25 Matching Annotations
  1. Last 7 days
    1. This document outlines the main develpments in natural language processing which is in the core of our AI approach.

      This text could be useful for bridging our language and AI activities via concept of contextual meaning.

      Diplomacy is an area with - most likely - most layered and contextual meaning. If we manage to move ahead by combining our language research with AI, we may push frontier in the NLP further.

      There are also features that relate to use of AI in two other areas:

      • summarising texts and tagging in DW
      • help desk in ConfTech

      We will brainstorm about ways how we can make these potentials operational.

      ||Jovan||

    2. Combining Supervised & Unsupervised Methods

      This is our approach for 'augmented intelligence'.

    3. Automation In NLP

      to follow closely

      ||Jovan||||JovanNj||

    4. The Turing multilingual language model incorporates Information-Theoretic Framework for Cross-Lingual Language Model Pre-Training for language understanding and generation for representing 94 languages in the same vector space. 

      What exists here?

    5. However, NLP still struggles with conflation deficiency, which is the inability to discriminate between different meanings of a word. This also extends to the contextual meaning of words and sentences and also identifying sarcasm or ironic statements. Another challenge that NLP currently faces is for analysing statements with multiple meanings, often contradictory. 

      This will be crticial for diplomacy.

      ||JovanNj|| what is realistic in this field to be expected soon. You may think of one presentation for our language team, faculty on 'reality' and 'myth' in discovering hidden text meaning in AI. Can AI read between lines?

      ||Jovan||

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

  2. Jan 2021
  3. Dec 2020
    1. The language of World Bank Reports,

      Very interesting report by the Stanford Literary Lab on the evolution of World Bank language. They look into words that were used frequently in the first two and the last two decades. ||Jovan||

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. “distant reading”, by getting computers to detect undercurrents in oceans of text.

      ||JovanNj|| Do they use AI? This could be interesting for us in spotting trends in texts.

      ||Katarina_An||

    2. The Stanford Literary Lab

      ||Katarina_An||Could you make summary of their work (links, and indication of what we may use).

    3. Digital humanities are used a lot in EU projects. Shall we start using it more for - for example - our linguistical analysis. We may have category 'digital humanities' on Diplo website.

      ||kat_hone|| Could you indicate what this term 'digital humanities' cover? In this example is more textual analysis which we already do a lot.

      ||Jovan||

    4. to annotate the syntax of every sentence in the Index Thomisticus database.

      How was this done?

      ||Katarina_An||||Andrej||

    5. Soon he had switched from handwritten cards to IBM’s punch-card machines, before adopting magnetic tape in the 1950s.

      ||GingerP|| here it is. With Darija we may make this type of excursion in the past of computers.

      ||Jovan||

    6. Digital humanities - How data analysis can enrich the liberal arts | Christmas Specials

      ||Andrej|| this is interesting for language and diplomacy course. Also see with ||Katarina_An|| about Sketch Engine as tool for counting words.

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

  4. Nov 2020
    1. I enlisted its help in drafting myself a new biography, infused with the spirit of Star Wars hero Luke Skywalker.

      ||JovanNj|| How is this done?

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

  5. Oct 2020
    1. Facebook released a new open source translator which translates directly between language. It bypasses English as previously used 'bridge language' for translations from - for example - Chinese to French.

      ||JovanNj||||Cecile||||StephanieBP||||AndrijanaG||

    1. what about creating something similar for digital policy or cybersecurity (based on Speech Generator)?

      ||Jovan||||VladaR||||JovanNj||

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

    1. Fascinating text on the work of interpreters at the UN with many funny anecdotes. We may use it for our language and diplomacy course.

      ||Andrej|| ||Dragana||

    2. We never do something, we implement. We don’t repeat, we reiterate and underscore. We are never happy, we are gratified or satisfied. You are never doing a great job: you are performing your duties in the outstanding manner in which you have always discharged them. There is no theft or embezzlement, but rather failure to ensure compliance with proper accounting and auditing procedures in the handling of financial resources. This is a language the interpreter must master very early on.

      great input for our 'UN language'.

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

  6. Aug 2020
    1. There is no treatment at hand, and no certainty of a vaccine on the near horizon. The fastest vaccine ever developed was for mumps. It took four years.

      Text has nice 'music' and cadenza with short sentences.

    2. In a single season, civilization has been brought low by a microscopic parasite 10,000 times smaller than a grain of salt.

      use of nice metaphore.

    3. Never in our lives

      Great start of the text. I tis also useful that it is in bold.

      ||Jovan||

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

  7. Jul 2020
    1. GPT-2 was described as a “chameleon-like” synthetic text generator, but it wasn’t state-of-the-art in downstream tasks like question answering, summarization, or translation.

      It is an interesting point of difference between text generation, question answering, summarization or translation? Here, we can discuss a potential usability of GPT-2 ||JovanNj||||djordjej||||NatasaPerucica||||Katarina_An||

    2. GPT achieved state-of-the-art in language tasks by pairing supervised learning with unsupervised pre-training (or using the parameters from an unsupervised step as a starting point for the supervised step).

      How can GPT perform supervised and unsupervised pre-training? How it can work practiclally? ||JovanNj||||Jovan||

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL

  8. Jun 2020
    1. This is an interesting text on e-mail writing style during the COVIDD-19 crisis.

      ||Jovan||

    Created with Sketch. Visit annotations in context

    Created with Sketch. Tags

    Created with Sketch. Annotators

    Created with Sketch. URL