Search

The authors explore the impact of digitization on historical newspaper studies, focusing on the British Library's efforts since 2004. They highlight the challenges posed by the colonial provenance of these sources, especially in the context of expanding digital collections. While digitization offers exciting opportunities for global historical research, it also raises concerns about reproducing colonial asymmetries and the exploitation of labor. The discussion underscores the need for ethical engagement and collaboration in digitization initiatives, particularly concerning colonial legacies and multilingual sources like early Indian newspapers.

Highlights • We present a unified multilingual system for handwriting recognition. • The system is based on a two step-level recognition system using an optical model and a language model. • We propose to model language modeling using sub-lexical units. • One investigates bilingual recognition on French and English datasets. • Using multigrams units allows to build a multilingual system which benefits to the language combination. Abstract We address the design of a unified multilingual system for handwriting recognition. Most of multilingual systems rests on specialized models that are trained on a single language and one of them is selected at test time. While some recognition systems are based on a unified optical model, dealing with a unified language model remains a major issue, as traditional language models are generally trained on corpora composed of large word lexicons per language. Here, we bring a solution by considering language models based on sub-lexical units, called multigrams. Dealing with multigrams strongly reduces the lexicon size and thus decreases the language model complexity. This makes possible the design of an end-to-end unified multilingual recognition system where both a single optical model and a single language model are trained on all the languages. We discuss the impact of the language unification on each model and show that our system reaches state-of-the-art methods performance with a strong reduction of the complexity. [ABSTRACT FROM AUTHOR]

Highlights • An analysis of Cookie Theft narratives in English and Swedish is presented. • Multilingual word embeddings are clustered to generate multilingual topics. • Features extracted from the topic model help detect mild cognitive impairment. • Classification accuracy is 63% (English) and 72% (Swedish). • Multilingual topic models outperform monolingual models in both languages. Abstract We analyze the information content of narrative speech samples from individuals with mild cognitive impairment (MCI), in both English and Swedish, using a combination of supervised and unsupervised learning techniques. We extract information units using topic models trained on word embeddings in monolingual and multilingual spaces, and find that the multilingual approach leads to significantly better classification accuracies than training on the target language alone. In many cases, we find that augmenting the topic model training corpus with additional clinical data from a different language is more effective than training on additional monolingual data from healthy controls. Ultimately we are able to distinguish MCI speakers from healthy older adults with accuracies of up to 63% (English) and 72% (Swedish) on the basis of information content alone. We also compare our method against previous results measuring information content in Alzheimer's disease, and report an improvement over other topic-modeling approaches. Furthermore, our results support the hypothesis that subtle differences in language can be detected in narrative speech, even at the very early stages of cognitive decline, when scores on screening tools such as the Mini-Mental State Exam are still in the "normal" range. [ABSTRACT FROM AUTHOR]

VerbNet—the most extensive online verb lexicon currently available for English—has proved useful in supporting a variety of NLP tasks. However, its exploitation in multilingual NLP has been limited by the fact that such classifications are available for few languages only. Since manual development of VerbNet is a major undertaking, researchers have recently translated VerbNet classes from English to other languages. However, no systematic investigation has been conducted into the applicability and accuracy of such a translation approach across different, typologically diverse languages. Our study is aimed at filling this gap. We develop a systematic method for translation of VerbNet classes from English to other languages which we first apply to Polish and subsequently to Croatian, Mandarin, Japanese, Italian, and Finnish. Our results on Polish demonstrate high translatability with all the classes (96% of English member verbs successfully translated into Polish) and strong inter-annotator agreement, revealing a promising degree of overlap in the resultant classifications. The results on other languages are equally promising. This demonstrates that VerbNet classes have strong cross-lingual potential and the proposed method could be applied to obtain gold standards for automatic verb classification in different languages. We make our annotation guidelines and the six language-specific verb classifications available with this paper. [ABSTRACT FROM AUTHOR]

Training speech recognizer with under-resourced language data still proves difficult. Indonesian language is considered under-resourced because the lack of a standard speech corpus, text corpus, and dictionary. In this research, the efficacy of augmenting limited Indonesian speech training data with highly - resourced-language training data, such as English, to train Indonesian speech recogniz er was analyzed. The training was performed in form of shared-hidden-layer deep-neural-network (SHL-DNN) training. An SHL-DNN has language-independent hidden layers and can be pre-trained and trained using multilingual training data without any difference with a monolingual deep neural network. The SHL-DNN using Indonesian and English speech training data proved effective for decreasing word error rate (WER) in decoding Indonesian dictated-speech by achieving 3.82% absolute decrease compared to a monolingual Indonesian hidden Markov model using Gaussian mixture model emission (GMM-HMM). The case was confirmed when the SHL-DNN was also employed to decode Indonesian spontaneous-speech by achieving 4.19% absolute WER decrease. [ABSTRACT FROM AUTHOR]

