We report on the participation of the JRC Text Mining and Analysis Competence Centre (TMA-CC) in the BSNLP-2019 Shared Task, which focuses on named-entity recognition, lemmatisation and cross-lingual linking. We propose a hybrid system combining a rule-based approach and light ML techniques. We use multilingual lexical resources such as JRC-NAMES and BABELNET together with a named entity guesser to recognise names. In a second step, we combine known names with wild cards to increase recognition recall by also capturing inflection variants. In a third step, we increase precision by filtering these name candidates with automatically learnt inflection patterns derived from name occurrences in large news article collections. Our major requirement is to achieve high precision. We achieved an average of 65% F-measure with 93% precision on the four languages.

We work on detecting positive or negative sentiment towards named entities in very large volumes of news articles. The aim is to monitor changes over time, as well as to work towards media bias detection by com-paring differences across news sources and countries. With view to applying the same method to dozens of languages, we use lin-guistically light-weight methods: searching for positive and negative terms in bags of words around entity mentions (also consid-ering negation). Evaluation results are good and better than a third-party baseline sys-tem, but precision is not sufficiently high to display the results publicly in our multilin-gual news analysis system Europe Media Monitor (EMM). In this paper, we focus on describing our effort to improve the English language results by avoiding the biggest sources of errors. We also present new work on using a syntactic parser to identify safe opinion recognition rules, such as predica-tive structures in which sentiment words di-rectly refer to an entity. The precision of this method is good, but recall is very low.

This paper describes an approach for the classification of millions of existing multi-word entities (MWEntities), such as organisation or event names, into thirteen category types, based only on the tokens they contain. In order to classify our very large in-house collection of multilingual MWEntities into an application-oriented set of entity categories, we trained and tested distantly-supervised classifiers in 43 languages based on MWEntities extracted from BabelNet. The best-performing classifier was the multi-class SVM using a TF.IDF-weighted data representation. Interestingly, one unique classifier trained on a mix of all languages consistently performed better than classifiers trained for individual languages, reaching an averaged F1-value of 88.8%. In this paper, we present the training and test data, including a human evaluation of its accuracy, describe the methods used to train the classifiers, and discuss the results.

This paper reports on an approach and experiments to automatically build a cross-lingual multi-word entity resource. Starting from a collection of millions of acronym/expansion pairs for 22 languages where expansion variants were grouped into monolingual clusters, we experiment with several aggregation strategies to link these clusters across languages. Aggregation strategies make use of string similarity distances and translation probabilities and they are based on vector space and graph representations. The accuracy of the approach is evaluated against Wikipedia’s redirection and cross-lingual linking tables. The resulting multi-word entity resource contains 64,000 multi-word entities with unique identifiers and their 600,000 multilingual lexical variants. We intend to make this new resource publicly available.

Various recent studies show that the performance of named entity recognition (NER) systems developed for well-formed text types drops significantly when applied to tweets. The only existing study for the highly inflected agglutinative language Turkish reports a drop in F-Measure from 91% to 19% when ported from news articles to tweets. In this study, we present a new named entity-annotated tweet corpus and a detailed analysis of the various tweet-specific linguistic phenomena. We perform comparative NER experiments with a rule-based multilingual NER system adapted to Turkish on three corpora: a news corpus, our new tweet corpus, and another tweet corpus. Based on the analysis and the experimentation results, we suggest system features required to improve NER results for social media like Twitter.

The Europe Media Monitor (EMM) is a fully-automatic system that analyses written online news by gathering articles in over 70 languages and by applying text analysis software for currently 21 languages, without using linguistic tools such as parsers, part-of-speech taggers or morphological analysers. In this paper, we describe the effort of adding to EMM Hungarian text mining tools for news gathering; document categorisation; named entity recognition and classification for persons, organisations and locations; name lemmatisation; quotation recognition; and cross-lingual linking of related news clusters. The major challenge of dealing with the Hungarian language is its high degree of inflection and agglutination. We present several experiments where we apply linguistically light-weight methods to deal with inflection and we propose a method to overcome the challenges. We also present detailed frequency lists of Hungarian person and location name suffixes, as found in real-life news texts. This empirical data can be used to draw further conclusions and to improve existing Named Entity Recognition software. Within EMM, the solutions described here will also be applied to other morphologically complex languages such as those of the Slavic language family. The media monitoring and analysis system EMM is freely accessible online via the web page http://emm.newsbrief.eu/overview.html.

Multi-word entities, such as organisation names, are frequently written in many different ways. We have previously automatically identified over one million acronym pairs in 22 languages, consisting of their short form (e.g. EC) and their corresponding long forms (e.g. European Commission, European Union Commission). In order to automatically group such long form variants as belonging to the same entity, we cluster them, using bottom-up hierarchical clustering and pair-wise string similarity metrics. In this paper, we address the issue of how to evaluate the named entity variant clusters automatically, with minimal human annotation effort. We present experiments that make use of Wikipedia redirection tables and we show that this method produces good results.

We are presenting a new highly multilingual document-aligned parallel corpus called DCEP - Digital Corpus of the European Parliament. It consists of various document types covering a wide range of subject domains. With a total of 1.37 billion words in 23 languages (253 language pairs), gathered in the course of ten years, this is the largest single release of documents by a European Union institution. DCEP contains most of the content of the European Parliament’s official Website. It includes different document types produced between 2001 and 2012, excluding only the documents already exist in the Europarl corpus to avoid overlapping. We are presenting the typical acquisition steps of the DCEP corpus: data access, document alignment, sentence splitting, normalisation and tokenisation, and sentence alignment efforts. The sentence-level alignment is still in progress but based on some first experiments; we showed that DCEP is very useful for NLP applications, in particular for Statistical Machine Translation.

