Refine
Year of publication
Document Type
- Conference Proceeding (100)
- Part of a Book (23)
- Article (17)
- Book (5)
- Contribution to a Periodical (3)
- Doctoral Thesis (3)
- Master's Thesis (2)
- Other (2)
- Preprint (2)
- Working Paper (2)
Language
- English (161) (remove)
Keywords
- Computerlinguistik (161) (remove)
Publicationstate
- Veröffentlichungsversion (98)
- Zweitveröffentlichung (22)
- Postprint (20)
- Preprint (1)
Reviewstate
Publisher
- Association for Computational Linguistics (21)
- Springer (14)
- European Language Resources Association (10)
- LiU Electronic Press (6)
- European Language Resources Association (ELRA) (5)
- CLARIN (4)
- Incoma Ltd. (3)
- Linköping University Electronic Press (3)
- Oxford University Press (3)
- The Association for Computational Linguistics (3)
The availability of electronic corpora of historical stages of languages has been wel- comed as possibly attenuating the inherent problem of diachronic linguistics, i.e. that we only have access to what has chanced to come down to us - the problem which was memorably named by Labov (1992) as one of “Bad Data”. However, such corpora can only give us access to an increased amount ot historical material and this can essentially still only be a partial and possibly distorted picture of the actual language at a particular period of history. Corpora can be improved by taking a more representative sample of extant texts if these are available (as they are in significant number for periods after the invention of printing). But, as examples from the recently compiled GerManC corpus of seventeenth and eighteenth century German show, the evidence from such corpora can still fail to yield definitive answers to our questions about earlier stages of a language. The data still require expert interpretation, and it is important to be realistic about what can legitimately be expected from an electronic historical corpus.
A constructicon, i.e., a structured inventory of constructions, essentially aims at documenting functions of lexical and grammatical constructions. Among other parameters, so-called constructional collo-profiles, as introduced by Herbst (2018, 2020), are conclusive for determining constructional meanings. They provide information on how relevant individual words are for construction slots, they hint at usage preferences of constructions and serve as a helpful indicator for semantic peculiarities of constructions. However, even though collo-profiles constitute an indispensable component of constructicon entries, they pose major challengers for constructicographers: For a constructicographic enterprise it is not feasible to conduct collostructional analyses for hundreds or even thousands of constructions. In this article, we introduce a procedure based on the large language model BERT that allows to predict collo-profiles without having to extensively annotate instances of constructions in a given corpus. Specifically, by discussing the constructions X macht Y ADJP (‘x makes Y ADJ’, e.g. he drives him crazy) and N1 PREP N1 (e.g., bumper to bumper, constructions over constructions), we show how the developed automated system generates collo-profiles based on a limited number of annotated instances. Finally, we place collo-profiles alongside other dimensions of constructional meanings included in the German Constructicon.
The Data Governance Act was proposed in late 2020 as part of the European Strategy for Data, and adopted on 30 May 2022 (as Regulation 2022/868). It will enter into application on 24 September 2023. The Data governance Act is a major development in the legal framework affecting CLARIN and the whole language community. With its new rules on the re-use of data held by the public sector bodies and on the provision of data sharing services, and especially its encouragement of data altruism, the Data Governance Act creates new opportunities and new challenges for CLARIN ERIC. This paper analyses the provisions of the Data Governance Act, and aims at initiating the debate on how they will impact CLARIN and the whole language community.
CLARIAH-DE cross-service search - prospects and benefits of merging subject-specific services
(2021)
CLARIAH-DE combines services and offerings of CLARIN-D and DARIAH-DE. This includes various search applications which are made directly available to researchers. These search applications are presented in this working paper based on their main characteristics and compared with a focus on possible harmonizations. Opportunities and risks of different forms of technical integration are highlighted. Identified challenges can be explained in particular considering the background of different organizational and technical frameworks as well as highly specific and discipline-dependent requirements. The integration work that has already been carried out and the experiences gained with regard to future work and possible integration of further applications are also discussed. The experiences made in CLARIAH-DE can especially be of interest for other projects in the field of digital research infrastructures.
In order to differentiate between figurative and literal usage of verb-noun combinations for the shared task on the disambiguation of German Verbal Idioms issued for KONVENS 2021, we apply and extend an approach originally developed for detecting idioms in a dataset consisting of random ngram samples. The classification is done by implementing a rather shallow, statistics-based pipeline without intensive preprocessing and examinations on the morphosyntactic and semantic level. We describe the overall approach, the differences between the original dataset and the dataset of the KONVENS task, provide experimental classification results, and analyse the individual contributions of our feature sets.
The 12th Web as Corpus workshop (WAC-XII) looks at the past, present, and future of web corpora given the fact that large web corpora are nowadays provided mostly by a few major initiatives and companies, and the diversity of the early years appears to have faded slightly. Also, we acknowledge the fact that alternative sources of data (such as data from Twitter and similar platforms) have emerged, some of them only available to large companies and their affiliates, such as linguistic data from social media and other forms of the deep web. At the same time, gathering interesting and relevant web data (web crawling) is becoming an ever more intricate task as the nature of the data offered on the web changes (for example the death of forums in favour of more closed platforms).
In a recent paper published in the Journal of Language Evolution, Kauhanen, Einhaus & Walkden (KEW) challenge the results presented in one of my papers (Koplenig, Royal Society Open Science, 6, 181274 (2019)), in which I tried to show through a series of statistical analyses that large numbers of L2 (second language) speakers do not seem to affect the (grammatical or statistical) complexity of a language. To this end, I focus on the way in which the Ethnologue assesses language status: a language is characterised as vehicular if, in addition to being used by L1 (first language) speakers, it should also have a significant number of L2 users. KEW criticise both the use of vehicularity as a (binary) indicator of whether a language has a significant number of L2 users and the idea of imputing a zero proportion of L2 speakers to non-vehicular languages whenever a direct estimate of that proportion is unavailable. While I recognise the importance of post-publication commentary on published research, I show in this rejoinder that both points of criticism are explicitly mentioned and analysed in my paper. In addition, I also comment on other points raised by KEW and demonstrate that both alternative analyses offered by KEW do not stand up to closer scrutiny.
L’article intitulé «Traitement de l’information: Spinfo, HKI et humanités numériques - l’expérience de Cologne» présente l’histoire du développement des humanités numériques au sein de l’Université de Cologne. L'institutionnalisation des humanités numériques a commencé encore à l’époque où dans le monde germanophone le périmètre de la discipline était en train d’être défini par les travaux de quelques pionniers. Parmi eux, il convient de souligner le rôle d’Elisabeth Burr, active notamment à Tubingue, Duisbourg, Brême et Leipzig.L’article retrace le développement des humanités numériques à Cologne à partir de leurs débuts dans les années soixante du 20ème siècle, en passant par leur consolidation dans les années quatre-vingt-dix, jusqu’aux deux dernières décennies, quand Cologne est devenu un centre important de cette discipline. Le processus illustre comment une nouvelle discipline scientifique peut s’institutionnaliser au sein d’une université allemande. L’article décrit la perspective de deux domaines fondateurs: le traitement linguistique de l’information (en allemand: Sprachliche Informationsverarbeitung, Spinfo) et le traitement historico-culturel de l’information (en allemand: Historisch Kulturwissenschaftliche Informationsverarbeitung, HKI) et leur synthèse, qui a abouti en 2017 à la création de l’Institut des Humanités Numériques (Digital Humanities), qui aujourd’hui est - du point de vue interne - une composante de la Faculté de Philosophie de l’Université de Cologne et - du point de vue externe - une partie intégrante de la communauté internationale des humanités numériques.