Modern data-driven spoken language systems (SLS) require manual semantic annotation for training spoken language understanding parsers. Multilingual porting of SLS demands significant manual effort and language resources, as this manual annotation has to be replicated. Crowdsourcing is an accessible and cost-effective alternative to traditional methods of collecting and annotating data. The application of crowdsourcing to simple tasks has been well investigated. However, complex tasks, like cross-language semantic annotation transfer, may generate low judgment agreement and/or poor performance. The most serious issue in cross-language porting is the absence of reference annotations in the target language; thus, crowd quality control and the evaluation of the collected annotations is difficult. In this paper we investigate targeted crowdsourcing for semantic annotation transfer that delegates to crowds a complex task such as segmenting and labeling of concepts taken from a domain ontology; and evaluation using source language annotation. To test the applicability and effectiveness of the crowdsourced annotation transfer we have considered the case of close and distant language pairs: Italian–Spanish and Italian–Greek. The corpora annotated via crowdsourcing are evaluated against source and target language expert annotations. We demonstrate that the two evaluation references (source and target) highly correlate with each other; thus, drastically reduce the need for the target language reference annotations. [ABSTRACT FROM AUTHOR]

Multilingual training of neural networks has proven to be simple yet effective way to deal with multilingual training corpora. It allows to use several resources to jointly train a language independent representation of features, which can be encoded into low-dimensional feature set by embedding narrow bottleneck layer to the network. In this paper, we analyze such features on the task of spoken language recognition (SLR), focusing on practical aspects of training bottleneck networks and analyzing their integration in SLR. By comparing properties of mono and multilingual features we show the suitability of multilingual training for SLR. The state-of-the-art performance of these features is demonstrated on the NIST LRE09 database. [ABSTRACT FROM AUTHOR]

Creating links among online encyclopedia articles in different languages is crucial in the construction and integration of large multilingual knowledge bases. Most research to date has focused on linking among different language versions of Wikipedia, yet other large online encyclopedias in a variety of languages exist. In this work, we present a cross-language article-linking method using a bilingual topic model and translation features based on an SVM model to link articles in English Wikipedia and Chinese Baidu Baike, the most widely used Wiki-like encyclopedia in China. To evaluate our approach, we compile data sets from Baidu Baike articles and their corresponding English Wikipedia articles. The evaluation results show that our approach achieves at most 0.8158 in MRR, outperforming the baseline system by 0.1328 (+19.44%) in MRR. Our method does not heavily depend on linguistic characteristics, and it can be easily extended to generate cross-language article links among different online encyclopedias in other languages. [ABSTRACT FROM AUTHOR]

In many applications, observations are available with different views. This is, for example, the case with image-text classification, multilingual document classification or document classification on the web. In addition, unlabeled multiview examples can be easily acquired, but assigning labels to these examples is usually a time consuming task. We describe a multiview self-learning strategy which trains different voting classifiers on different views. The margin distributions over the unlabeled training data, obtained with each view-specific classifier are then used to estimate an upper-bound on their transductive Bayes error. Minimizing this upper-bound provides an automatic margin-threshold which is used to assign pseudo-labels to unlabeled examples. Final class labels are then assigned to these examples, by taking a vote on the pool of the previous pseudo-labels. New view-specific classifiers are then trained using the labeled and pseudo-labeled training data. We consider applications to image-text classification and to multilingual document classification. We present experimental results on the NUS-WIDE collection and on Reuters RCV1-RCV2 which show that despite its simplicity, our approach is competitive with other state-of-the-art techniques. [ABSTRACT FROM AUTHOR]

Grounded theory offers a useful method for gaining an understanding of the context of the user experience when searching on an experimental multilingual image retrieval system. Observational, cognitive and affective data were collected while users searched for images in a multilingual environment. Straussian grounded theory was used to identify the elements of the dynamic process of information searching behaviour. The stages in which the data were coded are outlined to show how users' thoughts and actions were integrated in the analysis and to present and visualise the emerging concepts as representative of the context of the user experience. [ABSTRACT FROM AUTHOR]

