scholarly journals About Certain Semantic Annotation in Parallel Corpora

2015 ◽  
pp. 67-78
Author(s):  
Violetta Koseska-Toszewa
Keyword(s):  
Semantic Annotation ◽  
Contemporary Theory ◽  
Natural Languages ◽  
Parallel Corpora ◽  
Language Form ◽  
Sequence Of Events ◽  
Sentence Level ◽  
Computer Scientists ◽  
Automatic Methods ◽  
The Given

About Certain Semantic Annotation in Parallel CorporaThe semantic notation analyzed in this works is contained in the second stream of semantic theories presented here – in the direct approach semantics. We used this stream in our work on the Bulgarian-Polish Contrastive Grammar. Our semantic notation distinguishes quantificational meanings of names and predicates, and indicates aspectual and temporal meanings of verbs. It relies on logical scope-based quantification and on the contemporary theory of processes, known as “Petri nets”. Thanks to it, we can distinguish precisely between a language form and its contents, e.g. a perfective verb form has two meanings: an event or a sequence of events and states, finally ended with an event. An imperfective verb form also has two meanings: a state or a sequence of states and events, finally ended with a state. In turn, names are quantified universally or existentially when they are “undefined”, and uniquely (using the iota operator) when they are “defined”. A fact worth emphasizing is the possibility of quantifying not only names, but also the predicate, and then quantification concerns time and aspect.  This is a novum in elaborating sentence-level semantics in parallel corpora. For this reason, our semantic notation is manual. We are hoping that it will raise the interest of computer scientists working on automatic methods for processing the given natural languages. Semantic annotation defined like in this work will facilitate contrastive studies of natural languages, and this in turn will verify the results of those studies, and will certainly facilitate human and machine translations.

scholarly journals On Semantic Annotation in Clarin-PL Parallel Corpora

2015 ◽  
pp. 211-236
Author(s):  
Violetta Koseska-Toszewa ◽  
Roman Roszko
Keyword(s):  
Semantic Annotation ◽  
Automatic Processing ◽  
Natural Languages ◽  
Parallel Corpora ◽  
Sentence Level ◽  
The Given ◽  
The Way

On Semantic Annotation in Clarin-PL Parallel CorporaIn the article, the authors present a proposal for semantic annotation in Clarin-PL parallel corpora: Polish-Bulgarian-Russian and Polish-Lithuanian ones. Semantic annotation of quantification is a novum in developing sentence level semantics in multilingual parallel corpora. This is why our semantic annotation is manual. The authors hope it will be interesting to IT specialists working on automatic processing of the given natural languages. Semantic annotation defined the way it is defined here will make contrastive studies of natural languages more efficient, which in turn will help verify the results of those studies, and will certainly improve human and machine translations.

scholarly journals Trilingual aligned corpus – current state and new applications

2014 ◽  
pp. 13-20
Author(s):  
Ludmila Dimitrova ◽  
Violetta Koseska ◽  
Danuta Roszko ◽  
Roman Roszko
Keyword(s):  
Short Stories ◽  
Semantic Annotation ◽  
Theoretical Background ◽  
The Other ◽  
Literary Texts ◽  
Natural Languages ◽  
Parallel Corpus ◽  
Current State ◽  
Sentence Level ◽  
New Applications

Trilingual aligned corpus – current state and new applicationsThis article describes current state of a trilingual parallel corpus consisted of texts in two Slavic (Bulgarian and Polish) and one Baltic language (Lithuanian). The corpus contains original literary texts (fiction, novels, and short stories) in one of the three languages with translations to the other two, and texts in other languages translated into Bulgarian, Polish, and Lithuanian. A part of the texts are aligned at the sentence level. The authors propose a semantic annotation of verbs appearing in these aligned texts that will facilitate contrastive studies of natural languages. A theoretical background for the proposed semantic annotation is briefly also discussed.

scholarly journals Języki słowiańskie i litewski w korpusach równoległych Clarin-PL