So far, there have been few descriptions on creating structures capable of storing lexicographic data, ISO 24613:2008 being one of the latest. Another one is by Spohr (2012), who designs a multifunctional lexical resource which is able to store data of different types of dictionaries in a user-oriented way. Technically, his design is based on the principle of a hierarchical XML/OWL (eXtensible Markup Language/Web Ontology Language) representation model. This article follows another route in describing a model based on entities and relations between them; MySQL (usually referred to as: Structured Query Language) describes a database system of tables containing data and definitions of relations between them. The model was developed in the context of the project "Scientific eLexicography for Africa" and the lexicographic database to be built thereof will be implemented with MySQL. The principles of the ISO model and of Spohr's model are adhered to with one major difference in the implementation strategy: we do not place the lemma in the centre of attention, but the sense description — all other elements, including the lemma, depend on the sense description. This article also describes the contained lexicographic data sets and how they have been collected from different sources. As our aim is to compile several prototypical internet dictionaries (a monolingual Northern Sotho dictionary, a bilingual learners' Xhosa–English dictionary and a bilingual Zulu–English dictionary), we describe the necessary microstructural elements for each of them and which principles we adhere to when designing different ways of accessing them. We plan to make the model and the (empty) database with all graphical user interfaces that have been developed, freely available by mid-2015.
In this article, we examine the current situation of data dissemination and provision for CMC corpora. By that we aim to give a guiding grid for future projects that will improve the transparency and replicability of research results as well as the reusability of the created resources. Based on the FAIR guiding principles for research data management, we evaluate the 20 European CMC corpora listed in the CLARIN CMC Resource family, individuate successful strategies among the existing corpora and establish best practices for future projects. We give an overview of existing approaches to data referencing, dissemination and provision in European CMC corpora, and discuss the methods, formats and strategies used. Furthermore, we discuss the need for community standards and offer recommendations for best practices when creating a new CMC corpus.
In this Paper, we describe a schema and models which have been developed for the representation of corpora of computer-mediated communicatin (CMC corpora) using the representation framework provided by the Text Encoding Initiative (TEI). We characterise CMC discourse as dialogic, sequentially organised interchange between humans and point out that many features of CMC are not adequately handled by current corpus encoding schemas and tools. We formulate desiderata for a representation of CMC in encoding schemes and argue why the TEI is a suitable framework for the encoding of CMC corpora. We propose a model of basic CMC units (utterances, posts, and nonverbal activities) and the macro- and micro-level structures of interactions in CMC environments. Based on these models, we introduce CMC-core, a TEI customisation for the encoding of CMC corpora, which defines CMC-specific encoding features on the four levels of elements, model classes, attribute classes, and modules of the TEI infrastructure. The description of our customisation is illustrated by encoding examples from corpora by researchers of the TEI SIG CMC, representing a variety of CMC genres, i.e. chat, wiki talk, twitter, blog, and Second Life interactions. The material described, i.e. schemata, encoding examples, and documentation, is available from the of the TEI CMC SIG Wiki and will accompany a feature request to the TEI council in late 2019.
The QUEST (QUality ESTablished) project aims at ensuring the reusability of audio-visual datasets (Wamprechtshammer et al., 2022) by devising quality criteria and curating processes. RefCo (Reference Corpora) is an initiative within QUEST in collaboration with DoReCo (Documentation Reference Corpus, Paschen et al. (2020)) focusing on language documentation projects. Previously, Aznar and Seifart (2020) introduced a set of quality criteria dedicated to documenting fieldwork corpora. Based on these criteria, we establish a semi-automatic review process for existing and work-in-progress corpora, in particular for language documentation. The goal is to improve the quality of a corpus by increasing its reusability. A central part of this process is a template for machine-readable corpus documentation and automatic data verification based on this documentation. In addition to the documentation and automatic verification, the process involves a human review and potentially results in a RefCo certification of the corpus. For each of these steps, we provide guidelines and manuals. We describe the evaluation process in detail, highlight the current limits for automatic evaluation and how the manual review is organized accordingly.
Metadata provides important information relevant both to finding and understanding corpus data. Meaningful linguistic data requires both reasonable annotations and documentation of these annotations. This documentation is part of the metadata of a dataset. While corpus documentation has often been provided in the form of accompanying publications, machinereadable metadata, both containing the bibliographic information and documenting the corpus data, has many advantages. Metadata standards allow for the development of common tools and interfaces. In this paper I want to add a new perspective from an archive’s point of view and look at the metadata provided for four learner corpora and discuss the suitability of established standards for machine-readable metadata. I am are aware that there is ongoing work towards metadata standards for learner corpora. However, I would like to keep the discussion going and add another point of view: increasing findability and reusability of learner corpora in an archiving context.
Automatic summarization systems usually are trained and evaluated in a particular domain with fixed data sets. When such a system is to be applied to slightly different input, labor- and cost-intensive annotations have to be created to retrain the system. We deal with this problem by providing users with a GUI which allows them to correct automatically produced imperfect summaries. The corrected summary in turn is added to the pool of training data. The performance of the system is expected to improve as it adapts to the new domain.
This thesis is a corpus linguistic investigation of the language used by young German speakers online, examining lexical, morphological, orthographic, and syntactic features and changes in language use over time. The study analyses the language in the Nottinghamer Korpus deutscher YouTube‐Sprache ("Nottingham corpus of German YouTube language", or NottDeuYTSch corpus), one of the first large corpora of German‐language comments taken from the videosharing website YouTube, and built specifically for this project. The metadatarich corpus comprises c.33 million tokens from more than 3 million comments posted underneath videos uploaded by mainstream German‐language youthorientated YouTube channels from 2008‐2018.
The NottDeuYTSch corpus was created to enable corpus linguistic approaches to studying digital German youth language (Jugendsprache), having identified the need for more specialised web corpora (see Barbaresi 2019). The methodology for compiling the corpus is described in detail in the thesis to facilitate future construction of web corpora. The thesis is situated at the intersection of Computer‐Mediated Communication (CMC) and youth language, which have been important areas of sociolinguistic scholarship since the 1980s, and explores what we can learn from a corpus‐driven, longitudinal approach to (online) youth language. To do so, the thesis uses corpus linguistic methods to analyse three main areas:
1. Lexical trends and the morphology of polysemous lexical items. For this purpose, the analysis focuses on geil, one of the most iconic and productive words in youth language, and presents a longitudinal analysis, demonstrating that usage of geil has decreased, and identifies lexical items that have emerged as potential replacements. Additionally, geil is used to analyse innovative morphological productiveness, demonstrating how different senses of geil are used as a base lexeme or affixoid in compounding and derivation.
2. Syntactic developments. The novel grammaticalization of several subordinating conjunctions into both coordinating conjunctions and discourse markers is examined. The investigation is supported by statistical analyses that demonstrate an increase in the use of non‐standard syntax over the timeframe of the corpus and compares the results with other corpora of written language.
3. Orthography and the metacommunicative features of digital writing. This analysis identifies orthographic features and strategies in the corpus, e.g. the repetition of certain emoji, and develops a holistic framework to study metacommunicative functions, such as the communication of illocutionary force, information structure, or the expression of identities. The framework unifies previous research that had focused on individual features, integrating a wide range of metacommunicative strategies within a single, robust system of analysis.
By using qualitative and computational analytical frameworks within corpus linguistic methods, the thesis identifies emergent linguistic features in digital youth language in German and sheds further light on lexical and morphosyntactic changes and trends in the language of young people over the period 2008‐2018. The study has also further developed and augmented existing analytical frameworks to widen the scope of their application to orthographic features associated with digital writing.
