Feature Stories - Language processing tools to help teachers and students
'There is a growing problem in the education community, particularly in higher education: the size of classes. With more students per class, one resource that becomes particularly scarce is teacher time,' says Wolfgang Greller, Associate Professor for New Media Technologies and Knowledge Innovation at the Open University of the Netherlands. Dr. Greller, the coordinator of the LTfLL project, believes that a combination of advanced 'natural language processing' (NLP) solutions and other technologies can help alleviate the problem. Across OECD countries, student enrolment in universities has increased by 4 % per year on average since the mid-1990s and some university lectures are now attended by hundreds of students. Worryingly, studies have found that increases in class size have a direct negative impact on student performance, with students in smaller classes doing markedly better than students in larger ones, regardless of the subject they are studying. Within the LTfLL project, the researchers focused on three key areas where technology can provide important benefits to both students and teachers alike: student positioning and conceptual development helps ensure that students are enrolled at the right level, depending on their existing knowledge, and that their progress can be monitored; dialogue and text analysis, coupled with intelligent, automated feedback, helps guide students in their studies and reduce the workload on teachers; and semantic and social resource discovery makes subject-specific information and resources more relevant and easier to find. The open source LTfLL tools, developed with the help of EUR 2.85 million in research funding from the European Commission, are designed to be integrated seamlessly with existing learning management systems such as Moodle in any technology-enhanced learning environment. They can be implemented together or individually depending on user needs. 'The technologies and tools we developed are useful in almost any educational environment, not just universities but also secondary schools, adult education and professional training,' Dr. Greller notes. Nonetheless, some of the applications really come into their own in cases where a single teacher or lecturer must deal with many students. For instance, some universities are giving classes to up to 400 students at a time. In order to engage students more actively in the learning process, the students may be divided into groups of a few people tasked with discussing the same subject, but that still leaves a hundred different groups to monitor. 'That's a hundred conversations going on in parallel. It's impossible for a single lecturer or tutor to monitor, evaluate and provide feedback to all the groups, the most they can do is dip in occasionally here and there,' Dr. Greller says. 'Then at the end of the discussion period the lecturer has to assess and analyse what all the different groups talked about and the conclusions they drew - it's a horrendous amount of work if you consider the number of groups, discussions and exchanges that have taken place.' Analyse my dialogue A dialogue analysis tool based on innovative NLP technology developed by the LTfLL researchers addresses the problem. Tested in trials at universities in the Netherlands and Romania in three languages (Dutch, English and Romanian), it analyses students’ text-based interactions in online chats or forums, looking not only at how much they participate but how they participate. 'It helps tutors qualitatively analyse students' conversations by seeing who has been active, and in what ways they have been active: Did they ask questions? Did they respond to other students' questions? Were they isolated in their position? Did they help other people? Or did they play a role as mediator? and so on,' Dr. Greller explains. 'It tells the lecturer how well they covered the main points on a given subject or if they have deviated in their discussion and ended up talking about football or something else totally off-topic.' In this way, lecturers and teachers can save time when it comes to assessing how well their students are progressing, but, more significantly, students can obtain on-demand feedback about their progress directly from the analysis tool. This kind of on-demand feedback is equally essential in the case of students writing essays or articles, for which the LTfLL research team developed a separate but related text analysis tool. 'The tool provides continuous support and feedback about how well they are capturing the ideas and concepts related to the subject they are studying. This helps to avoid the risk that they deviate too much from the focus of the subject, something that otherwise they might only realise too late when they hand in a final draft of the essay to their tutor,' Dr. Greller says. The analysis tools work on the basis of comparing the language used by the students to that of experts in the field, looking not just at how often students mention certain words but how they use them and in what context, as well as carrying out stylistic and textual coherence checks similar to plagiarism software. The tools not only use advanced NLP technology, but also semantic analysis and social discovery. Semantic search helps provide students with easy access to related content, while a social component engine dedicated to the subject they are studying helps them communicate and share information from trusted counterparts. 'In this way it's not a computer algorithm that's selecting the content that is offered but rather a trusted source: your tutor, your fellow students or friends,' Dr. Greller says. 'It aids student engagement and helps people to study in both formal and informal settings.' As open source software, the tools have been made available to the research community for further development and several of the project partners are continuing to work on them. Meanwhile, Austrian partner Bit Media is planning to offer hosting and support services to schools and colleges that are unable to implement the technology by themselves. LTfLL received research funding under the European Commission's Seventh Framework Programme (FP7). Useful links: - 'Language technologies for lifelong learning' - LTfLL website - LTfLL project factsheet on CORDIS Related articles: - Feature Stories - A qualitative leap forward in Natural Language Processing for education - Speech-activated databases facilitate access to information - ICT project to enhance human-system interactivity