2016 ◽  
Vol 51 ◽  
pp. 191-217
Author(s):  
Violetta Koseska-Toszewa ◽  
Roman Roszko
Keyword(s):  
Corpus Linguistics ◽  
Semantic Annotation ◽  
Original System ◽  
Language Resources ◽  
Natural Languages ◽  
Parallel Corpora ◽  
Parallel Corpus ◽  
Slavic Languages ◽  
European Languages ◽  
Polish Language

Slavic languages and the Lithuanian language in the Clarin-PL parallel corporaThe Clarin Eric and Clarin-PL strategic scientific purpose is to support humanistic research in a multicultural and multilingual Europe. Polish researchers put the emphasis on building a bridge between the Polish language and Polish linguistic technologies and other European languages and their linguistic technologies. So far, the Polish scientific community has mainly focused on Polish-English connections. Clarin-PL has been developing the first and only multilingual corpora of the Polish language in conjunction with other Slavic languages and the Lithuanian language: the Polish-Bulgarian-Russian Parallel Corpus and the Polish- Lithuanian Parallel Corpus. The parallel corpora created by the ISS PAS Corpus Linguistics and Semantics Team break through the existing “canons” and allow scientists access to interlinked multilingual language resources – in the first phase limited to the languages of the three Slavic groups and the Lithuanian language. In the article, the authors present very detailed information on their original system of the semantic annotation of scope quantification in multilingual parallel corpora, hitherto unused in the subject literature. Due to the system’s originality, the semantic annotation is carried out manually. Identification of particular values of scope quantification in a sentence and the hereby presented attempts of its recording are supported by long-term research conducted by an international team of linguists and computer scientists / mathematicians developing the issue of quantification of names, time and aspect in natural languages. Języki słowiańskie i litewski w korpusach równoległych Clarin-PLStrategicznym celem naukowym Clarin ERIC i Clarin-PL jest wspieranie badań humanistycznych w wielokulturowej i wielojęzycznej Europie. Dla polskich badaczy ważna jest budowa pomostu między językiem polskim, polskimi technologiami językowymi a innymi językami europejskimi i na ich rzecz opracowanymi technologiami językowymi. Dotychczas w nauce polskiej największy nacisk był kładziony na powiązania polsko-angielskie. Clarin-PL opracowuje zatem pierwsze jak dotąd wielojęzyczne korpusy języka polskiego w zestawieniu z innymi językami słowiańskimi oraz z językiem litewskim: Korpus równoległy polsko-bułgarsko-rosyjski i Korpus równoległy polsko-litewski. Tworzone przez Zespół Lingwistyki Korpusowej i Semantyki (IS PAN) korpusy równoległe przełamują dotychczasowe „kanony” i udostępniają nauce powiązane wielojęzyczne zasoby – w pierwszym etapie ograniczone do języków trzech grup słowiańskich oraz języka litewskiego. W artykule autorzy przedstawiają bardzo szczegółową informację o zastosowanej po raz pierwszy w literaturze przedmiotu anotacji semantycznej dotyczącej kwantyfikacji zakresowej w wielojęzycznych korpusach równoległych. Z powodu swojego rozległego zakresu i nowatorstwa ta anotacja semantyczna jest nanoszona ręcznie. Identyfikacja poszczególnych wartości kwantyfikacji zakresowej w zdaniu oraz przedstawiane tu próby jej zapisu są poparte wieloletnimi badaniami międzynarodowego zespołu lingwistów i matematyków-informatyków opracowujących zagadnienie kwantyfikacji imion, czasu i aspektu w językach naturalnych.

