Skip to content

2022-11-23-HackyHour-5.md

Latest commit

 

History

History
185 lines (147 loc) · 7.68 KB

2022-11-23-HackyHour-5.md

File metadata and controls

185 lines (147 loc) · 7.68 KB

Markdown-based-Note-Taking

  • Talk title: Markdown based note taking apps
  • Presenter: Lukas Jelonek
  • Date: 23 November 2022
  • Notes by: Christian Krippes, Johannes Keyser

Idea

  • The Idea: Take notes only in Markdown
  • Simple file format, no dependency on specific software.
  • Avoid proprietary cloud providers if possible.

Obsidian:

  • Obsidians key feature is to connect notes with each other by tags
  • Those tags are then used to produce a graph representation of your notes
  • Community plugins for customization
  • Sync between devices costs you and runs on Obsidians Cloud. Cloud can be a downside for some people.

Joplin

  • Joplin is an open source software
  • Brings similar look and feature set as Obsidian but with the possibility to use well known solutions for data syncronization (e.g. Nextcloud, WebDAV and more)
  • No proprietary cloud needed. Setup your own syncronization between your devices
  • No graph representation of your notes
  • Web-Clipper: Save Websites as notes
  • Plugins are available

Dendron

  • Dendron is an open source software
  • Plugin for VSCode, not standalone.
  • Lots of similarities with Joplin and Obsidian
  • Graph representation of your notes is possible
  • Focus more on the way developers work

Mention from the audience

  • zettlr was a recommendation from the audiance. Its focus is more on users who write lager texts.

Mermaid

  • Mermaid creates graphs out of markdown-like text
  • Mermaid becomes handy when you already use markdown and want to quickly visualize your ideas without having to use graphics software.

Note: This section doesn't render as intendend on our GitHub Pages website.
See https://github.com/HackyHour/Giessen/blob/gh-pages/notes/2022-11-23-HackyHour-5.md#mermaid for the correct representation of the Mermaid graph.

graph TD
    A[Start] -->|Laufen| B[Ziel]
Loading

The syntax for the above diagram looks like this:

```mermaid
graph TD
    A[Start] -->|Laufen| B[Ziel]```

Visualization of Relational Data

What are relational data?

  • entities whos relationships can be given in matrix form
  • matrix examples: pairwise physical feature distances - the lower the more similar; score data ("similarities") - the larger the more similar
  • good to know: a matrix of distances between n-dimensional vectors can be perfectly reconstructed as n-dimensional point cloud up to rotation, translation, scaling
  • preferences are one example: I prefer apple over oranges; there is no common physical coordinate origin such as 0 meter length
  • matrix data might be incomplete: if relation E to B is unknown, but E to A is, and B to C is, as well as and C to A is known, then E to B might be recovered
  • scoring data my be asymmetric: Eve likes Paul, but Paul dislikes Eve (a bit)

Visualization of relational data

  • goal: intuitive visual inspection of abstract data tables / matrices
  • for example, "embedding" distance data into Euclidean space (on sheet of paper, computer screen) to get point clouds
  • distance preservation: pairwise matrix distances to be preserved in point clouds. Such distance preservation is called (metric) "multidimensional scaling". Usually symmetric inputs, i.e. dist(A,B) = dist(B,A)
  • neighborhood preservation: for example entity B as third nearest neighbor to A in the scoring matrix to be reconstructed as point B being third nearest neighbor in corresponding scatter point representation (if possible). Neighborhood preservation can be called neighbor embedding. Famous approaches are "Kruskal scaling" and t-distributed neighbor embedding (t-SNE).

Presented method: correlation based multidimensional scaling

  • scatter plot coordinates X depend on distance matrix depend on maximization goal of correlation r between each row in the source scoring matrix S and the negative distance D matrix of point scatter coordinates, e.g.
  • r(-D(X),S) to be maximized for each row in -D(X) leading to a possibly sub-optimal point cloud configuration

Nerdy technical remarks regarding optimization

  • for N points in k dimensions this leads to non-linear optimization for (N x k) free parameters for the scatter point positions
  • to be optimized by the Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm for continuous steepest descend problems
  • using partial nested derivatives of correlation r wrt point coordinates X
  • if correlation is non-linear (i.e. NOT the common Pearson correlation), such as Spearman or Kendall correlation, discrete steps in the optimization function are smoothened out by soft approximations of the ranking function

Sources

"License me if you can... or if you are allowed to"

  • Talk title: "License me if you can... or if you are allowed to"
  • Presenter: Andre Pietsch
  • Date: 23 November 2022
  • Notes by: Andre Pietsch

Ownership

  • concept of ownership in Germany only applies to material objects
  • from a legal perspective, data is not owned by someone
  • medium, where the data is actually saved on, is owned by someone

Authorship ≠ Copyright

  • authorship = state of being an author of a work or data
    • can not be transferred to others
  • copyright = right by law to be the entity which determines who may publish, copy and distribute a piece of writing, music, picture or other work of authorship
    • can be transferred to others
  • Neither is the copyright necessarily maintained by the author of a work nor does every piece of data fall under the copyright law
    • If you for example work on something and you get paid for it, then you are probably the author of it, but the copyright could lie at your employer.
  • for scientific discourse: Always cite, which data you have used, even if it is published under a Public Domain license where attributing is not needed!

The German Act on Copyright and Related Rights (UrhG)

  • copyright protects the form of a work and not the content
    • works within the meaning of the Copyright Act are “only the author’s own intellectual creations” (Div. 2 Sec. 2 No. 2 UrhG)
    • protected is the creative performance of arranging the content in exactly this or that way
  • therefore: copyright most often can not be applied to raw data or quantiative data
  • Related Rights are rights that protect objects that are not directly subject to copyright.
    • database producer right: Protects a database if a substantial qualitative or quantitative investment was made to create this database.
  • copyfraud = unlawful claim to copyright

Copyright licenses

General licenses

  • Creative Commons
    • Publish as open as possible and as closed as necessary!
    • Preferably use a CC0 or CC BY license!
  • Open Data Commons
    • Were inteded as an equivalent alternative to Creative Commons, but have not caught on.
    • Suggestion: Just use CC licenses!
  • Don't use CC licenses for software!

Software licenses

  • Four types:
    • Strong Copyleft (GNU-GPL, GNU-AGPL)
      • CC equivalent = CC BY-SA
    • Weak Copyleft (LGPL, MPL, EPL)
      • used mainly for software libraries
    • Permissive (3-clause BSD, Apache, MIT)
      • CC equivalent: CC BY
    • Noncommercial (JRL, AFPL)
      • CC equivalent: CC BY-NC
  • Special case: ethical licenses ("The Hippocratic License")
⚡