Natural language processing (NLP) is a field of machine learning that deals with unprocessed natural language. Classifying documents, answering search engine queries and finding information in large amounts of text are all studied under NLP. In the recent years, with the help of newly available deep neural networks, there has been considerable advances in NLP. These advances include better understanding of semantic relations between words and methods for creating applications that can work on multiple languages simultaneously.
Although, NLP has great potential for certain domains, due to its certain limitations in understanding natural language, it must be used with great care. Therefore, we provide a brief survey on the capabilities and limitations of these methods. We illustrate these practical considerations in relation to tasks relevant to automated OER generation including automated question generation, related content retrieval and content summarization.
We identify language teaching as a potential application where automated OER creation holds considerable promise. We outline our findings for an automated system that can analyze a given English text, identify the level of text using a widely used language standard, and find open resources that can replace this text. We also provide guidance on how this system can be extended to other languages and be used for large scale generation of OER for language teaching.