Preface
(2020)
Learning new languages has a high relevance in today’s society with a globalized economy and the freedom to move abroad for work, study or other reasons. In this context new methods to teach and learn languages with the help of modern technology are becoming more relevant besides traditional language classes.
This work presents a new approach to combine a traditional language class with a modern computer-based approach for teaching. As a concrete example a web application to help teach and learn Latin was developed.
Preface
(2019)
We address the task of distinguishing implicitly abusive sentences on identity groups (“Muslims contaminate our planet”) from other group-related negative polar sentences (“Muslims despise terrorism”). Implicitly abusive language are utterances not conveyed by abusive words (e.g. “bimbo” or “scum”). So far, the detection of such utterances could not be properly addressed since existing datasets displaying a high degree of implicit abuse are fairly biased. Following the recently-proposed strategy to solve implicit abuse by separately addressing its different subtypes, we present a new focused and less biased dataset that consists of the subtype of atomic negative sentences about identity groups. For that task, we model components that each address one facet of such implicit abuse, i.e. depiction as perpetrators, aspectual classification and non-conformist views. The approach generalizes across different identity groups and languages.
In this paper we present work in developing a computerized grammar for the Latin language. It demonstrates the principles and challenges in developing a grammar for a natural language in a modern grammar formalism. The grammar presented here provides a useful resource for natural language processing applications in different fields. It can be easily adopted for language learning and use in language technology for Cultural Heritage like translation applications or to support post-correction of document digitization.
Content
1 Predicting learner knowledge of individual words using machine learning
Drilon Avdiu, Vanessa Bui, Klára Ptacinová Klimci´ková
2 Automatic Generation and Semantic Grading of Esperanto Sentences in a Teaching Context
Eckhard Bick
3 Toward automatic improvement of language produced by non-native language learners
Mathias Creutz, Eetu Sjöblom
4 Linguistic features and proficiency classification in L2 Spanish and L2 Portuguese
Iria del Ri´o
5 Integrating large-scale web data and curated corpus data in a search engine supporting German literacy education
Sabrina Dittrich, Zarah Weiss, Hannes Schröter, Detmar Meurers
6 Formalism for a language agnostic language learning game and productive grid generation
Sylvain Hatier, Arnaud Bey, Mathieu Loiseau
7 Understanding Vocabulary Growth Through An Adaptive Language Learning System
Elma Kerz, Andreas Burgdorf, Daniel Wiechmann, Stefan Meeger,Yu Qiao, Christian Kohlschein, Tobias Meisen
8 Summarization Evaluation meets Short-Answer Grading
Margot Mieskes, Ulrike Padó
9 Experiments on Non-native Speech Assessment and its Consistency
Ziwei Zhou, Sowmya Vajjala, Seyed Vahid Mirnezami
10 The Impact of Spelling Correction and Task Context on Short Answer Assessment for Intelligent Tutoring Systems
Ramon Ziai, Florian Nuxoll, Kordula De Kuthy, Björn Rudzewitz, Detmar Meurers
Content
1 Substituto - A Synchronous Educational Language Game for Simultaneous Teaching and Crowdsourcing
Marianne Grace Araneta, Gülsen Eryigit, Alexander König, Ji-Ung Lee, Ana Luís, Verena Lyding, Lionel Nicolas, Christos Rodosthenous and Federico Sangati
2 The Teacher-Student Chatroom Corpus
Andrew Caines, Helen Yannakoudakis, Helena Edmondson, Helen Allen, Pascual Pérez-Paredes, Bill Byrne and Paula Buttery
3 Polygloss - A conversational agent for language practice
Etiene da Cruz Dalcol and Massimo Poesio
4 Show, Don’t Tell: Visualising Finnish Word Formation in a Browser-Based Reading Assistant
Frankie Robertson
MULLE is a tool for language learning that focuses on teaching Latin as a foreign language. It is aimed for easy integration into the traditional classroom setting and syllabus, which makes it distinct from other language learning tools that provide standalone learning experience. It uses grammar-based lessons and embraces methods of gamification to improve the learner motivation. The main type of exercise provided by our application is to practice translation, but it is also possible to shift the focus to vocabulary or morphology training.
In this paper, we investigate the practical applicability of Co-Training for the task of building a classifier for reference resolution. We are concerned with the question if Co-Training can significantly reduce the amount of manual labeling work and still produce a classifier with an acceptable performance.
We present a language learning application that relies on grammars to model the learning outcome. Based on this concept we can provide a powerful framework for language learning exercises with an intuitive user interface and a high reliability. Currently the application aims to augment existing language classes and support students by improving the learner attitude and the general learning outcome. Extensions beyond that scope are promising and likely to be added in the future.
In this paper we investigate the problem of grammar inference from a different perspective. The common approach is to try to infer a grammar directly from example sentences, which either requires a large training set or suffers from bad accuracy. We instead view it as a problem of grammar restriction or sub-grammar extraction. We start from a large-scale resource grammar and a small number of examples, and find a sub-grammar that still covers all the examples. To do this we formulate the problem as a constraint satisfaction problem, and use an existing constraint solver to find the optimal grammar. We have made experiments with English, Finnish, German, Swedish and Spanish, which show that 10–20 examples are often sufficient to learn an interesting domain grammar. Possible applications include computer-assisted language learning, domain-specific dialogue systems, computer games, Q/A-systems, and others.
Controlled Natural Languages (CNLs) have many applications including document authoring, automatic reasoning on texts and reliable machine translation, but their application is not limited to these areas. We explore a new application area of CNLs, the use of CNLs in computer-assisted language learning. In this paper we present a a web application for language learning using CNLs as well as a detailed description of the properties of the family of CNLs it uses.
We present a light-weight tool for the annotation of linguistic data on multiple levels. It is based on the simplification of annotations to sets of markables having attributes and standing in certain relations to each other. We describe the main features of the tool, emphasizing its simplicity, customizability and versatility
We present an implemented XML data model and a new, simplified query language for multi-level annotated corpora. The new query language involves automatic conversion of queries into the underlying, more complicated MMAXQL query language. It supports queries for sequential and hierarchical, but also associative (e.g. coreferential) relations. The simplified query language has been designed with non-expert users in mind.
We present a supervised machine learning AND system which tackles semantic similarity between publication titles by means of word embeddings. Word embeddings are integrated as external components, which keeps the model small and efficient, while allowing for easy extensibility and domain adaptation. Initial experiments show that word embeddings can improve the Recall and F score of the binary classification sub-task of AND. Results for the clustering sub-task are less clear, but also promising and overall show the feasibility of the approach.
We present an implemented machine learning system for the automatic detection of nonreferential it in spoken dialog. The system builds on shallow features extracted from dialog transcripts. Our experiments indicate a level of performance that makes the system usable as a preprocessing filter for a coreference resolution system. We also report results of an annotation study dealing with the classification of it by naive subjects.
The thesis describes a fully automatic system for the resolution of the pronouns 'it', 'this', and 'that' in English unrestricted multi-party dialog. Referential relations considered include both normal NP-antecedence as well as discourse-deictic pronouns. The thesis contains a theoretical part with a comprehensive empiricial study, and a practical part describing machine learning experiments.
We present WOMBAT, a Python tool which supports NLP practitioners in accessing word embeddings from code. WOMBAT addresses common research problems, including unified access, scaling, and robust and reproducible preprocessing. Code that uses WOMBAT for accessing word embeddings is not only cleaner, more readable, and easier to reuse, but also much more efficient than code using standard in-memory methods: a Python script using WOMBAT for evaluating seven large word embedding collections (8.7M embedding vectors in total) on a simple SemEval sentence similarity task involving 250 raw sentence pairs completes in under ten seconds end-to-end on a standard notebook computer.
