It’s the use of computers to analyse text. This is a broad area of endeavour, depending on what the question is that you’re trying to answer.
Below we list four things that people sometimes bracket under natural language processing. The solutions to problems under these four headings are sometimes related, but just as often are not.
At present this repo summarises some findings we’ve made from creating products to answer questions relating to Search and Topics. You can click the headings to see our thoughts in these areas. If you are unsure about a piece of jargon or terminology you might find it represented in our glossary.
Finding parts of the text that are about a particular topic of interest (e.g. allow the user to search for parts of the text that are about biscuits).
Summarise what topics the texts are about (e.g. automatically note that our topics are about biscuits, people, and cars, and assign each sentence or document to one or more of these topics).
Find key identifiers in the text (e.g. tag all occurrences of people’s names, or of names of biscuits).
Understand some type of linguistic input and translate it to some action (e.g. a chatbot responding to typed input).
Written by Sam Tazzyman, DaSH, MoJ