Informatics Practices · Ch 2 — Emerging Trends
Natural Language Processing (NLP)
Natural Language Processing (NLP)
Every time your search engine suggests the next word of your sentence while you are still typing the keywords, or a spell-checking feature flags a misspelt word, you are watching Natural Language Processing (NLP) at work. NLP is the field that deals with the interaction between humans and computers using human spoken languages — languages such as Hindi or English — rather than artificial programming languages.
The consequences are all around us. It is now possible to search the web by voice, and to operate or control our devices using our voice alone. All of this has been made possible by NLP: the computer takes in ordinary human language, makes sense of it, and responds usefully.
Two core conversions make voice interaction possible, and an NLP system can perform both (this is what Figure 2.1 depicts):
- Speech-to-text — spoken words are converted into written text that the computer can process.
- Text-to-speech — written text is converted into spoken output that the user can hear.
Beyond voice control, two application areas of NLP are growing especially fast:
- Machine translation — a rapidly emerging field in which machines translate text from one language to another with a fair amount of correctness — good, though not yet perfect.
- Automated customer service — computer software that can interact with customers directly, understanding and serving their queries or complaints without a human agent handling every conversation. …
Figure 2.1, captioned Use of natural language processing, is a flat vector illustration of a person controlling a computer entirely by voice. The person is seated in an armchair at a desk, with one hand raised toward a monitor mounted on the wall in front of them. Beside the raised hand is a wireless, sound-wave style signal symbol — the visual shorthand for spoken words travelling through the air toward the device.
The monitor itself tells the other half of the story. Its screen displays an audio waveform with a microphone icon at its centre: the machine is listening, capturing the incoming speech as a sound signal it can convert and process. Nothing in the scene involves a keyboard or a mouse — the interaction runs on voice alone. …