During the recent pandemic many organisations have produced documents that are designed to be easier to read such as the UK Government guidance about ‘National Lockdown: Stay at Home’ and we went even further with our symbol versions. This type of support may help those with intellectual disabilities cope with complex content and Inclusion Europe offer guidelines for making text more understandable.
However, making documents easier to read using text simplification techniques can take time. It is not just the choice of words that needs to be considered (the lexicon) but also the way sentences are written (the syntax) to remove complex grammatical structures.
John Rochford, has been working in this area for a long time with people who have cogntive impairments and has a blog called Clear Helper. He has been exploring the use of machine learning and natural language processing to speed automated text simplification.
This March saw John sharing his ideas for EasyText AI with a group of us, discussing how one could achieve improved automated text simplification and add keyword support using symbols to help those with intellectual disabilities cope better with web page content. The challenges include overcoming polysemy (multiple meanings) – words that look the same but may require very different symbols and the degree to which symbols can help rather than confuse. Using different forms of artificial intelligence and training data, allows us to think that it is possible to overcome some issues around ‘disambiguation’ (removing uncertainty of meaning), but ensuring web content personalisation, in terms of symbol choices, is quite a different challenge!
Spring is here, she felt a spring in her step as she jumped into the spring.