w, is also used to classify w. However, for active learning in NER, it is not reasonable to select a single word without context for human to label. Even if ... sent. (223K words) PER 5 sent. (131 words) 7809 sent. (157K words) LOC 5 sent. (130 words) 7809 sent. (157K words) Newswire ORG MUC-6 5 sent. (113 words) 602 sent. (14K words) ... multi-criteria-based active learning approach and effec-tively apply it to named entity recognition. Active learning targets to minimize the human annotation efforts by selecting ex-amples for labeling....

Tài liệu Báo cáo khoa học: "An Equivalent Pseudoword Solution to Chinese Word Sense Disambiguation" ppt

... supervised learning or semi-supervised learning method. This rein-forcement algorithm dates back to Gale et al. (1992a). Their investigation was based on a 6- word test set with 2 senses for each word. ... two words. An ambiguous word has the same number of EPs as of senses. Each EP's sense maps to a sense of ambiguous word. The semantic equivalence demands further equivalence at each sense ... similarity with the word. 4) The morpheme of a pseudoword stands for a sense, while the sense should consist of one or more morphemes. 5) The morpheme should be a monosemous word. The fourth...

Báo cáo khoa học: "Relieving The Data Acquisition Bottleneck In Word Sense Disambiguation" ppt

... clusters;SALAAM identiﬁes the appropriate senses for the words in those clusters based on the wordssenses’ proximity in WordNet. The word sense proximity is measured in information theo-retic terms based ... Resnik(Resnik, 1999);A sense selection criterion is applied to choosethe appropriate sense label or set of sense la-bels for each word in the cluster;The chosen sense tags for the words in thecluster ... Acquisition Bottleneck In Word Sense DisambiguationMona DiabLinguistics DepartmentStanford Universitymdiab@stanford.eduAbstractSupervised learning methods for WSD yield betterperformance than unsupervised...

Báo cáo khoa học: "An Unsupervised Morpheme-Based HMM for Hebrew Morphological Disambiguation" pdf

... word. As a result, for this case,75% of the words remain with one analysis with 95% accuracy, 20% with two analyses and 5% with three analyses.Segal (2000) built a transformation-based tag-ger ... method for languages with aﬃxational morphology inwhich the knowledge of word formationrules (which are quite restricted in He-brew) helps in the disambiguation. Weadapt HMM algorithms for learning ... and79.12% for unsupervised word- based HMM trainedon parts of the LOB corpora, with a tagset of134 tags. With good initial conditions, such asgood approximation of the tag distribution for each word, ...

Báo cáo khoa học: "Domain Adaptation for Machine Translation by Mining Unseen Words" doc

... same pre-processng for all words, the training words and theOOV words. And the resulting feature vectors for each word are used for learning the CCA projectionsSince a word can have multiple ... trans-lations for the OOV German words (Haghighi et al.,2008). From the target domain corpus we extract themost frequent words (approximately 5000) for boththe languages. Of these, words that ... project all the words into thesub-space identiﬁed by CCA and mine translations for the OOV words. We will describe each of thesesteps in detail in this section. For each of the frequent words we extract...

Báo cáo khoa học: "Domain Adaptation of Maximum Entropy Language Models" potx

... optimized on the heldout data. Usually,larger values are used for global parameters and for domains with more data, while for domains with less data, the variance is typically set to besmaller, ... this∗Currently with Tallinn University of Technology, Esto-niapaper is that we show how the suggested hierar-chical adaptation can be used with suitable pri-ors and combined with the class-based ... speech. For training the LMs, two sources were used:ﬁrst 5M sentences from the Gigaword (2nd ed.)corpus (99.5M words), and broadcast news tran-scriptions from the TDT4 corpus (1.19M words).The...

Báo cáo khoa học: "A Combination of Active Learning and Semi-supervised Learning Starting with Positive and Unlabeled Examples for Word Sense Disambiguation: An Empirical Study on Japanese Web Search Query" pdf

... 2009.c2009 ACL and AFNLPA Combination of Active Learning and Semi-supervised Learning Starting with Positive and Unlabeled Examples for Word Sense Disambiguation: An Empirical Study on ... data for word sense disambiguation (WSD) in the do-main of web queries, where a complete set of ambiguous word senses are unknown. In this paper, we present a combination of active learning ... linguis-tic features for Japanese sentence analysis, (a) Word feature within sentences, (b) Preceding word feature within bunsetsu (Japanese base phrase), (c) Backward word feature within bun-setsu,...

Xem thêm