In this paper we focus on unsupervised sentiment analysis in Spanish. The lack of resources for languages other than English, as for example Spanish, adds more complexity to the task. However, we take advantage of some good already existing lexical resources. We have carried out several experiments using different unsupervised approaches in order to compare the different methodologies for solving the problem of the Spanish polarity classification in a corpus of movie reviews. Among all these approaches, perhaps the newest one integrates SentiWordNet with the Multilingual Central Repository to tackle polarity detection directly over the Spanish corpus. However, the results obtained were not as promising as we expected, and so we carried out another group of experiments combining all the methods using meta-classifiers. The results obtained with stacking outperformed the individual experiments and encourage us to continue in this way. [ABSTRACT FROM PUBLISHER]

Since we have been facing multilingual contents, it is difficult for recommender systems (RecSys) to efficiently collect user feedbacks (e.g., ratings). Thus, we expect that multilingual entities matching can improve the performance of recommendation services. Particularly, in movie recommendation services, the movies have several titles in different languages. Thereby, we are focusing on interlinking some possible data sources including traditional tabular data (e.g., IMDB) and Linked Open Data (LOD) (e.g., DBpedia and LinkedMDB). This paper shows meaningful experiences that we have observed during experimentation; i) discovering identical movies which have multilingual titles by interlinking LOD, and ii) improving the performance of multilingual recommendation. [ABSTRACT FROM AUTHOR]

This paper investigates the ability of multilingual BERT (mBERT) language model to transfer syntactic knowledge cross-lingually, verifying if and to which extent syntactic dependency relationships learnt in a language are maintained in other languages. In detail, the main contributions of this paper are: (i) an analysis of the cross-lingual syntactic transfer capability of mBERT model; (ii) a detailed comparison of cross-language syntactic transfer among languages belonging to different branches of the Indo-European languages, namely English, Italian and French, which present very different syntactic constructions; (iii) a study on the transferability of a syntactic phenomenon peculiar of Italian language, namely the pronoun dropping (pro-drop), also known as omissibility of the subject. To this end, a structural probe devoted to reconstruct the dependency parse tree of a sentence has been exploited, representing the input sentences with the contextual embeddings from mBERT layers. The results of the experimental assessment have shown a transfer of syntactic knowledge of the mBERT model among these languages. Moreover, the behaviour of the probe in the transition from pro-drop to non-pro-drop languages and vice versa has proven to be more effective in case of languages sharing a common linguistic matrix. The possibility of transferring syntactical knowledge, especially in the case of specific phenomena, meets both a theoretical need and can have important practical implications in syntactic tasks, such as dependency parsing. • An analysis of cross-lingual syntax transfer capability of multilingual BERT model. • Use of a structural probe to reconstruct the dependency parse tree of a sentence. • A cross-lingual test on an aligned and parallel annotated treebank. • Three different languages considered: English, Italian and French. • Study of cross-language transfer of omissibility of the subject syntactic phenomenon. [ABSTRACT FROM AUTHOR]

Highlights: [•] We study the possibility to employ Machine Translation (MT) systems and supervised methods for multilingual sentiment analysis. [•] Experiments are done for English, German, Spanish and French. [•] We use three MT systems – Google, Bing and Moses –, different supervised learning algorithms and various types of features. [•] We show how meta-classifiers can be employed to mitigate the noise introduced by translation. [•] Our extensive evaluations show that MT systems can be used for multilingual sentiment analysis. [Copyright &y& Elsevier]

Highlights: [•] A model to represent patent request by a set of concepts related to a multilingual knowledge ontology. [•] Analysis of patent extraction process from 104,296 patents from US patent office. [•] Analysis of multilingual decision process from Korean, US, and Chinese patent offices. [•] High recall and precision: increasing number of languages only has minor effects on model results. [Copyright &y& Elsevier]

This research article provides a summary of our approach in providing integrated multilingual access to diverse digital archives of Japanese Ukiyo-e prints. We propose a novel method, which generates links across databases using Linked Data. A retrieval system has been developed for realizing the proposed method, which dynamically creates links to miscellaneous Linked Data resources from authority data. By using the proposed method, users would allowed to access to additional data about a certain record in multiple databases not depending on languages and formats of each database. Experimental evaluations have conducted to measure reliability of the generated links. The experimental results show high accuracy rate and prove the effectiveness of the proposed method. Links to major Linked Data resources were created successfully. [ABSTRACT FROM AUTHOR]

Psycholinguistic research shows that key properties of the human sentence processor are incrementality, connectedness (partial structures contain no unattached nodes), and prediction (upcoming syntactic structure is anticipated). There is currently no broad-coverage parsing model with these properties, however. In this article, we present the first broad-coverage probabilistic parser for PLTAG, a variant of TAG that supports all three requirements. We train our parser on a TAG-transformed version of the Penn Treebank and show that it achieves performance comparable to existing TAG parsers that are incremental but not predictive. We also use our PLTAG model to predict human reading times, demonstrating a better fit on the Dundee eye-tracking corpus than a standard surprisal model. [ABSTRACT FROM AUTHOR]

