57 Matching Annotations
  1. Nov 2022
    1. n recent years, the neural network based topic modelshave been proposed for many NLP tasks, such as infor-mation retrieval [11], aspect extraction [12] and sentimentclassification [13]. The basic idea is to construct a neuralnetwork which aims to approximate the topic-word distri-bution in probabilistic topic models. Additional constraints,such as incorporating prior distribution [14], enforcing di-versity among topics [15] or encouraging topic sparsity [16],have been explored for neural topic model learning andproved effective.

      Neural topic models are often trained to mimic the behaviours of probabilistic topic models - I should come back and look at some of the works:

      • R. Das, M. Zaheer, and C. Dyer, “Gaussian LDA for topic models with word embeddings,”
      • P. Xie, J. Zhu, and E. P. Xing, “Diversity-promoting bayesian learning of latent variable models,”
      • M. Peng, Q. Xie, H. Wang, Y. Zhang, X. Zhang, J. Huang, and G. Tian, “Neural sparse topical coding,”
    2. e argue that mutual learningwould benefit sentiment classification since it enriches theinformation required for the training of the sentiment clas-sifier (e.g., when the word “incredible” is used to describe“acting” or “movie”, the polarity should be positive)

      By training a topic model that has "similar" weights to the word vector model the sentiment task can also be improved (as per the example "incredible" should be positive when used to describe "acting" or "movie" in this context

    3. . However, such a framework is not applicablehere since the learned latent topic representations in topicmodels can not be shared directly with word or sentencerepresentations learned in classifiers, due to their differentinherent meanings

      Latent word vectors and topic models learn different and entirely unrelated representations

  2. Oct 2022
    1. Perhaps there are al-ready too many formal discourses on method, and cer-tainly there are too many inspirational pieces on how tothink. Neither seem to be of much use to those for whomthey are apparently intended. The first does not usuallytouch the realities of the problem as the beginning studentencounters them: the second is usually vulgar and oftennonsense.

      A description of the problem.

      Also missing are concrete examples and modeling of behavior for students to see and follow.

  3. Sep 2022
    1. IntertextsAs Jonathan Culler writes: “Liter-ary works are not to be consideredautonomous entities, ‘organicwholes,’ but as intertextual con-structs: sequences which havemeaning in relation to other textswhich they take up, cite, parody,refute, or generally transform.” ThePursuit of Signs (Ithaca, NY: CornelUniversity Press, 1981), 38.

      Throughout Rewriting: How To Do Things With Texts (Utah State University Press, 2006) Joseph Harris presents highlighted sidebar presentations he labels "Intertexts".

      They simultaneously serve the functions of footnotes, references, (pseudo-)pull quotes, and conversation with his own text. It's not frequently seen this way, but these intertexts serve the function of presenting his annotations of his own text to model these sorts of annotations and intertextuality which he hopes the reader (student) to be able to perform themselves. He explicitly places them in a visually forward position within the text rather than hiding them in the pages' footnotes or end notes where the audience he is addressing can't possibly miss them. In fact, the reader will be drawn to them above other parts of the text when doing a cursory flip through the book upon picking it up, a fact that underlines their importance in his book's thesis.


      This really is a fantastic example of the marriage of form and function as well as modelling behavior.


      cc: @remikalir

    1. Harris further illustrates hisown idea of voices adding to an author’s text; each chapter contains multiple “intertexts,”which are small graphics with citation references to outside materials addressed nearby inthe text. These intertexts reinforce the practice of adding voices to the author’s docu-ment. These illustrations are effective; essentially, Harris is reflecting and modeling thepractice.

      I quite like the idea of intertexts, which have the feeling of annotating one's own published work with the annotations of others. A sort of reverse annotation. Newspapers and magazines often feature pull quotes to draw in the reader, but why not have them as additional voices annotating one's stories or arguments.

      This could certainly be done without repeating the quote twice within the piece.

  4. Aug 2022
    1. Ausführlicher Überblick zu den erwarteten Klimaveränderungen auf kroatischem Staatsgebiet bis 2040, mit einem Ausblick auf 2070. Berücksichtigt vor allem RCP4.5, aber auch RCP8.5. Macht (bei sehr oberflächlichem Durchsehen mit mangelnden Sprachkenntnissen) an manchen Stellen einen etwas verharmlosenden Eindruck.