This paper presents an evaluation of the use of machine translation to obtain and employ data for training multilingual sentiment classifiers. We show that the use of machine translated data obtained similar results as the use of native-speaker translations of the same data. Additionally, our evaluations pinpoint to the fact that the use of multilingual data, including that obtained through machine translation, leads to improved results in sentiment classification. Finally, we show that the performance of the sentiment classifiers built on machine translated data can be improved using original data from the target language and that even a small amount of such texts can lead to significant growth in the classification performance.

The European Commission's (EC) Directorate General for Translation, together with the EC's Joint Research Centre, is making available a large translation memory (TM; i.e. sentences and their professionally produced translations) covering twenty-two official European Union (EU) languages and their 231 language pairs. Such a resource is typically used by translation professionals in combination with TM software to improve speed and consistency of their translations. However, this resource has also many uses for translation studies and for language technology applications, including Statistical Machine Translation (SMT), terminology extraction, Named Entity Recognition (NER), multilingual classification and clustering, and many more. In this reference paper for DGT-TM, we introduce this new resource, provide statistics regarding its size, and explain how it was produced and how to use it.

EuroVoc (2012) is a highly multilingual thesaurus consisting of over 6,700 hierarchically organised subject domains used by European Institutions and many authorities in Member States of the European Union (EU) for the classification and retrieval of official documents. JEX is JRC-developed multi-label classification software that learns from manually labelled data to automatically assign EuroVoc descriptors to new documents in a profile-based category-ranking task. The JEX release consists of trained classifiers for 22 official EU languages, of parallel training data in the same languages, of an interface that allows viewing and amending the assignment results, and of a module that allows users to re-train the tool on their own document collections. JEX allows advanced users to change the document representation so as to possibly improve the categorisation result through linguistic pre-processing. JEX can be used as a tool for interactive EuroVoc descriptor assignment to increase speed and consistency of the human categorisation process, or it can be used fully automatically. The output of JEX is a language-independent EuroVoc feature vector lending itself also as input to various other Language Technology tasks, including cross-lingual clustering and classification, cross-lingual plagiarism detection, sentence selection and ranking, and more.

We present a fully functional Arabic information extraction (IE) system that is used to analyze large volumes of news texts every day to extract the named entity (NE) types person, organization, location, date and number, as well as quotations (direct reported speech) by and about people. The Named Entity Recognition (NER) system was not developed for Arabic, but - instead - a highly multilingual, almost language-independent NER system was adapted to also cover Arabic. The Semitic language Arabic substantially differs from the Indo-European and Finno-Ugric languages currently covered. This paper thus describes what Arabic language-specific resources had to be developed and what changes needed to be made to the otherwise language-independent rule set in order to be applicable to the Arabic language. The achieved evaluation results are generally satisfactory, but could be improved for certain entity types. The results of the IE tools can be seen on the Arabic pages of the freely accessible Europe Media Monitor (EMM) application NewsExplorer, which can be found at http://press.jrc.it/overview.html.

Recent years have brought a significant growth in the volume of research in sentiment analysis, mostly on highly subjective text types (movie or product reviews). The main difference these texts have with news articles is that their target is clearly defined and unique across the text. Following different annotation efforts and the analysis of the issues encountered, we realised that news opinion mining is different from that of other text types. We identified three subtasks that need to be addressed: definition of the target; separation of the good and bad news content from the good and bad sentiment expressed on the target; and analysis of clearly marked opinion that is expressed explicitly, not needing interpretation or the use of world knowledge. Furthermore, we distinguish three different possible views on newspaper articles ― author, reader and text, which have to be addressed differently at the time of analysing sentiment. Given these definitions, we present work on mining opinions about entities in English language news, in which we apply these concepts. Results showed that this idea is more appropriate in the context of news opinion mining and that the approaches taking this into consideration produce a better performance.

We present a new, unique and freely available parallel corpus containing European Union (EU) documents of mostly legal nature. It is available in all 20 official EU languages, with additional documents being available in the languages of the EU candidate countries. The corpus consists of almost 8,000 documents per language, with an average size of nearly 9 million words per language. Pair-wise paragraph alignment information produced by two different aligners (Vanilla and HunAlign) is available for all 190+ language pair combinations. Most texts have been manually classified according to the EUROVOC subject domains so that the collection can also be used to train and test multi-label classification algorithms and keyword-assignment software. The corpus is encoded in XML, according to the Text Encoding Initiative Guidelines. Due to the large number of parallel texts in many languages, the JRC-Acquis is particularly suitable to carry out all types of cross-language research, as well as to test and benchmark text analysis software across different languages (for instance for alignment, sentence splitting and term extraction).

We are presenting a method to recognise geographical references in free text. Our tool must work on various languages with a minimum of language-dependent resources, except a gazetteer. The main difficulty is to disambiguate these place names by distinguishing places from persons and by selecting the most likely place out of a list of homographic place names world-wide. The system uses a number of language-independent clues and heuristics to disambiguate place name homographs. The final aim is to index texts with the countries and cities they mention and to automatically visualise this information on geographical maps using various tools.