In this contribution we present some work of the R&D European project “LIRICS” and of the ISO/TC 37/SC 4 committee related to the topic of interoperability and re-use of language resources. We introduce some basic mechanisms of the standardization work in ISO and describe in more details the general approach on how to cope with the annotation of language data within ISO.
We describe a simple procedure for the automatic creation of word-level alignments between printed documents and their respective full-text versions. The procedure is unsupervised, uses standard, off-the-shelf components only, and reaches an F-score of 85.01 in the basic setup and up to 86.63 when using pre- and post-processing. Potential areas of application are manual database curation (incl. document triage) and biomedical expression OCR.
pyMMAX2 is an API for processing MMAX2 stand-off annotation data in Python. It provides a lightweight basis for the development of code which opens up the Java- and XML-based ecosystem of MMAX2 for more recent, Python-based NLP and data science methods. While pyMMAX2 is pure Python, and most functionality is implemented from scratch, the API re-uses the complex implementation of the essential business logic for MMAX2 annotation schemes by interfacing with the original MMAX2 Java libraries. pyMMAX2 is available for download at http://github.com/nlpAThits/pyMMAX2.
We introduce a novel scientific document processing task for making previously inaccessible information in printed paper documents available to automatic processing. We describe our data set of scanned documents and data records from the biological database SABIO-RK, provide a definition of the task, and report findings from preliminary experiments. Rigorous evaluation proved challenging due to lack of gold-standard data and a difficult notion of correctness. Qualitative inspection of results, however, showed the feasibility and usefulness of the task.
This paper will address the challenge of creating a knowledge graph from a corpus of historical encyclopedias with a special focus on word sense alignment (WSA) and disambiguation (WSD). More precisely, we examine WSA and WSD approaches based on article similarity to link messy historical data, utilizing Wikipedia as aground-truth component – as the lack of a critical overlap in content paired with the amount of variation between and within the encyclopedias does not allow for choosing a ”baseline” encyclopedia to align the others to. Additionally, we are comparing the disambiguation performance of conservative methods like the Lesk algorithm to more recent approaches, i.e. using language models to disambiguate senses.
The current state of the art for metadata provision allows for a very flexible approach, catering for the needs of different archives and communities, referring to common data category registries that describe the meaning of a data category at least to authors of metadata. Component models for metadata provisions are for example used by CLARIN and META-SHARE, but there is also an increased flexibility in other metadata schemas such as Dublin Core, which is usually not seen as appropriate for meaningful description of language resources.
Making resources available for others and putting this to a second use in other projects has never been more widely accepted as a sensible efficient way to avoid a waste of efforts and resources. However, when it comes to the details, there is still a vast number of problems. This workshop has aimed at being a forum to address issues and challenges in the concrete work with metadata for LRs, not restricted to a single initiative for archiving LRs. It has allowed for exchange and discussion and we hope that the reader finds the articles here compiled interesting and useful.
Measuring the quality of metadata is only possible by assessing the quality of the underlying schema and the metadata instance. We propose some factors that are measurable automatically for metadata according to the CMD framework, taking into account the variability of schemas that can be defined in this framework. The factors include among others the number of elements, the (re-)use of reusable components, the number of filled in elements. The resulting score can serve as an indicator of the overall quality of the CMD instance, used for feedback to metadata providers or to provide an overview of the overall quality of metadata within a repository. The score is independent of specific schemas and generalizable. An overall assessment of harvested metadata is provided in form of statistical summaries and the distribution, based on a corpus of harvested metadata. The score is implemented in XQuery and can be used in tools, editors and repositories.
Creating and maintaining metadata for various kinds of resources requires appropriate tools to assist the user. The paper presents the metadata editor ProFormA for the creation and editing of CMDI (Component Metadata Infrastructure) metadata in web forms. This editor supports a number of CMDI profiles currently being provided for different types of resources. Since the editor is based on XForms and server-side processing, users can create and modify CMDI files in their standard browser without the need for further processing. Large parts of ProFormA are implemented as web services in order to reuse them in other contexts and programs.
Linguistics is facing the challenge of many other sciences as it continues to grow into increasingly complex subfields, each with its own separate or overarching branches. While linguists are certainly aware of the overall structure of the research field, they cannot follow all developments other than those of their subfields. It is thus important to help specialists but also newcomers alike to bushwhack through evolved or unknown territory of linguistic data. A considerable amount of research data in linguistics is described with metadata. While studies described and published in archived journals and conference proceedings receive a quite homogeneous set of metadata tags — e.g., author, title, publisher —, this does not hold for the empirical data and analyses that underlie such studies. Moreover, lexicons, grammars, experimental data, and other types of resources come in different forms; and to make things worse, their description in terms of metadata is also not uniform, if existing at all. These problems are well-known and there are now a number of international initiatives — e.g., CLARIN, FlareNet, MetaNet, DARIAH — to build infrastructures for managing linguistic resources. The NaLiDa project, funded by the German Research Foundation, aims at facilitating the management and access to linguistic resources originating from German research institutions. In cooperation with the German SFB 833 research center, we are developing a combination of faceted and full-text search to give integrated access through heterogeneous metadata sets. Our approach is supported by a central registry for metadata field descriptors, and a component repository for structured groups of data categories as larger building blocks.
The paper’s purpose is to give an overview of the work on the Component Metadata Infrastructure (CMDI) that was implemented in the CLARIN research infrastructure. It explains, the underlying schema, the accompanying tools and services. It also describes the status and impact of the CMDI developments done within the CLARIN project and past and future collaborations with other projects.
This paper describes the status of the standardization efforts of a Component Metadata approach for describing Language Resources with metadata. Different linguistic and Language & Technology communities as CLARIN, META-SHARE and NaLiDa use this component approach and see its standardization of as a matter for cooperation that has the possibility to create a large interoperable domain of joint metadata. Starting with an overview of the component metadata approach together with the related semantic interoperability tools and services as the ISOcat data category registry and the relation registry we explain the standardization plan and efforts for component metadata within ISO TC37/SC4. Finally, we present information about uptake and plans of the use of component metadata within the three mentioned linguistic and L&T communities.
XML has been designed for creating structured documents, but the information that is encoded in these structures are, by definition, out of scope for XML. Additional sources, normally not easily interpretable by computers, such as documentation are needed to determine the intention of specific tags in a tag-set. The Component Metadata Infrastructure (CMDI) takes a rather pragmatic approach to foster interoperability between XML instances in the domain of metadata descriptions for language resources. This paper gives an overview of this approach.
The Component MetaData Infrastructure (CMDI) provides a lego-brick framework for the creation, use and re-use of self-defined metadata formats. The design of CMDI can be a force forgood, but history shows that it has often been misunderstood or badly executed. Consequently,it has led the community towards the dark ages of metadata clutter rather than the bright side of semantic interoperability. In this abstract, we report on the condition of CMDI but also outlinean agenda to make the CMDI world a better place to use, share and profit from metadata.
In this chapter, we discuss steps toward extending CMDI’s semantic interoperability beyond the Social Sciences and Humanities: We stress the need for an initial data curation step, in part supported by a relation registry that helps impose some structure on CMDI vocabulary; we describe the use of authority file information and other controlled vocabulary to help connecting CMDI-based metadata to existing Linked Data; we show how significant parts of CMDI-based metadata can be converted to bibliographic metadata standards and hence entered into library catalogs; and finally we describe first steps to convert CMDI-based metadata to RDF. The initial grassroots approach of CMDI (meaning that anybody can define metadata descriptors and components) mirrors the AAA slogan of the Semantic Web (“Anyone can say Anything about Any topic”). Ironically, this makes it hard to fully link CMDI-based metadata to other Semantic Web datasets. This paper discusses the challenges of this enterprise.