Sentence Level Alignment of Digitized Books Parallel Corpora

Informatica ◽  
2018 ◽  
Vol 29 (4) ◽  
pp. 693-710
Author(s):  
Algirdas Laukaitis ◽  
Darius Plikynas ◽  
Egidijus Ostasius
Keyword(s):  
Parallel Corpora ◽  
Sentence Level

scholarly journals Context-Aware Neural Machine Translation for Korean Honorific Expressions

Electronics ◽  
2021 ◽  
Vol 10 (13) ◽  
pp. 1589
Author(s):  
Yongkeun Hwang ◽  
Yanghoon Kim ◽  
Kyomin Jung
Keyword(s):  
Machine Translation ◽  
Deep Neural Networks ◽  
Contextual Information ◽  
Context Aware ◽  
Neural Machine Translation ◽  
Translation Quality ◽  
Sentence Level ◽  
Proposed Model ◽  
The Given ◽  
The Relationship

Neural machine translation (NMT) is one of the text generation tasks which has achieved significant improvement with the rise of deep neural networks. However, language-specific problems such as handling the translation of honorifics received little attention. In this paper, we propose a context-aware NMT to promote translation improvements of Korean honorifics. By exploiting the information such as the relationship between speakers from the surrounding sentences, our proposed model effectively manages the use of honorific expressions. Specifically, we utilize a novel encoder architecture that can represent the contextual information of the given input sentences. Furthermore, a context-aware post-editing (CAPE) technique is adopted to refine a set of inconsistent sentence-level honorific translations. To demonstrate the efficacy of the proposed method, honorific-labeled test data is required. Thus, we also design a heuristic that labels Korean sentences to distinguish between honorific and non-honorific styles. Experimental results show that our proposed method outperforms sentence-level NMT baselines both in overall translation quality and honorific translations.

scholarly journals Semantics, contrastive linguistics and parallel corpora

2014 ◽  
pp. 85-100
Author(s):  
Violetta Koseska
Keyword(s):  
Lexical Semantics ◽  
Semantic Annotation ◽  
Semantic Structure ◽  
Automatic Annotation ◽  
Parallel Corpora ◽  
Parallel Corpus ◽  
Linguistic Form ◽  
Semantic Categories ◽  
Contrastive Linguistics

Semantics, contrastive linguistics and parallel corporaIn view of the ambiguity of the term “semantics”, the author shows the differences between the traditional lexical semantics and the contemporary semantics in the light of various semantic schools. She examines semantics differently in connection with contrastive studies where the description must necessary go from the meaning towards the linguistic form, whereas in traditional contrastive studies the description proceeded from the form towards the meaning. This requirement regarding theoretical contrastive studies necessitates construction of a semantic interlanguage, rather than only singling out universal semantic categories expressed with various language means. Such studies can be strongly supported by parallel corpora. However, in order to make them useful for linguists in manual and computer translations, as well as in the development of dictionaries, including online ones, we need not only formal, often automatic, annotation of texts, but also semantic annotation - which is unfortunately manual. In the article we focus on semantic annotation concerning time, aspect and quantification of names and predicates in the whole semantic structure of the sentence on the example of the “Polish-Bulgarian-Russian parallel corpus”.

scholarly journals Pseudotext Injection and Advance Filtering of Low-Resource Corpus for Neural Machine Translation

2021 ◽  
Vol 2021 ◽  
pp. 1-10
Author(s):  
Michael Adjeisah ◽  
Guohua Liu ◽  
Douglas Omwenga Nyabuga ◽  
Richard Nuetey Nortey ◽  
Jinling Song
Keyword(s):  
Machine Translation ◽  
Language Processing ◽  
Training Data ◽  
Target Language ◽  
Similarity Metrics ◽  
Mahalanobis Distances ◽  
Parallel Corpora ◽  
Parallel Corpus ◽  
Low Resource ◽  
Sentence Level

Scaling natural language processing (NLP) to low-resourced languages to improve machine translation (MT) performance remains enigmatic. This research contributes to the domain on a low-resource English-Twi translation based on filtered synthetic-parallel corpora. It is often perplexing to learn and understand what a good-quality corpus looks like in low-resource conditions, mainly where the target corpus is the only sample text of the parallel language. To improve the MT performance in such low-resource language pairs, we propose to expand the training data by injecting synthetic-parallel corpus obtained by translating a monolingual corpus from the target language based on bootstrapping with different parameter settings. Furthermore, we performed unsupervised measurements on each sentence pair engaging squared Mahalanobis distances, a filtering technique that predicts sentence parallelism. Additionally, we extensively use three different sentence-level similarity metrics after round-trip translation. Experimental results on a diverse amount of available parallel corpus demonstrate that injecting pseudoparallel corpus and extensive filtering with sentence-level similarity metrics significantly improves the original out-of-the-box MT systems for low-resource language pairs. Compared with existing improvements on the same original framework under the same structure, our approach exhibits tremendous developments in BLEU and TER scores.

