Writing Assistance and Language Learning
Mastering a language is the result of a long process of learning and practice. Intelligent systems can effectively assist learners in this process by providing learning material, automatic tutoring, and feedback on the learning progress. Our research interests at UKP include:
- Automatic generation and analysis of exercises,
- Contextual linking to learning materials and language resources,
- Modelling text quality and providing feedback on quality issues,
- Automatic assessment of free-text answers and learner proficiency.
We put a special focus on studying written discourse and assisting writing tasks. Recent works deal with the difficulty prediction of exercises, the analysis of argumentation structures, the modeling of writing quality in collaborative texts, and the generation of instant quality feedback. To evaluate the effectiveness of our assistance systems, we conduct empirical user studies with cooperation partners from psychology, didactics, and language learning research.
- a! – Automated Language Instruction: Our goal is to automatically generate C-tests that match a learner's current proficiency. We therefore predict the CEFR level (Common European Framework of Reference for Languages) of C-tests and the difficulty of every gap. We then match the C-tests with the learner's goals and learning progress. Several factors play a role in C-test difficulty assessment and generation, including readability assessment of texts, word and spelling complexity, and the learner's answers in previous C-tests. We do not limit ourselves to a single language, but explore cross-lingual approaches towards predicting and manipulating exercise difficulty.
- Fostering diagnostic competence in medical and teacher education via adaptive online-case-simulations (FAMULUS): The interdisciplinary FAMULUS project aims to study how online case simulations that provide automatic adaptive feedback can foster students' diagnostic skills. To generate automatic feedback, we will develop novel methods for identifying and evaluating diagnostic reasoning (e.g. hypothesis generation, evidence generation and evaluation, hypothesis acceptance or rejection) in student essays. The effect of such feedback on the development of diagnostic skills will then be evaluated in a user study with students from medicine and education.
- Argumentative Writing Support: Formulating persuasive and well-formed arguments is a challenging task and a crucial aspect in writing skills acquisition. However, current writing support is limited to feedback about grammar or spelling and there is no system that provides formative feedback about argumentative writing. In this project, we aim to research novel methods for assisting authors in writing persuasive arguments.
- Automated Exercise Generation for Language Learners: In a labor market that is increasingly globalized, knowledge of one or even more than one foreign language is more relevant than ever before. New research technologies from the field of natural language processing can support self-directed learning as they offer tools for the assessment of text difficulty and enable the automated generation of adequate exercises.
- Automatic Quality Assessment and Feedback in eLearning 2.0 (AQUA): The project investigates the use of Natural Language Processing and Machine Learning techniques to automatically measure the quality of user-generated text in the Web 2.0, such as forum posts or blog entries. This can be utilized to recommend the user high-quality materials, to implement quality-aware information retrieval, or to predict the popularity of web sites for computational advertising.
- Educational Text Analytics: Automatically Grading Text Responses: Initial work in automatically grading text responses using text analytics concerns participation in the 2013 SemEval challenge on textual entailment, applying automatic grading methods to a novel corpus of children's essays, and publishing a survey article on the state of the art of short answer grading methods.
- Educational Web 2.0: In the EduWeb project, we seek to implement our vision of technology enhanced education of the 21st century. A vast amount of content is produced by many people every day, but despite their interconnection through the World Wide Web, their efforts are often isolated from each other. To overcome this problem, the UKP Lab will provide and explore new algorithms to simplify tedious, recurring tasks as well as improving the coordination with the community.
- Feature-based Visualization and Analysis of Natural Language Documents (VisADoc): The amount of digital text data, e.g. created by the Web users, has been rapidly growing over the recent years yielding heavy information overload. Search engines help the user to find the relevant documents, but do not provide advanced tools for analyzing and understanding the dimensions of text relevant to the users' needs.
- LOEWE Research Center “Digital Humanities”, TP 2.3 “Text as a Process”: This project aims at gaining insights into the collaboration, production, and reception processes of collaboratively created and edited texts using the example of Wikipedia. We are particularly interested in article revisions (i.e., how articles change over time), discussions (i.e., the hidden relations between article discussions and content), and writing quality.
- Mining Lexical-Semantic Knowledge from Dynamic and Linguistic Sources and Integration into Question Answering for Discourse-Based Knowledge Acquisition in eLearning (QA-EL): The project investigates novel applications of dynamic lexical-semantic resources for information search in eLearning. We develop novel ways of mining knowledge from Wikipedia and other Web 2.0 knowledge repositories, and we apply question answering in the area of discourse-based knowledge acquisition in eLearning for the first time.
- Open window: open window is concerned to give the opportunity for learners to look into interlinked educational content in the World Wide Web. As part of the Open Window project, technologies for automatic linking educational content with different collaboratively created media are developed. These collaborative created media include encyclopedias, such as Wikipedia, and social media services, such as Twitter.
- Sentiment Analysis for User-Generated Discourse in eLearning 2.0 (SENTAL): The project aims to support easy exploration of subjective content and feedback generation to content providers. We develop components for subjectivity identification, opinion and topic extraction.
- Utilizing Web Knowledge: Language Technologies and Psychological Processes (WiWeb): The project examines the usefulness of selected, innovative language technologies according to psychological processes and models. This research project will provide important groundwork by bringing together scientists from computer science, industrial science, and psychology.
- Wikulu – Self-Organizing Wikis: Wikulu assists the user while creating, editing, or searching content. The self-organizing abilities of the wiki are enabled through Natural Language Processing algorithms like keyphrase extraction, document summarization, document clustering, or graph-based term weighting.
- Argumentation mining
- Cognate production
- Discourse analysis
- Sentence simplification
- Spelling correction
- Text-quality assessment
- Textual revisions
- Darmstadt Knowledge Processing Repository (DKPro)
- Interactive summarization
- Spelling difficulty prediction
Cooperations with other areas
- AIPHES research area D: Area D of the research training group “Adaptive Preparation of Information from Heterogeneous Sources” (AIPHES) is concerned with the quality requirements for the selection of heterogeneous document sources and the automatic document summarization. Together with our partners, we research intelligent writing aids for online journalism and online monitoring of heterogeneous information sources.
- Argumentation Mining: Formulating persuasive and well-formed arguments is a challenging task and a crucial aspect in writing skills acquisition. In this project, we aim to research novel methods for assisting authors in writing persuasive arguments.
- Dr. Christian Stab
Argumentative Writing Support by means of Natural Language Processing, Technische Universität Darmstadt, 2017.
Reviewer: Prof. Dr. Iryna Gurevych
Co-reviewers: Prof. Marie-Francine Moens, Ph.D. (Katholieke Universiteit Leuven), Prof. Dr. Manfred Stede, (Universität Potsdam)
- Dr. Lisa Beinborn
Predicting and Manipulating the Difficulty of Text-Completion Exercises for Language Learning, Technische Universität Darmstadt, 2016.
Reviewer: Prof. Dr. Iryna Gurevych
Co-reviewers: Prof. Dr. Torsten Zesch (Universität Duisburg-Essen), Prof. Dr. Detmar Meurers (Universität Tübingen)
- Dr. Nicolai Erbs
Approaches to Automatic Text Structuring, Technische Universität Darmstadt, 2015.
Reviewer: Prof. Dr. Iryna Gurevych
Co-reviewers: Prof. Dr. Eneko Agirre (University of the Basque Country), Prof. Dr. Torsten Zesch (Universität Duisburg-Essen)