Data Management is one of the core activities of all CLARIN centres providing data and services for the academia. In PARTHENOS, European initiatives and projects in the area of the humanities and social sciences assembled to compare policies and procedures. One of the areas of interest is data management. The data management landscape shows a lot of proliferation, for which an abstraction level is introduced to help centres, such as CLARIN centres, in the process of providing the best possible services to users with data management needs.
This paper describes the TEI-based ISO standard 2462:2016 “Transcription of spoken language” and other formats used within CLARIN for spoken language resources. It assesses the current state of support for the standard and the interoperability between these formats and with relevant tools and services. The main idea behind the paper is that a digital infrastructure providing language resources and services to researchers should also allow the combined use of resources and/or services from different contexts. This requires syntactic and semantic interoperability. We propose a solution based on the ISO/TEI format and describe the necessary steps for this format to work as an exchange format with basic semantic interoperability for spoken language resources across the CLARIN infrastructure and beyond.
The article focuses on determining responsible parties and the division of potential liability arising from sharing language data (LD) containing personal data (PD). A key issue here is to identify who has to make sure and guarantee the GDPR compliance. The authors aim to answer 1) whether an individual researcher is a controller and 2) whether sharing LD results in joint controllership or separate controllership (whether the data's transferee becomes the controller, the joint controller or the processor). The article also analyses the legal relations of parties involved in data sharing and potential liability. The final section outlines data sharing in the CLARIN context. The analysis serves as a preliminary analytical background for redesigning the CLARIN contractual framework for sharing data.
N-grams are of utmost importance for modern linguistics and language technology. The legal status of n-grams, however, raises many practical questions. Traditionally, text snippets are considered copyrightable if they meet the originality criterion, but no clear indicators as to the minimum length of original snippets exist; moreover, the solutions adopted in some EU Member States (the paper cites German and French law as examples) are considerably different. Furthermore, recent developments in EU law (the CJEU's Pelham decision and the new right of press publishers) also provide interesting arguments in this debate. The paper presents the existing approaches to the legal protection of n-grams and tries to formulate some clear guidelines as to the length of n-grams that can be freely used and shared.
Ungoliant: An optimized pipeline for the generation of a very large-scale multilingual web corpus
(2021)
Since the introduction of large language models in Natural Language Processing, large raw corpora have played a crucial role in Computational Linguistics. However, most of these large raw corpora are either available only for English or not available to the general public due to copyright issues. Nevertheless, there are some examples of freely available multilingual corpora for training Deep Learning NLP models, such as the OSCAR and Paracrawl corpora. However, they have quality issues, especially for low-resource languages. Moreover, recreating or updating these corpora is very complex. In this work, we try to reproduce and improve the goclassy pipeline used to create the OSCAR corpus. We propose a new pipeline that is faster, modular, parameterizable, and well documented. We use it to create a corpus similar to OSCAR but larger and based on recent data. Also, unlike OSCAR, the metadata information is at the document level. We release our pipeline under an open source license and publish the corpus under a research-only license.
Contents:
1. Julien Abadji, Pedro Javier Ortiz Suárez, Laurent Romary and Benoît Sagot: "Ungoliant: An Optimized Pipeline for the Generation of a Very Large-Scale Multilingual Web Corpus", S.1-9.
2. Markus Gärtner, Felicitas Kleinkopf, Melanie Andresen and Sibylle Hermann: "Corpus Reusability and Copyright - Challenges and Opportunities", S.10-19.
3. Nils Diewald, Eliza Margaretha and Marc Kupietz: "Lessons learned in Quality Management for Online Research Software Tools in Linguistics", S.20-26.
Editorial
(2020)
Journal for language technology and computational linguistics. Special Issue on offensive language
(2020)
Recent years have seen a sharp increase in studies of offensive language (and related notions such as abusive language, hate speech, verbal aggression etc.) as well as of patterns of online behavior such as cyberbullying and trolling. Multiple efforts have been launched for the exploration of computational approaches and the establishment of benchmark datasets for various languages (Basile et al. (2019), Wiegand et al. (2018), Zampieri et al. (2019)).
In order to determine priorities for the improvement of timing in synthetic speech this study looks at the role of segmental duration prediction and the role of phonological symbolic representation in the perceptual quality of a text-to-speech system. In perception experiments using German speech synthesis, two standard duration models (Klatt rules and CART) were tested. The input to these models consisted of a symbolic representation which was either derived from a database or a text-to-speech system. Results of the perception experiments show that different duration models can only be distinguished when the symbolic representation is appropriate. Considering the relative importance of the symbolic representation, post-lexical segmental rules were investigated with the outcome that listeners differ in their preferences regarding the degree of segmental reduction. As a conclusion, before fine-tuning the duration prediction, it is important to derive an appropriate phonological symbolic representation in order to improve timing in synthetic speech.
As a part of the ZuMult-project, we are currently modelling a backend architecture that should provide query access to corpora from the Archive of Spoken German (AGD) at the Leibniz-Institute for the German Language (IDS). We are exploring how to reuse existing search engine frameworks providing full text indices and allowing to query corpora by one of the corpus query languages (QLs) established and actively used in the corpus research community. For this purpose, we tested MTAS - an open source Lucene-based search engine for querying on text with multilevel annotations. We applied MTAS on three oral corpora stored in the TEI-based ISO standard for transcriptions of spoken language (ISO 24624:2016). These corpora differ from the corpus data that MTAS was developed for, because they include interactions with two and more speakers and are enriched, inter alia, with timeline-based annotations. In this contribution, we report our test results and address issues that arise when search frameworks originally developed for querying written corpora are being transferred into the field of spoken language.
We evaluate a graph-based dependency parser on DeReKo, a large corpus of contemporary German. The dependency parser is trained on the German dataset from the SPMRL 2014 Shared Task which contains text from the news domain, whereas DeReKo also covers other domains including fiction, science, and technology. To avoid the need for costly manual annotation of the corpus, we use the parser’s probability estimates for unlabeled and labeled attachment as main evaluation criterion. We show that these probability estimates are highly correlated with the actual attachment scores on a manually annotated test set. On this basis, we compare estimated parsing scores for the individual domains in DeReKo, and show that the scores decrease with increasing distance of a domain to the training corpus.
In order to satisfy the information needs of a wide range of researchers across a number of disciplines, large textual datasets require careful design, collection, cleaning, encoding, annotation, storage, retrieval, and curation. This daunting set of tasks has coalesced into a number of key themes and questions that are of interest to the contributing research communities: (a) what sampling techniques can we apply? (b) what quality issues should we be aware of? (c) what infrastructures and frameworks are being developed for the efficient storage, annotation, analysis and retrieval of large datasets? (d) what affordances do visualisation techniques offer for the exploratory analysis approaches of corpora? (e) what legal paths can be followed in dealing with IPR and data protection issues governing both the data sources and the query results? (f) how to guarantee that corpus data remain available and usable in a sustainable way?
In this paper, we describe a data processing pipeline used for annotated spoken corpora of Uralic languages created in the INEL (Indigenous Northern Eurasian Languages) project. With this processing pipeline we convert the data into a loss-less standard format (ISO/TEI) for long-term preservation while simultaneously enabling a powerful search in this version of the data. For each corpus, the input we are working with is a set of files in EXMARaLDA XML format, which contain transcriptions, multimedia alignment, morpheme segmentation and other kinds of annotation. The first step of processing is the conversion of the data into a certain subset of TEI following the ISO standard ’Transcription of spoken language’ with the help of an XSL transformation. The primary purpose of this step is to obtain a representation of our data in a standard format, which will ensure its long-term accessibility. The second step is the conversion of the ISO/TEI files to a JSON format used by the “Tsakorpus” search platform. This step allows us to make the corpora available through a web-based search interface. As an addition, the existence of such a converter allows other spoken corpora with ISO/TEI annotation to be made accessible online in the future.