Current investigations in data-driven models of parsing have shifted from purely syntactic analysis to richer semantic representations, showing that the successful recovery of the meaning of text requires structured analyses of both its grammar and its semantics. In this article, we report on a joint generative history-based model to predict the most likely derivation of a dependency parser for both syntactic and semantic dependencies, in multiple languages. Because these two dependency structures are not isomorphic, we propose a weak synchronization at the level of meaningful subsequences of the two derivations. These synchronized subsequences encompass decisions about the left side of each individual word. We also propose novel derivations for semantic dependency structures, which are appropriate for the relatively unconstrained nature of these graphs. To train a joint model of these synchronized derivations, we make use of a latent variable model of parsing, the Incremental Sigmoid Belief Network (ISBN) architecture. This architecture induces latent feature representations of the derivations, which are used to discover correlations both within and between the two derivations, providing the first application of ISBNs to a multi-task learning problem. This joint model achieves competitive performance on both syntactic and semantic dependency parsing for several languages. Because of the general nature of the approach, this extension of the ISBN architecture to weakly synchronized syntactic-semantic derivations is also an exemplification of its applicability to other problems where two independent, but related, representations are being learned. [ABSTRACT FROM AUTHOR]

Highlights: [•] The task of multi-lingual tweet generation is analyzed. [•] Headlines may not be the most suitable way to communicate information via Twitter. [•] Current tweets delivered by news providers are informative, but not interesting. [•] Automatic tweets were found to be indicative as well as interesting. [•] Overall, COMPENDIUM and the Language-specific summarizer were the best systems. [ABSTRACT FROM AUTHOR]

The author conveys his concerns on whether the "Oxford English Dictionary" (OED) have a philosophy in the sense of a series of guiding principles. Topics discussed include how the concept of incremental logical progression over time works for the English term "culture," compatibility of computational linguistics and online editing with the OED's philosophy of language, and a discussion on illustrative quotations and the risks of inexactitude.

Sentiment ambiguous adjectives, which have been neglected by most previous researches, pose a challenging task in sentiment analysis. We present an evaluation task at SemEval-2010, designed to provide a framework for comparing different approaches on this problem. The task focuses on 14 Chinese sentiment ambiguous adjectives, and provides manually labeled test data. There are 8 teams submitting 16 systems in this task. In this paper, we define the task, describe the data creation, list the participating systems, and discuss different approaches. [ABSTRACT FROM AUTHOR]

Abstract: It is commonly acknowledged that ever-increasing video archives should be conveniently indexed with the conveyed semantic information to facilitate later video retrieval. Domain-independent semantic video indexing is usually carried out through manual means which is too time-consuming and labor-intensive to be employed in practical settings. On the other hand, fully automated approaches are usually proposed for very specialized domains such as team sports videos. In this paper, we propose a generic text-based semi-automatic system for off-line semantic indexing and retrieval of news videos, since video texts such as speech transcripts stand as a plausible source of semantic information. The proposed system has a pipelined flow of execution where the sole manual intervention takes place during text extraction, yet it could execute in fully automated mode in case the associated video text is already available or a convenient text extractor is available to be incorporated into the system. At the core of the system is an information extraction component – a named entity recognizer – which extracts representative semantic information from the video texts. Based on the proposed generic system, a novel semantic annotation and retrieval system for Turkish is designed, implemented, and evaluated on two distinct news video data sets. By equipping it with the necessary components, the ultimate system is also turned into a multilingual video retrieval system and executed on a video data set in English, thereby facilitating multilingual semantic video retrieval. [Copyright &y& Elsevier]

The objective of this paper is to report on a comparative user evaluation of two multilingual thesaurus-enhanced visual user interfaces--namely, T-saurus and Searchling, designed and developed for digital libraries. The study used 25 academic users carrying out three search tasks on both user interfaces. It applied usability and affordance strength questionnaires, interviews, think-alouds, and direct observation to investigate users' evaluation of the key components of both user interfaces--namely, multilingual features, thesaurus and search functions, and visualization and visual appeal. Results of the study show that users were able to successfully carry out the search tasks using thesaurus-enhanced search interfaces. However, they preferred Searchling, a faceted search user interface, for its flexible language option, thesaurus browsing, and visualization. Users preferred the interface to show the thesaural relationships along with the selected term without interacting with the interface. The empirical data gathered and the design ideas implemented within the two user interfaces will be useful for designers of search interfaces that make use of thesaurus and multilingual features. The design and methodological framework of the developed user interfaces and the study could be used to create more information-rich multilingual academic search user interfaces for teaching and learning purposes. The design and representation of thesaural terms and structures in these two interfaces are designed using novel visualization techniques. [ABSTRACT FROM AUTHOR]