  5. Jun 2022
    1. By asking students to share their annotations openly, we help students to see a wide range of annotation practices, thus demystifying what has often been a private, individual practice.

      Teachers can model their own reading and annotating practices for students, but this can be expanded when using social annotation. This will allow students to show each other a wider variety of potential note taking and annotation strategies which help to reinforce the teacher's own modeling. This can be useful modelling of a practice in public which has historically been done privately.

      By featuring notes which might be reused for papers or developing later research, teachers can also feature the portions of the note taking process which can be reused for developing new ideas. How might annotations within a text be linked to each other outside of the particular flow of the paper? Might there have been different orderings for the arguments that may have been clearer?

      What ideas in the broader class might the ideas within a particular text be linked to? What ideas outside of the class can be linked to those found within the text?

      In less experienced groups, teachers might occasionally call out individual annotations in discussion to ask the group for what purposes a student might have annotated specific portions to highlight the various methods and reasons.

      What are the list of particular note types here? - Paraphrasing segments to self-test for understanding - Creation of spaced repetition type notes for memorizing definitions and facts - Conversations with the text/original author and expansion of the ideas - Questioning the original text, do we agree/disagree? - Linking ideas from the text into one's broader knowledge base - Highlighting quotes for later reuse - others??


      Link to - double-entry journaling in Bruce Ballenger<br /> - types of questions one might ask within a text, Ballenger again

  6. Apr 2022
    1. Denise Dewald, MD 🗽. (2021, August 12). Here are some modeling predictions for the delta variant from COVSIM (group at North Carolina State): PLEASE CHECK THIS OUT - RESOURCES TO SHARE WITH YOUR SCHOOL DISTRICT School-level COVID-19 Modeling Results for North Carolina for #DeltaVariant https://t.co/zU5hB9bKlY [Tweet]. @denise_dewald. https://twitter.com/denise_dewald/status/1425626289399009288

    1. Builds on previous research Robust spreadsheet modelling techniques • design principles • modularity • quality controls and diagnostics • version control • formulae conventions • format conventions • logical thought. This list is complemented by additional skills including: • understanding and applying complex financial principles to models and demonstrating that experience • understanding the implications of risk and can model for this • ability to communicate and work with people from various business functions • ability to manage a project • capable of interpreting the results of the models and presenting them to a wide audience that includes bank, investors, senior staff and the public, and • bound by some code of ethics, which does not mean the modeller must have an accounting background but has followed some professional training route in the past. (Avon 2013, p. 447)

      Skills required to be a successful financial modeler.

    1. Much of Barthes’ intellectual and pedagogical work was producedusing his cards, not just his published texts. For example, Barthes’Collège de France seminar on the topic of the Neutral, thepenultimate course he would take prior to his death, consisted offour bundles of about 800 cards on which was recorded everythingfrom ‘bibliographic indications, some summaries, notes, andprojects on abandoned figures’ (Clerc, 2005: xxi-xxii).

      In addition to using his card index for producing his published works, Barthes also used his note taking system for teaching as well. His final course on the topic of the Neutral, which he taught as a seminar at Collège de France, was contained in four bundles consisting of 800 cards which contained everything from notes, summaries, figures, and bibliographic entries.


      Given this and the easy portability of index cards, should we instead of recommending notebooks, laptops, or systems like Cornell notes, recommend students take notes directly on their note cards and revise them from there? The physicality of the medium may also have other benefits in terms of touch, smell, use of colors on them, etc. for memory and easy regular use. They could also be used physically for spaced repetition relatively quickly.