Question Answering Systems for retrieving information from Knowledge Graphs (KG) have become a major area of interest in recent years. Current systems search for words and entities but cannot search for grammatical phenomena. The purpose of this paper is to present our research on developing a QA System that answers natural language questions about German grammar.
Our goal is to build a KG which contains facts and rules about German grammar, and is also able to answer specific questions about a concrete grammatical issue. An overview of the current research in the topic of QA systems and ontology design is given and we show how we plan to construct the KG by integrating the data in the grammatical information system Grammis, hosted by the Leibniz-Institut für Deutsche Sprache (IDS). In this paper, we describe the construction of the initial KG, sketch our resulting graph, and demonstrate the effectiveness of such an approach. A grammar correction component will be part of a later stage. The paper concludes with the potential areas for future research.
We present web services implementing a workflow for transcripts of spoken language following TEI guidelines, in particular ISO 24624:2016 "Language resource management - Transcription of spoken language". The web services are available at our website and will be available via the CLARIN infrastructure, including the Virtual Language Observatory and WebLicht.
We propose a Cross-lingual Encoder-Decoder model that simultaneously translates and generates sentences with Semantic Role Labeling annotations in a resource-poor target language. Unlike annotation projection techniques, our model does not need parallel data during inference time. Our approach can be applied in monolingual, multilingual and cross-lingual settings and is able to produce dependencybased and span-based SRL annotations. We benchmark the labeling performance of our model in different monolingual and multilingual settings using well-known SRL datasets. We then train our model in a cross-lingual setting to generate new SRL labeled data. Finally, we measure the effectiveness of our method by using the generated data to augment the training basis for resource-poor languages and perform manual evaluation to show that it produces high-quality sentences and assigns accurate semantic role annotations. Our proposed architecture offers a flexible method for leveraging SRL data in multiple languages.
This paper discusses the semi-formal language of mathematics and presents the Naproche CNL, a controlled natural language for mathematical authoring. Proof Representation Structures, an adaptation of Discourse Representation Structures, are used to represent the semantics of texts written in the Naproche CNL. We discuss how the Naproche CNL can be used in formal mathematics, and present our prototypical Naproche system, a computer program for parsing texts in the Naproche CNL and checking the proofs in them for logical correctness.
In this article, we examine the effectiveness of bootstrapping supervised machine-learning polarity classifiers with the help of a domain-independent rule-based classifier that relies on a lexical resource, i.e., a polarity lexicon and a set of linguistic rules. The benefit of this method is that though no labeled training data are required, it allows a classifier to capture in-domain knowledge by training a supervised classifier with in-domain features, such as bag of words, on instances labeled by a rule-based classifier. Thus, this approach can be considered as a simple and effective method for domain adaptation. Among the list of components of this approach, we investigate how important the quality of the rule-based classifier is and what features are useful for the supervised classifier. In particular, the former addresses the issue in how far linguistic modeling is relevant for this task. We not only examine how this method performs under more difficult settings in which classes are not balanced and mixed reviews are included in the data set but also compare how this linguistically-driven method relates to state-of-the-art statistical domain adaptation.
In this article, we explore the feasibility of extracting suitable and unsuitable food items for particular health conditions from natural language text. We refer to this task as conditional healthiness classification. For that purpose, we annotate a corpus extracted from forum entries of a food-related website. We identify different relation types that hold between food items and health conditions going beyond a binary distinction of suitability and unsuitability and devise various supervised classifiers using different types of features. We examine the impact of different task-specific resources, such as a healthiness lexicon that lists the healthiness status of a food item and a sentiment lexicon. Moreover, we also consider task-specific linguistic features that disambiguate a context in which mentions of a food item and a health condition co-occur and compare them with standard features using bag of words, part-of-speech information and syntactic parses. We also investigate in how far individual food items and health conditions correlate with specific relation types and try to harness this information for classification.
Sentiment Analysis is the task of extracting and classifying opinionated content in natural language texts. Common subtasks are the distinction between opinionated and factual texts, the classification of polarity in opinionated texts, and the extraction of the participating entities of an opinion(-event), i.e. the source from which an opinion emanates and the target towards which it is directed. With the emerging Web 2.0 which describes the shift towards a highly user-interactive communication medium, the amount of subjective content on the World Wide Web is steadily increasing. Thus, there is a growing need for automatically processing this type of content which is provided by sentiment analysis. Both natural language processing, which is the task of providing computational methods for the analysis and representation of natural language, and machine learning, which is the task of building task-specific classification models on the basis of empirical data, may be instrumental in mastering the challenges of the automatic sentiment analysis of written text. Many problems in sentiment analysis have been proposed to be solved with machine learning methods exclusively using a fairly low-level feature design, such as bag of words, containing little linguistic information. In this thesis, we examine the effectiveness of linguistic features in various subtasks of sentiment analysis. Thus, we heavily draw from the insights gained by natural language processing. The application of linguistic features can be applied on various classification methods, be it in rule-based classification, where the linguistic features are directly encoded as a classifier, in supervised machine learning, where these features complement basic low-level features, or in bootstrapping methods, where these features form a rule-based classifier generating a labeled training set from which a supervised classifier can be trained. In this thesis, we will in particular focus on scenarios where the combination of linguistic features and machine learning methods is effective. We will look at common text classification tasks, both coarse-grained and fine-grained, and extraction tasks.
While good results have been achieved for named entity recognition (NER) in supervised settings, it remains a problem that for low resource languages and less studied domains little or no labelled data is available. As NER is a crucial preprocessing step for many natural language processing tasks, finding a way to overcome this deficit in data remains of great interest. We propose a distant supervision approach to NER that is both language and domain independent where we automatically generate labelled training data using gazetteers that we previously extracted from Wikipedia. We test our approach on English, German and Estonian data sets and contribute further by introducing several successful methods to reduce the noise in the generated training data. The tested models beat baseline systems and our results show that distant supervision can be a promising approach for NER when no labelled data is available. For the English model we also show that the distant supervision model is better at generalizing within the same domain of news texts by comparing it against a supervised model on a different test set.
Opinion holder extraction is one of the most important tasks in sentiment analysis. We will briefly outline the importance of predicates for this task and categorize them according to part of speech and according to which semantic role they select for the opinion holder. For many languages there do not exist semantic resources from which such predicates can be easily extracted. Therefore, we present alternative corpus-based methods to gain such predicates automatically, including the usage of prototypical opinion holders, i.e. common nouns, denoting for example experts or analysts, which describe particular groups of people whose profession or occupation is to form and express opinions towards specific items.
