Natural Language Processing (NLP) aims to create a communication bridge between man and machine in a different way than programming. The goal is to be able to communicate with computers in the same way we talk to people.
This is not a new concept, as it has been around for almost half a century already, but it is still under development. The application in education have already been mapped out but is waiting for technology to catch up. Some of the use cases within education include automatic test scoring, substitute teachers and support for administrative tasks. The difference between using NLP and conventional software to perform these tasks boils down to the steepness of the learning curve.
Top uses for NLP in education:
Addressing a machine as you would a colleague or a secretary can speed up administrative processes. As an example, if students were able to ask questions as if speaking to an AI assistant like Siri and receive the same level of qualified answers a teacher would provide.
Organization and administration
Just think about the application process in most schools or colleges. Students write essays, submit their report cards, recommendation letters and a CV. All these documents are carefully analyzed by the admission committee and sometimes an interview takes place, of which the results are part of a report.
What if instead of having all the data sorted by traditional manual processes, machines can perform the set task within a reduced timeframe than the traditional norm. It could free valuable time for teachers to develop a better curriculum or create personal development tasks for the student.
Visit DATAx London on May 2–3, 2018
This same idea of evaluation can be replicated during the school year. Right now, machines are helpful in evaluating single answer or multiple-choice questions, but are not sophisticated enough to tackle paragraph reading, essay scoring or even assessing solutions given to science problems.
There are already some tools, like Grammarly which provide services such as context-accurate suggestions for spelling and accurate tense use. Other tools like Yoast, which is used for SEO, can detect the degree of complexity of text and check if it is right for the intended audience.
The aforementioned tools can be seen as a scratch on the surface of what AI-powered grading tools could do. Not only would it mean less work for teachers, but it could provide a much more objective evaluation, clear feedback and improvement directions for the student.
Teaching and learning
NLP currently helps some educational methods, but NLP could very well even replace some parts of the act entirely at some point in the future. For example, an NLP powered system could act as a substitute tutor right from an app.
More importantly, there is also the reverse process. To create great NLP, it is necessary to understand how the human mind processes information and context. This step will help replicate the process through the neural networks. The first result will be a fine-tuned NLP system. A secondary outcome will be a better understanding of the way people learn languages.
The research work for NLP in different languages could be in fact a mapping of that specific language, which could then be reused in teaching the language to foreigners, not only a text analysis solution.
Educational tools powered by NLP:
The future is already here, and some tools are already available on the market for students to use. Here is a selection of some that can be utilized.
Writing Mentor and Language Muse
Similar to Grammarly, the Writing Mentor is, in fact, an add-on for Google Docs. It aims to help those polishing their English language skills. It offers feedback on writing, coherence and even comes with suggestions related to the best choice of words. It gives constant feedback and also builds a report for tracking improvements.
The Muse aims to help English learners by creating a range of different activities to stimulate more than one learning style. It includes vocabulary, writing, sentence structure and summary writing.
Visit Innovation Enterprise's Digital Strategy Innovation Summit in New York on March 1–2, 2019
This is a focused educational tool to help teachers select the best texts for the instructional process. It takes into consideration multiple variables, starting with readability but looking further into the level of argumentation, complexity, challenging aspects and more.
This is a text editor which helps writers improve their style. It focuses on pointing out lengthy sentences which could use trimming. It also highlights sentences which are too complicated and as a result may make the reader confused. Furthermore, it suggests better synonyms and helps get rid of passive voice.
Challenges of NLP for education
The greatest challenge NLP has to face whenever there is text analysis involved is ambiguity and this comes at three different levels: lexical (the words), phrase and semantic.
The fact that the same word can mean different things, depending on the context makes NLP vulnerable to misinterpretations and possible faulty results. For example, the word "table" could mean a desk or spreadsheet.
Next, the semantic structure can be less than clear, especially in longer phrases where the parts of the same phrase are separated by additional words. This is a characteristic of philosophical essays, where the idea is not always expressed straightforward. Just imagine such a tool for analyzing Shakespeare.
These problems need to be addressed before teachers can hand over evaluation and therefore the future of the students, to machines.
Most likely, in the next few years, all NLP tools for education will mostly have a supporting role instead of a decisive one.