This paper deals with the problems of language detection and tracking in multilingual online short word-of-mouth (WoM) discussions. This problem is particularly unusual and difficult from a pattern recognition perspective because, in these discussions, the participants and content involve the opinions of users from all over the world. The nature of these discussions, consisting of multiple topics in different languages, presents us with a problem of finding training and classification strategies when the class-conditional distributions are nonstationary. The difficulties in solving the problem are many-fold. First of all, the analyst has no knowledge of when one language stops and when the next starts. Further, the features which one uses for any one language (for example, the n-grams) will not be valid to recognize another. Finally, and most importantly, in most real-life applications, such as in WoM, the fragments of text available before the switching, are so small that it renders any meaningful classification using traditional estimation methods almost futile. Earlier, the authors [B. J. Oommen and L. Rueda, Patt. Recogn.39(1) (2006) 328-341.] had recommended that for a variety of problems, the use of strong estimators (i.e. estimators that converge with probability 1) is sub-optimal. In this vein, we propose to solve the current problem using novel estimators that are pertinent for nonstationary environments. The classification results obtained for various data sets which involve as many as eight languages demonstrates that our proposed methodology is both powerful and efficient. [ABSTRACT FROM AUTHOR]

Abstract: The online encyclopedia Wikipedia is a vast, constantly evolving tapestry of interlinked articles. For developers and researchers it represents a giant multilingual database of concepts and semantic relations, a potential resource for natural language processing and many other research areas. This paper introduces the Wikipedia Miner toolkit, an open-source software system that allows researchers and developers to integrate Wikipediaʼs rich semantics into their own applications. The toolkit creates databases that contain summarized versions of Wikipediaʼs content and structure, and includes a Java API to provide access to them. Wikipediaʼs articles, categories and redirects are represented as classes, and can be efficiently searched, browsed, and iterated over. Advanced features include parallelized processing of Wikipedia dumps, machine-learned semantic relatedness measures and annotation features, and XML-based web services. Wikipedia Miner is intended to be a platform for sharing data mining techniques. [Copyright &y& Elsevier]

Abstract: We automatically create enormous, free and multilingual silver-standard training annotations for named entity recognition (ner) by exploiting the text and structure of Wikipedia. Most ner systems rely on statistical models of annotated data to identify and classify names of people, locations and organisations in text. This dependence on expensive annotation is the knowledge bottleneck our work overcomes. We first classify each Wikipedia article into named entity (ne) types, training and evaluating on 7200 manually-labelled Wikipedia articles across nine languages. Our cross-lingual approach achieves up to 95% accuracy. We transform the links between articles into ne annotations by projecting the target articleʼs classifications onto the anchor text. This approach yields reasonable annotations, but does not immediately compete with existing gold-standard data. By inferring additional links and heuristically tweaking the Wikipedia corpora, we better align our automatic annotations to gold standards. We annotate millions of words in nine languages, evaluating English, German, Spanish, Dutch and Russian Wikipedia-trained models against conll shared task data and other gold-standard corpora. Our approach outperforms other approaches to automatic ne annotation (Richman and Schone, 2008 [61], Mika et al., 2008 [46]) competes with gold-standard training when tested on an evaluation corpus from a different source; and performs 10% better than newswire-trained models on manually-annotated Wikipedia text. [Copyright &y& Elsevier]

Abstract: Many applications in the context of natural language processing have been proven to achieve a significant performance when exploiting semantic information extracted from high-quality annotated resources. However, the practical use of such resources is often biased by their limited coverage. Furthermore, they are generally available only for English and few other languages. We propose a novel methodology that, starting from the mapping between FrameNet lexical units and Wikipedia pages, automatically leverages from Wikipedia new lexical units and example sentences. The goal is to build a reference data set for the semi-automatic development of new FrameNets. In addition, this methodology can be adapted to perform frame identification in any language available in Wikipedia. Our approach relies on a state-of-the-art word sense disambiguation system that is first trained on English Wikipedia to assign a page to the lexical units in a frame. Then, this mapping is further exploited to perform frame identification in English or in any other language available in Wikipedia. Our approach shows a high potential in multilingual settings, because it can be applied to languages for which other lexical resources such as WordNet or thesauri are not available. [Copyright &y& Elsevier]