scholarly journals The development of stemming algorithm for the Uzbek language

2021 ◽  
pp. 1-12
Author(s):  
Ilkhom Izatovich Bakaev
Keyword(s):  
English Language ◽  
Hybrid Approach ◽  
Data Retrieval ◽  
Natural Languages ◽  
Effective Implementation ◽  
Linguistic Rules ◽  
Word Forms ◽  
The Subject ◽  
Normal Word ◽  
The Given

The automatic processing of unstructured texts in natural languages is one of the relevant problems of computer analysis and text synthesis. Within this problem, the author singles out a task of text normalization, which usually suggests such processes as tokenization, stemming, and lemmatization. The existing stemming algorithms for the most part are oriented towards the synthetic languages with inflectional morphemes. The Uzbek language represents an example of agglutinative language, characterized by polysemanticity of affixal and auxiliary morphemes. Although the Uzbek language largely differs from, for example, English language, it is successfully processed by stemming algorithms. There are virtually no examples of effective implementation of stemming algorithms for the Uzbek language; therefore, this questions is the subject of scientific interest and defines the goal of this work. In the course of this research, the author solved the task of bringing the given texts in the Uzbek language to normal form, which on the preliminary stage were tokenized and cleared of stop words. To author developed the method of normalization of texts in the Uzbek language based on the stemming algorithm. The development of stemming algorithm employed hybrid approach with application of algorithmic method, lexicon of linguistic rules and database of the normal word forms of the Uzbek language. The precision of the proposed algorithm depends on the precision of tokenization algorithm. At the same time, the article did not explore the question of finding the roots of paired words separated by spaces, as this task is solved at the stage of tokenization. The algorithm can be integrated into various automated systems for machine translation, information extraction, data retrieval, etc.

scholarly journals Dialog Structure Through the Lens of Gender, Gender Environment, and Power

2017 ◽  
Vol 8 (2) ◽  
pp. 21-55
Author(s):  
Vinodkumar Prabhakaran ◽  
Owen Rambow
Keyword(s):  
Social Context ◽  
Human Behavior ◽  
Social Contexts ◽  
Social Scientists ◽  
Gender Information ◽  
The Social ◽  
Computer Scientists ◽  
Automatic Methods ◽  
And Gender ◽  
Textual Content

Understanding how the social context of an interaction affects our dialog behavior is of great interest to social scientists who study human behavior, as well as to computer scientists who build automatic methods to infer those social contexts. In this paper, we study the interaction of power, gender, and dialog behavior in organizational interactions. In order to perform this study, we first construct the Gender Identified Enron Corpus of emails, in which we semi-automatically assign the gender of around 23,000 individuals who authored around 97,000 email messages in the Enron corpus. This corpus, which is made freely available, is orders of magnitude larger than previously existing gender identified corpora in the email domain. Next, we use this corpus to perform a largescale data-oriented study of the interplay of gender and manifestations of power. We argue that, in addition to one’s own gender, the “gender environment” of an interaction, i.e., the gender makeup of one’s interlocutors, also affects the way power is manifested in dialog. We focus especially on manifestations of power in the dialog structure — both, in a shallow sense that disregards the textual content of messages (e.g., how often do the participants contribute, how often do they get replies etc.), as well as the structure that is expressed within the textual content (e.g., who issues requests and how are they made, whose requests get responses etc.). We find that both gender and gender environment affect the ways power is manifested in dialog, resulting in patterns that reveal the underlying factors. Finally, we show the utility of gender information in the problem of automatically predicting the direction of power between pairs of participants in email interactions.