In der natürlichen Sprachverarbeitung haben Frage-Antwort-Systeme in der letzten Dekade stark an Bedeutung gewonnen. Vor allem durch robuste Werkzeuge wie statistische Syntax-Parser und Eigennamenerkenner ist es möglich geworden, linguistisch strukturierte Informationen aus unannotierten Textkorpora zu gewinnen. Zusätzlich werden durch die Text REtrieval Conference (TREC) jährlich Maßstäbe für allgemeine domänen-unabhängige Frage-Antwort-Szenarien definiert. In der Regel funktionieren Frage-Antwort-Systeme nur gut, wenn sie robuste Verfahren für die unterschiedlichen Fragetypen, die in einer Fragemenge vorkommen, implementieren. Ein charakteristischer Fragetyp sind die sogenannten Ereignisfragen. Obwohl Ereignisse schon seit Mitte des vorigen Jahrhunderts in der theoretischen Linguistik, vor allem in der Satzsemantik, Gegenstand intensive Forschung sind, so blieben sie bislang im Bezug auf Frage-Antwort-Systeme weitgehend unerforscht. Deshalb widmet sich diese Diplomarbeit diesem Problem. Ziel dieser Arbeit ist zum Einen eine Charakterisierung von Ereignisstruktur in Frage-Antwort Systemen, die unter Berücksichtigung der theoretischen Linguistik sowie einer Analyse der TREC 2005 Fragemenge entstehen soll. Zum Anderen soll ein Ereignis-basiertes Antwort-Extraktionsverfahren entworfen und implementiert werden, das sich auf den Ergebnissen dieser Analyse stützt. Informationen von diversen linguistischen Ebenen sollen daten-getrieben in einem uniformen Modell integriert werden. Spezielle linguistische Ressourcen, wie z.B. WordNet und Subkategorisierungslexika werden dabei eine zentrale Rolle einnehmen. Ferner soll eine Ereignisstruktur vorgestellt werden, die das Abpassen von Ereignissen unabhängig davon, ob sie von Vollverben oder Nominalisierungen evoziert werden, erlaubt. Mit der Implementierung eines Ereignis-basierten Antwort-Extraktionsmoduls soll letztendlich auch die Frage beantwortet werden, ob eine explizite Ereignismodellierung die Performanz eines Frage-Antwort-Systems verbessern kann.
This paper describes a rule-based approach to detect direct speech without the help of any quotation markers. As datasets fictional and non-fictional texts were used. Our evaluation shows that the results appear stable throughout different datasets in the fictional domain and are comparable to the results achieved in related work.
We examine the combination of pattern-based and distributional similarity for the induction of semantic categories. Pattern-based methods are precise and sparse while distributional methods have a higher recall. Given these particular properties we use the prediction of distributional methods as a back-off to pattern-based similarity. Since our pattern-based approach is embedded into a semi-supervised graph clustering algorithm, we also examine how distributional information is best added to that classifier. Our experiments are carried out on 5 different food categorization tasks.
We examine the task of relation extraction in the food domain by employing distant supervision. We focus on the extraction of two relations that are not only relevant to product recommendation in the food domain, but that also have significance in other domains, such as the fashion or electronics domain. In order to select suitable training data, we investigate various degrees of freedom. We consider three processing levels being argument level, sentence level and feature level. As external resources, we employ manually created surface patterns and semantic types on all these levels. We also explore in how far rule-based methods employing the same information are competitive.
In this paper, we examine methods to extract different domain-specific relations from the food domain. We employ different extraction methods ranging from surface patterns to co-occurrence measures applied on different parts of a document. We show that the effectiveness of a particular method depends very much on the relation type considered and that there is no single method that works equally well for every relation type. As we need to process a large amount of unlabeled data our methods only require a low level of linguistic processing. This has also the advantage that these methods can provide responses in real time.
In this paper, we explore different linguistic structures encoded as convolution kernels for the detection of subjective expressions. The advantage of convolution kernels is that complex structures can be directly provided to a classifier without deriving explicit features. The feature design for the detection of subjective expressions is fairly difficult and there currently exists no commonly accepted feature set. We consider various structures, such as constituency parse structures, dependency parse structures, and predicate-argument structures. In order to generalize from lexical information, we additionally augment these structures with clustering information and the task-specific knowledge of subjective words. The convolution kernels will be compared with a standard vector kernel.
We explore the feasibility of contextual healthiness classification of food items. We present a detailed analysis of the linguistic phenomena that need to be taken into consideration for this task based on a specially annotated corpus extracted from web forum entries. For automatic classification, we compare a supervised classifier and rule-based classification. Beyond linguistically motivated features that include sentiment information we also consider the prior healthiness of food items.
Unknown words are a challenge for any NLP task, including sentiment analysis. Here, we evaluate the extent to which sentiment polarity of complex words can be predicted based on their morphological make-up. We do this on German as it has very productive processes of derivation and compounding and many German hapax words, which are likely to bear sentiment, are morphologically complex. We present results of supervised classification experiments on new datasets with morphological parses and polarity annotations.
One problem of data-driven answer extraction in open-domain factoid question answering is that the class distribution of labeled training data is fairly imbalanced. In an ordinary training set, there are far more incorrect answers than correct answers. The class-imbalance is, thus, inherent to the classification task. It has a deteriorating effect on the performance of classifiers trained by standard machine learning algorithms. They usually have a heavy bias towards the majority class, i.e. the class which occurs most often in the training set. In this paper, we propose a method to tackle class imbalance by applying some form of cost-sensitive learning which is preferable to sampling. We present a simple but effective way of estimating the misclassification costs on the basis of class distribution. This approach offers three benefits. Firstly, it maintains the distribution of the classes of the labeled training data. Secondly, this form of meta-learning can be applied to a wide range of common learning algorithms. Thirdly, this approach can be easily implemented with the help of state-of-the-art machine learning software.
This paper presents a survey on the role of negation in sentiment analysis. Negation is a very common linguistic construction that affects polarity and, therefore, needs to be taken into consideration in sentiment analysis.
We will present various computational approaches modeling negation in sentiment analysis. We will, in particular, focus on aspects such as level of representation used for sentiment analysis, negation word detection and scope of negation. We will also discuss limits and challenges of negation modeling on that task.
We present a major step towards the creation of the first high-coverage lexicon of polarity shifters. In this work, we bootstrap a lexicon of verbs by exploiting various linguistic features. Polarity shifters, such as ‘abandon’, are similar to negations (e.g. ‘not’) in that they move the polarity of a phrase towards its inverse, as in ‘abandon all hope’. While there exist lists of negation words, creating comprehensive lists of polarity shifters is far more challenging due to their sheer number. On a sample of manually annotated verbs we examine a variety of linguistic features for this task. Then we build a supervised classifier to increase coverage. We show that this approach drastically reduces the annotation effort while ensuring a high-precision lexicon. We also show that our acquired knowledge of verbal polarity shifters improves phrase-level sentiment analysis.
We examine the task of separating types from brands in the food domain. Framing the problem as a ranking task, we convert simple textual features extracted from a domain-specific corpus into a ranker without the need of labeled training data. Such method should rank brands (e.g. sprite) higher than types (e.g. lemonade). Apart from that, we also exploit knowledge induced by semi-supervised graph-based clustering for two different purposes. On the one hand, we produce an auxiliary categorization of food items according to the Food Guide Pyramid, and assume that a food item is a type when it belongs to a category unlikely to contain brands. On the other hand, we directly model the task of brand detection using seeds provided by the output of the textual ranking features. We also harness Wikipedia articles as an additional knowledge source.
In recent years, text classification in sentiment analysis has mostly focused on two types of classification, the distinction between objective and subjective text, i.e. subjectivity detection, and the distinction between positive and negative subjective text, i.e. polarity classification. So far, there has been little work examining the distinction between definite polar subjectivity and indefinite polar subjectivity. While the former are utterances which can be categorized as either positive or negative, the latter cannot be categorized as either of these two categories. This paper presents a small set of domain independent features to detect indefinite polar sentences. The features reflect the linguistic structure underlying these types of utterances. We give evidence for the effectiveness of these features by incorporating them into an unsupervised rule-based classifier for sentence-level analysis and compare its performance with supervised machine learning classifiers, i.e. Support Vector Machines (SVMs) and Nearest Neighbor Classifier (kNN). The data used for the experiments are web-reviews collected from three different domains.