Abstract: A knowledge base for real-world language processing applications should consist of a large base of facts and reasoning mechanisms that combine them to induce novel and more complex information. This paper describes an approach to deriving such a large scale and multilingual resource by exploiting several facets of the on-line encyclopedia Wikipedia. We show how we can build upon Wikipediaʼs existing network of categories and articles to automatically discover new relations and their instances. Working on top of this network allows for added information to influence the network and be propagated throughout it using inference mechanisms that connect different pieces of existing knowledge. We then exploit this gained information to discover new relations that refine some of those found in the previous step. The result is a network containing approximately 3.7 million concepts with lexicalizations in numerous languages and 49+ million relation instances. Intrinsic and extrinsic evaluations show that this is a high quality resource and beneficial to various NLP tasks. [Copyright &y& Elsevier]

Globalization and multilingualism contribute to code-switching--the phenomenon in which speakers produce utterances containing words or expressions from a second language. Processing code-switched sentences is a significant challenge for multilingual intelligent systems. This study proposes a language modeling approach to the problem of code-switching language processing, dividing the problem into two subtasks: the detection of code-switched sentences and the identification of codeswitched words in sentences. A code-switched sentence is detected on the basis of whether it contains words or phrases from another language. Once the code-switched sentences are identified, the positions of the code-switched words in the sentences are then identified. Experimental results show that the language modeling approach achieved an F-measure of 80.43% and an accuracy of 79.01% for detecting Mandarin-Taiwanese code-switched sentences. For the identification of code-switched words, the word-based and POS-based models, respectively, achieved F-measures of 41.09% and 53.08%. [ABSTRACT FROM AUTHOR]

Abstract: We present an automatic approach to the construction of BabelNet, a very large, wide-coverage multilingual semantic network. Key to our approach is the integration of lexicographic and encyclopedic knowledge from WordNet and Wikipedia. In addition, Machine Translation is applied to enrich the resource with lexical information for all languages. We first conduct in vitro experiments on new and existing gold-standard datasets to show the high quality and coverage of BabelNet. We then show that our lexical resource can be used successfully to perform both monolingual and cross-lingual Word Sense Disambiguation: thanks to its wide lexical coverage and novel semantic relations, we are able to achieve state-of the-art results on three different SemEval evaluation tasks. [Copyright &y& Elsevier]

This paper proposes to advance in the current state-of-the-art of automatic Language Resource (LR) building by taking into consideration three elements: (1) the knowledge available in existing LRs, (2) the vast amount of information available from the collaborative paradigm that has emerged from the Web 2.0 and (3) the use of standards to improve interoperability. We present a case study in which a set of LRs for different languages (WordNet for English and Spanish and Parole-Simple-Clips for Italian) are extended with Named Entities (NE) by exploiting Wikipedia and the aforementioned LRs. The practical result is a multilingual NE lexicon connected to these LRs and to two ontologies: SUMO and SIMPLE. Furthermore, the paper addresses an important problem which affects the Computational Linguistics area in the present, interoperability, by making use of the ISO LMF standard to encode this lexicon. The different steps of the procedure (mapping, disambiguation, extraction, NE identification and postprocessing) are comprehensively explained and evaluated. The resulting resource contains 974,567, 137,583 and 125,806 NEs for English, Spanish and Italian respectively. Finally, in order to check the usefulness of the constructed resource, we apply it into a state-of-the-art Question Answering system and evaluate its impact; the NE lexicon improves the system's accuracy by 28.1%. Compared to previous approaches to build NE repositories, the current proposal represents a step forward in terms of automation, language independence, amount of NEs acquired and richness of the information represented. [ABSTRACT FROM AUTHOR]

This paper, proposes a novel approach for feature extraction based on the segmentation and morphological alteration of handwritten multi-lingual characters. We explored multi-resolution and multi-directional transforms such as wavelet, curvelet and ridgelet transform to extract classifying features of handwritten multi-lingual images. Evaluating the pros and cons of each multi-resolution algorithm has been discussed and resolved that Curvelet-based features extraction is most promising for multi-lingual character recognition. We have also applied some morphological operation such as thinning and thickening then feature level fusion is performed in order to create robust feature vector for classification. The classification is performed with K-nearest neighbor (K-NN) and support vector machine (SVM) classifier with their relative performance. We experiment with our in-house dataset, compiled in our lab by more than 50 personnel. [ABSTRACT FROM AUTHOR]

Abstract: The process of Natural Language Generation for a Conversational Agent translates some semantic language to its surface form expressed in natural language. In this paper, we are going to show a Case Based Reasoning technique which is easily extensible and adaptable to multiple domains and languages, that generates coherent phrases and produces a natural outcome in the context of a Conversational Agent that maintains a dialogue with the user. [Copyright &y& Elsevier]

