This page belongs to: Action Plan for Critical Technologies

Natural Language Processing

Systems that enable computers to recognise, understand and use written and/or spoken language in the same ways that people use language to communicate with each other. Natural language processing is a type of artificial intelligence. Applications for natural language processing include predictive text, language translation, virtual assistants and chat bots, summarising long documents and making technologies more accessible and inclusive.

Key sectors

Influences all sectors of the economy, including:

  • Agriculture
  • Banking & Finance
  • Communications
  • Defence & Defence Industry
  • Health
  • Education & Research
  • Transport & Logistics
  • Intelligence
  • Law enforcement
  • Manufacturing

Estimated impact on national interest

Economic Prosperity - High
National Security - High

Key Australian Government actions

Initiatives

  • Artificial Intelligence Action Plan
  • AI-focused Cooperative Research Centre Projects
  • Digital Economy Strategy
  • Next Generation Technologies Fund & Defence Innovation Hub

Regulations

  • Defence and Strategic Goods List 2021

Example outcomes

  • Technology that is more accessible and inclusive.
  • Faster, more intuitive access to information.
  • More productive use of existing corporate, government and academic information.
  • Faster, more equitable access to government services.
  • Increased capacity for online and remote learning solutions.
  • More natural human-machine teaming.
  • Fewer language barriers.
  • More accessible and inclusive society.
  • More local and special interest news publishing.
  • Increased access to counselling and machine-driven psychological support services.
  • More timely regulation and compliance monitoring.

Underpinning science

ANZ Standard Research Classification Category

  • Artificial Intelligence
  • Computer vision and multimedia computation
  • Data management and data science
  • Graphics, augmented reality and games
  • Human-centred computing
  • Machine learning
  • Software engineering
  • Pure Mathematics
  • Statistics
  • Linguistics

Example applications

Readiness Level – Now

  • Text-to-speech translation in multiple languages and vice-versa
  • Multiple language translation in multiple forms of media
  • Writing news articles for simple, structured events (finances, sports, weather)
  • Increased cognition for search engine queries
  • Sentiment analysis for news and speech media
  • Simple chatbots and digital assistants like Alexa/Google Home
  • Low-code application development using simple language to code translation
  • Monitoring social media for signs of suicidal ideation
  • Identifying signs of mental illness from everyday speech and writing

Readiness Level – 2–5 years

  • Text summarisation and concept identification
  • Cross-language search engines
  • Regulatory design and compliance monitoring (RegTech)
  • Complex, creative and unique content creation from small snippets of media
  • Digital assistants, chatbots and search engines that can understand and respond to increasingly complicated queries
  • Create lyrics for music that match a given theme/beat/tune

Readiness Level – Beyond 5 years

  • Real-time idiomatic language machine translation
  • Personalised virtual assistants – can create media such as text or speech in your style/voice
  • Create a synthetic language
  • Comprehend, analyse and report on huge datasets consisting of multiple media sources
  • Thought to text/speech conversion and vice versa including privacy filters

Australia's place in the world

The United States, led by Alphabet Inc (aka Google), has the highest research impact in this field; China ranks second and Australia 8th. Among Australian institutions, the University of Technology Sydney, followed closely by the University of Sydney, have the highest research impact for natural language processing (NLP); both are also in the top 50 internationally. Global venture capital (VC) investment in NLP is concentrated in China and the United States. Patents in this area have been growing at around 39% p.a. since 2015, with the United States and China well ahead in patent numbers.

Australia’s research strength in this area, coupled with strengths in related technologies (including machine learning and AI algorithms), and strong international partnerships, means we are well positioned to take advantage of NLP’s opportunities.

Opportunities and risks

For decades, people have conformed to the limited and often unintuitive ways of computers. Natural language processing (NLP) turns this process around by enabling computers to ‘understand’ how humans naturally communicate and use information.

NLP can make workers more productive by easing their interactions with machines, reducing the time taken to find and analyse information, and automating simple or repetitive tasks. NLP can also improve Australia’s national security by supporting law enforcement and intelligence professionals in identifying relevant information amidst the unprecedented volume of new content that people create every day. NLP may also make Australia more inclusive by enabling more people to communicate using their preferred language and by making technology more like talking or writing.

NLP also presents risks for Australia. Poorly designed NLP-based bots, text generation or analysis tools can perpetuate harmful biases or stereotypes, or exclude people from participating equally in society. NLP has potentially malicious applications, including enabling small teams of people to mount sophisticated and widespread disinformation campaigns, and creating software that can reliably impersonate the writing style, quirks and mannerisms of well-known authors and public figures.

Research impact (RI)

The United States leads research impact, with Australia ranked 8th globally. Total volume of published research has increased at around 19% p.a. over the 5 year period 2016–2020, with 19% of research involving international collaboration.

  1. USA - 43332
  2. China - 26906
  3. UK - 10802
  4. India - 7209
  5. Germany - 6626
  1. Australia - 4346

The research impact provides an indication of the productivity of a country or institution. Here, productivity was assumed to be represented by the volume of publications (i.e. scholarly output) as an indicator of the resources & facilities, and the level of interest in the publications as an indicator of quality

VC investment

Australia is ranked 24th globally for relative venture capital (VC) investment, which is led by China. Investment in this area has been slowing since 2016.

  1. China
  2. USA
  3. France
  4. UK
  5. Canada
  1. Australia

Data from Crunchbase. The Crunchbase database provides a partial view of the global VC landscape. However the quantity, quality and richness of the data are considered to be statistically significant, and indicative of global trends.

Patents - international

The number of patents filed annually in this field has increased by 39% from 2015 to 2019. Most patents for this technology were filed by applicants or inventors from the United States and China, and Australia ranks 14th.

  1. USA - 2697
  2. China - 2206
  3. R. of Korea - 591
  4. Japan - 381
  5. India - 309
  1. Australia - 52

Research institutions - international

The United States has 7 institutes in the top 10 international institutions. Alphabet Inc (aka Google) is the top ranked institute.

Rank Top International Institution Research Impact
1 Alphabet Inc. | United States 4300
2 Stanford University | United States 4275
3 Chinese Academy of Sciences | China 4139
4 Harvard University | United States 3558
5 Microsoft USA | United States 2137
6 Massachusetts Institute of Technology | United States 1949
7 University of Washington | United States 1928
8 National Institutes of Health | United States 1926
9 Nanyang Technological University | Singapore 1867
10 University College London | United Kingdom 1816

Research Australian - Australia

Within Australia, the University of Technology Sydney has the highest research impact, which places it 39th internationally. The University of Sydney is ranked 48th internationally.

Rank Top Australian Institution Research Impact
1 University of Technology Sydney 819
2 University of Sydney 688
3 University of Adelaide 370
4 University of Melbourne 367
5 University of Western Australia 335
6 University of New South Wales 299
7 CSIRO 248
8 Monash University 240
9 University of Queensland 203
10 Australian National University 150

Patents - Australia

Eleven Australian companies have each filed one patent family in this field. Five of these are included in the table below.

Top 5 Australian Patent Applicants Patent Families
Atlassian 1
Vendome Consulting 1
Cochlear 1
The Utree Group 1
Safesoft 1

Patents filed by Australian businesses, 2015–2019.