scholarly journals Foreword

2011 ◽  
Vol 6 (3) ◽  
pp. 385 ◽  
Author(s):  
L.A. Zadeh
Keyword(s):  
Fuzzy Logic ◽  
Natural Language ◽  
Language Processing ◽  
Extension Principle ◽  
Computing With Words ◽  
World Knowledge ◽  
Special Issue ◽  
Large Measure ◽  
Natural Languages ◽  
Computer Scientists

<p>I feel honored by the dedication of the Special Issue of IJCCC to me. I should like to express my deep appreciation to the distinguished Co-Editors and my good friends, Professors Balas, Dzitac and Teodorescu, and to distinguished contributors, for honoring me. The subjects which are addressed in the Special Issue are on the frontiers of fuzzy logic.<br /> <br /> The Foreword gives me an opportunity to share with the readers of the Journal my recent thoughts regarding a subject which I have been pondering about for many years - fuzzy logic and natural languages. The first step toward linking fuzzy logic and natural languages was my 1973 paper," Outline of a New Approach to the Analysis of Complex Systems and Decision Processes." Two key concepts were introduced in that paper. First, the concept of a linguistic variable - a variable which takes words as values; and second, the concept of a fuzzy if- then rule - a rule in which the antecedent and consequent involve linguistic variables. Today, close to forty years later, these concepts are widely used in most applications of fuzzy logic.<br /> <br /> The second step was my 1978 paper, "PRUF - a Meaning Representation Language for Natural Languages." This paper laid the foundation for a series of papers in the eighties in which a fairly complete theory of fuzzy - logic-based semantics of natural languages was developed. My theory did not attract many followers either within the fuzzy logic community or within the linguistics and philosophy of languages communities. There is a reason. The fuzzy logic community is largely a community of engineers, computer scientists and mathematicians - a community which has always shied away from semantics of natural languages. Symmetrically, the linguistics and philosophy of languages communities have shied away from fuzzy logic.<br /> <br /> In the early nineties, a thought that began to crystallize in my mind was that in most of the applications of fuzzy logic linguistic concepts play an important, if not very visible role. It is this thought that motivated the concept of Computing with Words (CW or CWW), introduced in my 1996 paper "Fuzzy Logic = Computing with Words." In essence, Computing with Words is a system of computation in which the objects of computation are words, phrases and propositions drawn from a natural language. The same can be said about Natural Language Processing (NLP.) In fact, CW and NLP have little in common and have altogether different agendas.<br /> <br /> In large measure, CW is concerned with solution of computational problems which are stated in a natural language. Simple example. Given: Probably John is tall. What is the probability that John is short? What is the probability that John is very short? What is the probability that John is not very tall? A less simple example. Given: Usually Robert leaves office at about 5 pm. Typically it takes Robert about an hour to get home from work. What is the probability that Robert is home at 6:l5 pm.? What should be noted is that CW is the only system of computation which has the capability to deal with problems of this kind. The problem-solving capability of CW rests on two key ideas. First, employment of so-called restriction-based semantics (RS) for translation of a natural language into a mathematical language in which the concept of a restriction plays a pivotal role; and second, employment of a calculus of restrictions - a calculus which is centered on the Extension Principle of fuzzy logic.<br /> <br /> What is thought-provoking is that neither traditional mathematics nor standard probability theory has the capability to deal with computational problems which are stated in a natural language. Not having this capability, it is traditional to dismiss such problems as ill-posed. In this perspective, perhaps the most remarkable contribution of CW is that it opens the door to empowering of mathematics with a fascinating capability - the capability to construct mathematical solutions of computational problems which are stated in a natural language. The basic importance of this capability derives from the fact that much of human knowledge, and especially world knowledge, is described in natural language.<br /> <br /> In conclusion, only recently did I begin to realize that the formalism of CW suggests a new and challenging direction in mathematics - mathematical solution of computational problems which are stated in a natural language. For mathematics, this is an unexplored territory.</p>

Sign in / Sign up

Export Citation Format

Share Document