Multilingual text processing is useful because the information content found in different languages is complementary, both regarding facts and opinions. While Information Extraction and other text mining software can, in principle, be developed for many languages, most text analysis tools have only been applied to small sets of languages because the development effort per language is large. Self-training tools obviously alleviate the problem, but even the effort of providing training data and of manually tuning the results is usually considerable. In this paper, we gather insights by various multilingual system developers on how to minimise the effort of developing natural language processing applications for many languages. We also explain the main guidelines underlying our own effort to develop complex text mining software for tens of languages. While these guidelines-most of all: extreme simplicity-can be very restrictive and limiting, we believe to have shown the feasibility of the approach through the development of the Europe Media Monitor (EMM) family of applications (). EMM is a set of complex media monitoring tools that process and analyse up to 100,000 online news articles per day in between twenty and fifty languages. We will also touch upon the kind of language resources that would make it easier for all to develop highly multilingual text mining applications. We will argue that-to achieve this-the most needed resources would be freely available, simple, parallel and uniform multilingual dictionaries, corpora and software tools. [ABSTRACT FROM AUTHOR]

Many existing studies have investigated to discover a variety of co-occurrence patterns between entities (e.g. users, tags and resources) from a folksonomy system. The common purposes among them are (i) to understand collective behaviors between online users and (ii) to provide online services (e.g. tag recommendation and information searching) to users. However, most of the existing studies assume that all tags in the folksonomy should be written in an identical language. In this paper, we focus on analyzing a multilingual folksonomy generated by various lingual practices of online users, and discovering meaningful relationships between multilingual tags (e.g. between ‘Seoul’ in English and ‘Corée’ in French) co-occurred in the folksonomy. Thereby, we propose novel methods for (i) identifying lingual practices from user tagging patterns to build a community of lingual practice and (ii) exploiting the tag matchings to extend simple term-based queries. Thus, additional resources tagged by other languages can be retrieved. To evaluate the proposed multilingual tag matching method, we have collected real tagging datasets from several well-known social tagging websites (e.g. Del.icio.us), and applied to translating queries to other languages without any external dictionaries. [ABSTRACT FROM AUTHOR]

This paper presents a document representation strategy and a bio-inspired algorithm to cluster multilingual collections of documents in the field of economics and business. The proposed approach allows the user to identify groups of related economics documents written in Spanish and English using techniques inspired on clustering and sorting behaviours observed in some types of ants. In order to obtain a language independent vector representation of each document two multilingual resources are used: an economic glossary and a thesaurus. Each document is represented using four feature vectors: words, proper names, economic terms in the glossary and thesaurus descriptors. The proper name identification, word extraction and lemmatization are performed using specific tools. The tf-idf scheme is used to measure the importance of each feature in the document, and a convex linear combination of angular separations between feature vectors is used as similarity measure of documents. The paper shows experimental results of the application of the proposed algorithm in a Spanish-English corpus of research papers in economics and management areas. The results demonstrate the usefulness and effectiveness of the ant clustering algorithm and the proposed representation scheme. [ABSTRACT FROM AUTHOR]

Abstract: Novelty detection aims at reducing redundant information from a chronologically ordered list of documents or sentences. Other studies of novelty detection have been conducted on the English language, but few papers have addressed the problem of multilingual novelty detection. Likewise, research in multilingual information retrieval have rarely been applied to novelty detection. This paper attempts to bridge the two disciplines by first describing the preprocessing steps for English, Malay and Chinese, then applying document and sentence-level novelty detection for the three languages on APWSJ and TREC 2004 Novelty Track data. Experiments on sentence-level novelty detection show similar results for all three languages, which indicates that our algorithm is suitable for multilingual novelty detection at the sentence level. However, results for document-level novelty detection show a disparity across the different languages, with English and Malay outperforming Chinese. After applying sentence-level novelty detection to detect novel documents, we observe substantial improvements on all three languages. This demonstrates that segmenting documents into sentences improves document-level novelty detection in multiple languages, and has practical benefits for a real-time multilingual novelty detection system. [Copyright &y& Elsevier]

The article discusses how the phenomenon of increased inflow of borrowings in natural language being observed within last twenty years influences formative system of subject headings languages. Internationalisation and effect of globalization are mentioned as a trend indicating the direction of changes in natural and subject heading languages. [ABSTRACT FROM AUTHOR]