      Teachers using their index cards of notes physically in class or in discussions has the benefit of modeling the sort of note taking behaviors we might ask of our students. Imagine a classroom that has access to a teacher's public notes (electronic perhaps) which could be searched and cross linked by the students in real-time. This would also allow students to go beyond the immediate topic at hand, but see how that topic may dovetail with the teachers' other research work and interests. This also gives greater meaning to introductory coursework to allow students to see how it underpins other related and advanced intellectual endeavors and invites the student into those spaces as well. This sort of practice could bring to bear the full weight of the literacy space which we center in Western culture, for compare this with the primarily oral interactions that most teachers have with students. It's only in a small subset of suggested or required readings that students can use for leveraging the knowledge of their teachers while all the remainder of the interactions focus on conversation with the instructor and questions that they might put to them. With access to a teacher's card index, they would have so much more as they might also query that separately without making demands of time and attention to their professors. Even if answers aren't immediately forthcoming from the file, then there might at least be bibliographic entries that could be useful.

      I recently had the experience of asking a colleague for some basic references about the history and culture of the ancient Near East. Knowing that he had some significant expertise in the space, it would have been easier to query his proverbial card index for the lived experience and references than to bother him with the burden of doing work to pull them up.

      What sorts of digital systems could help to center these practices? Hypothes.is quickly comes to mind, though many teachers and even students will prefer to keep their notes private and not public where they're searchable.

      Another potential pathway here are systems like FedWiki or anagora.org which provide shared and interlinked note spaces. Have any educators attempted to use these for coursework? The closest I've seen recently are public groups using shared Roam Research or Obsidian-based collections for book clubs.

  7. Mar 2022
    1. Unwin, H. J. T., Hillis, S., Cluver, L., Flaxman, S., Goldman, P. S., Butchart, A., Bachman, G., Rawlings, L., Donnelly, C. A., Ratmann, O., Green, P., Nelson, C. A., Blenkinsop, A., Bhatt, S., Desmond, C., Villaveces, A., & Sherr, L. (2022). Global, regional, and national minimum estimates of children affected by COVID-19-associated orphanhood and caregiver death, by age and family circumstance up to Oct 31, 2021: An updated modelling study. The Lancet Child & Adolescent Health, 6(4), 249–259. https://doi.org/10.1016/S2352-4642(22)00005-0

  8. Feb 2022
    1. Students should not only learn to write papers, butalso learn facts, be able to discuss their ideas in seminars and listencarefully to lectures

      I wonder if there are any labs which not only have journal clubs, but have a shared note taking system or zettelkasten as well to keep as a community resource.

      I'm sure there are probably a few lab wikis in existence.

      Are professors keeping public note collections that they share with students or fellow researchers?

  9. Nov 2021
    1. Dr. Thomas Wilckens. (2021, October 31). JCVI facing calls from within for greater transparency over decision-making https://buff.ly/3GwVqCZ JCVI has been criticised for failing to publish detailed minutes, modelling and analysis behind its decision to advise vaccinating all over-16s in Britain #covid19 #coronavirus https://t.co/nWbnvci7LI [Tweet]. @Thomas_Wilckens. https://twitter.com/Thomas_Wilckens/status/1454798820156530689

  10. Oct 2021
  11. Sep 2021
  12. Jul 2021
    1. NIHR HPRU in Behavioural Science and Eval Bristol. (2021, May 27). Event: The CONQUEST study has collected data on the contacts, behaviour & symptoms of staff & students @BristolUni during #COVID19 to inform policy & math modelling. Join us for this webinar on 8 June for an update on the study, its impact & future plans. Https://t.co/DHrmferP0L https://t.co/25cOASdyKJ [Tweet]. @HPRU_BSE. https://twitter.com/HPRU_BSE/status/1397906695775473671

  13. May 2021
  14. Apr 2021
  15. Mar 2021
  16. Feb 2021
  17. Dec 2020
  18. Nov 2020
  19. Oct 2020
  20. Aug 2020
  21. Jun 2020
  22. Jan 2020
  23. Oct 2019
  24. May 2019
  25. Mar 2019
    1. Mithering about the unmodellable. "Sometime late last year I went to the Euro IA conference with Anya and Silver to give a talk on the domain modelling work we've been doing in UK Parliament."

  26. Jul 2018
    1. However, computers and algorithms – even the most sophisticated ones – cannot address the fallacy of obviousness. Put differently, they can never know what might be relevant.

      One goal of systems science and modelling, to explore what might be relevant and give us better heuristics.