We use a convolutional neural network to perform authorship identification on a very homogeneous dataset of scientific publications. In order to investigate the effect of domain biases, we obscure words below a certain frequency threshold, retaining only their POS-tags. This procedure improves test performance due to better generalization on unseen data. Using our method, we are able to predict the authors of scientific publications in the same discipline at levels well above chance.
Though polarity classification has been extensively explored at document level, there has been little work investigating feature design at sentence level. Due to the small number of words within a sentence, polarity classification at sentence level differs substantially from document-level classification in that resulting bag-of-words feature vectors tend to be very sparse resulting in a lower classification accuracy.
In this paper, we show that performance can be improved by adding features specifically designed for sentence-level polarity classification. We consider both explicit polarity information and various linguistic features. A great proportion of the improvement that can be obtained by using polarity information can also be achieved by using a set of simple domain-independent linguistic features.
Bootstrapping Supervised Machine-learning Polarity Classifiers with Rule-based Classification
(2010)
In this paper, we explore the effectiveness of bootstrapping supervised machine-learning polarity classifiers using the output of domain-independent rule-based classifiers. The benefit of this method is that no labeled training data are required. Still, this method allows to capture in-domain knowledge by training the supervised classifier on in-domain features, such as bag of words.
We investigate how important the quality of the rule-based classifier is and what features are useful for the supervised classifier. The former addresses the issue in how far relevant constructions for polarity classification, such as word sense disambiguation, negation modeling, or intensification, are important for this self-training approach. We not only compare how this method relates to conventional semi-supervised learning but also examine how it performs under more difficult settings in which classes are not balanced and mixed reviews are included in the dataset.
Opinion holder extraction is one of the important subtasks in sentiment analysis. The effective detection of an opinion holder depends on the consideration of various cues on various levels of representation, though they are hard to formulate explicitly as features. In this work, we propose to use convolution kernels for that task which identify meaningful fragments of sequences or trees by themselves. We not only investigate how different levels of information can be effectively combined in different kernels but also examine how the scope of these kernels should be chosen. In general relation extraction, the two candidate entities thought to be involved in a relation are commonly chosen to be the boundaries of sequences and trees. The definition of boundaries in opinion holder extraction, however, is less straightforward since there might be several expressions beside the candidate opinion holder to be eligible for being a boundary.
This paper presents a survey on hate speech detection. Given the steadily growing body of social media content, the amount of online hate speech is also increasing. Due to the massive scale of the web, methods that automatically detect hate speech are required. Our survey describes key areas that have been explored to automatically recognize these types of utterances using natural language processing. We also discuss limits of those approaches.
We present a gold standard for semantic relation extraction in the food domain for German. The relation types that we address are motivated by scenarios for which IT applications present a commercial potential, such as virtual customer advice in which a virtual agent assists a customer in a supermarket in finding those products that satisfy their needs best. Moreover, we focus on those relation types that can be extracted from natural language text corpora, ideally content from the internet, such as web forums, that are easy to retrieve. A typical relation type that meets these requirements are pairs of food items that are usually consumed together. Such a relation type could be used by a virtual agent to suggest additional products available in a shop that would potentially complement the items a customer has already in their shopping cart. Our gold standard comprises structural data, i.e. relation tables, which encode relation instances. These tables are vital in order to evaluate natural language processing systems that extract those relations.
Automatic Food Categorization from Large Unlabeled Corpora and Its Impact on Relation Extraction
(2014)
We present a weakly-supervised induction method to assign semantic information to food items. We consider two tasks of categorizations being food-type classification and the distinction of whether a food item is composite or not. The categorizations are induced by a graph-based algorithm applied on a large unlabeled domain-specific corpus. We show that the usage of a domain-specific corpus is vital. We do not only outperform a manually designed open-domain ontology but also prove the usefulness of these categorizations in relation extraction, outperforming state-of-the-art features that include syntactic information and Brown clustering.
In opinion mining, there has been only very little work investigating semi-supervised machine learning on document-level polarity classification. We show that semi-supervised learning performs significantly better than supervised learning when only few labelled data are available. Semi-supervised polarity classifiers rely on a predictive feature set. (Semi-)Manually built polarity lexicons are one option but they are expensive to obtain and do not necessarily work in an unknown domain. We show that extracting frequently occurring adjectives & adverbs of an unlabeled set of in-domain documents is an inexpensive alternative which works equally well throughout different domains.
In order to automatically extract opinion holders, we propose to harness the contexts of prototypical opinion holders, i.e. common nouns, such as experts or analysts, that describe particular groups of people whose profession or occupation is to form and express opinions towards specific items. We assess their effectiveness in supervised learning where these contexts are regarded as labelled training data and in rule-based classification which uses predicates that frequently co-occur with mentions of the prototypical opinion holders. Finally, we also examine in how far knowledge gained from these contexts can compensate the lack of large amounts of labeled training data in supervised learning by considering various amounts of actually labeled training sets.
We investigate the task of detecting reliable statements about food-health relationships from natural language texts. For that purpose, we created a specially annotated web corpus from forum entries discussing the healthiness of certain food items. We examine a set of task-specific features (mostly) based on linguistic insights that are instrumental in finding utterances that are commonly perceived as reliable. These features are incorporated in a supervised classifier and compared against standard features that are widely used for various tasks in natural language processing, such as bag of words, part-of speech and syntactic parse information.
In this paper, we examine methods to automatically extract domain-specific knowledge from the food domain from unlabeled natural language text. We employ different extraction methods ranging from surface patterns to co-occurrence measures applied on different parts of a document. We show that the effectiveness of a particular method depends very much on the relation type considered and that there is no single method that works equally well for every relation type. We also examine a combination of extraction methods and also consider relationships between different relation types. The extraction methods are applied both on a domain-specific corpus and the domain-independent factual knowledge base Wikipedia. Moreover, we examine an open-domain lexical ontology for suitability.
In this paper, we investigate the role of predicates in opinion holder extraction. We will examine the shape of these predicates, investigate what relationship they bear towards opinion holders, determine what resources are potentially useful for acquiring them, and point out limitations of an opinion holder extraction system based on these predicates. For this study, we will carry out an evaluation on a corpus annotated with opinion holders. Our insights are, in particular, important for situations in which no labelled training data are available and only rule-based methods can be applied.
In this paper, we compare three different generalization methods for in-domain and cross-domain opinion holder extraction being simple unsupervised word clustering, an induction method inspired by distant supervision and the usage of lexical resources. The generalization methods are incorporated into diverse classifiers. We show that generalization causes significant improvements and that the impact of improvement depends on the type of classifier and on how much training and test data differ from each other. We also address the less common case of opinion holders being realized in patient position and suggest approaches including a novel (linguistically-informed) extraction method how to detect those opinion holders without labeled training data as standard datasets contain too few instances of this type.
We address the detection of abusive words. The task is to identify such words among a set of negative polar expressions. We propose novel features employing information from both corpora and lexical resources. These features are calibrated on a small manually annotated base lexicon which we use to produce a large lexicon. We show that the word-level information we learn cannot be equally derived from a large dataset of annotated microposts. We demonstrate the effectiveness of our (domain-independent) lexicon in the crossdomain detection of abusive microposts.
We report on the two systems we built for Task 1 of the German Sentiment Analysis Shared Task, the task on Source, Subjective Expression and Target Extraction from Political Speeches (STEPS). The first system is a rule-based system relying on a predicate lexicon specifying extraction rules for verbs, nouns and adjectives, while the second is a translation-based system that has been obtained with the help of the (English) MPQA corpus.