A new algorithm for summarizing text is reported by INSIDE HIGHER ED. Their account is about making piles of academic papers more accessible. The same approach could be applied to many kinds of texts, one day making it useful for journalists and others who work with lots of source documents.
Distilling texts is complicated for machines that can’t retain context over long passages. Instead of trying to extract meaning from the whole text, this new algorithms looks for patterns between words throughout a document and knits them together. The approach is called RUM, for Rotational Unit of Memory, describing how it pulls meaning from around a document like a sweep-hand works on a clock. The research comes from MIT.
‘(It) could potentially be used by journalists to help communicate complex research to the public, though the authors say they aren’t going to be putting journalists out of a job any time soon.‘
- Unlike many systems that use a kind of ‘cut-and-paste’ approach, RUM appears to generate new sentences while retaining original meaning.
- These are preliminary results, like so many others, indicating new research directions.
- Summarizing is a key journalistic function: think of applications in radio newscasts, digital signage, and some aspects of daily news. Some forms of automated summarization in use today are limited to shorter documents or structured data.