Abstract: R&D project management systems are developed to facilitate the efficient management of R&D projects in government-funding agencies and universities. They should support R&D information sharing among researchers and research administrators from different nationalities using different languages. This paper presents an ontology-based methodology to the development of R&D project management systems with multilingual supports. In particular, a multilingual ontology is proposed, which consists of domain model, application model, user model and linguistic model. The methodology is used in the development of an R&D project management system for the Innovation and Technology Commission of the Hong Kong SAR Government. The system supports three languages and facilitates R&D information sharing among users with different culture backgrounds and usage preferences. [Copyright &y& Elsevier]

Abstract: This paper introduces a novel approach to the task of lexical translation between languages for which no translation dictionaries are available. We build a massive translation graph, automatically constructed from over 630 machine-readable dictionaries and Wiktionaries. In this graph each node denotes a word in some language and each edge denotes a word sense shared by and . Our current graph contains over 10,000,000 nodes and expresses more than 60,000,000 pairwise translations. The composition of multiple translation dictionaries leads to a transitive inference problem: if word A translates to word B which in turn translates to word C, what is the probability that C is a translation of A? The paper describes a series of probabilistic inference algorithms that solve this problem at varying precision and recall levels. All algorithms enable us to quantify our confidence in a translation derived from the graph, and thus trade precision for recall. We compile the results of our best inference algorithm to yield PanDictionary, a novel multilingual dictionary. PanDictionary contains more than four times as many translations as in the largest Wiktionary at precision 0.90 and over 200,000,000 pairwise translations in over 200,000 language pairs at precision 0.8. [Copyright &y& Elsevier]

Abstract: Due to the availability of a huge amount of textual data from a variety of sources, users of internationally distributed information regions need effective methods and tools that enable them to discover, retrieve and categorize relevant information, in whatever language and form it may have been stored. This drives a convergence of numerous interests from diverse research communities focusing on the issues related to multilingual text categorization. In this work, we implemented and measured the performance of the leading supervised and unsupervised approaches for multilingual text categorization. We selected support vector machines (SVM) as representative of supervised techniques as well as latent semantic indexing (LSI) and self-organizing maps (SOM) techniques as our selective ones of unsupervised methods for system implementation. The preliminary results show that our platform models including both supervised and unsupervised learning methods have the potentials for multilingual text categorization. [Copyright &y& Elsevier]

In this article, an evaluation of the behavioral change of groups collaborating towards the production of multilingual digital content is provided by using the data publication duration times as a proxy. In parallel, the assessment of feedback as a stimulant of social behavior in developing such systems is given. The evaluation is based on a system developed by a consortium of experts in the context of a European project towards the provision of an eServices platform on Organic Agriculture (OA). Group behavior of the partners involved is studied and analyzed using regression models for the extraction of trends and patterns while the results of their collaboration are discussed and valuable conclusions regarding the improvement of collaborative work over the Internet are presented. [ABSTRACT FROM AUTHOR]

• Real-world spoken data streams, e.g., from TV or radio, often contain utterances in several related languages. • A novel approach is presented allowing for online identification of the spoken language in this type of multilingual streams. • Our method utilizes a decoder and a language classifier fed by multilingual bottleneck features. • These features, as well as the classifier, make use of neural networks architecture that allow for learning long-term dependencies. • The entire online recognition scheme operates with an average latency of around 2.5 s. The accelerating flow of information we encounter around the world today makes many companies deploy speech recognition systems that, to an ever-growing extent, process data on-line rather than off-line. These systems, e.g., for real-time 24/7 broadcast transcription, often work with input-stream data containing utterances in more than one language. This multilingual data can correctly be transcribed in real-time only if the language used is identified with just a small latency for each input frame. For this purpose, a novel approach to on-line spoken language identification is proposed in this work. Its development is documented within a series of consecutive experiments starting in the off-line mode for 11 Slavic languages, going through artificially prepared multilingual data for the on-line scenario, and ending with real bilingual TV programs containing utterances in mutually similar Czech and Slovak. The resulting scheme that we propose operates frame-by-frame; it takes in a multilingual stream of speech frames and outputs a stream of the corresponding language labels. It utilizes a weighted finite-state transducer as a decoder, which smooths the output from a language classifier fed by multilingual and augmented bottleneck features. An essential factor from the accuracy point of view is that these features, as well as the classifier itself, are based on deep neural network architectures that allow the modeling of long-term time dependencies. The obtained results show that our scheme allows us to determine the language spoken in real-world bilingual TV shows with an average latency of around 2.5 seconds and with an increase in word error rate by a mere 2.9% over the reference 18.1% value yielded by using manually prepared language labels. [ABSTRACT FROM AUTHOR]