grep.ro

The conference had more than 70 small presentations by various people; bioinformatics is mostly new to me, and so is the TDWG community, so I experimented with taking notes in a format usable with melody.

Syntax

For each day there's a text file and the day's talks are sections in that file. Each section starts with a headline, some metadata (title and speaker/author) and then several lines of free-form text. I've also taken pictures of speakers and slides so the notes include links to these pictures. If the speaker is important to me, I'll store his information in a separate note, sort of like an address book entry, and link to that. All of this is done with some simple, machine-readable markup syntax (the code to read said syntax does not exist, but it's on my todo list). Here's an example:

    == SERNEC ==
    title: Developing a Cyberinfrastructure: Experiences From a Regional Network of Plant Taxonomists
    by: Zack Murrell [contact: zack_murrell], Derick Poindexter and Michael Denslow
    ~~
    [photo: 3650, 3651]
    trying to provide tools for curators to reduce their workload so they have more time to contribute to the database
    use existing social networks - ppl sharing technical knowledge; use the lessons to make new networks
    tdwg / gbif good as a means to share/communicate
    [photo: 3655]

So, what does this tell us? "SERNEC" is the title I used for the presentation – a short name, or slug, if you will. "title:" and "by:" are obvious. The [contact: ...] thingie next to Zack is a link saying "Zack is in my contact list, in the file zack_murrell". "~~" marks the end of key/value pairs of metadata; whatever remains is text describing the talk. The "[photo: ...]" links point to photographs of slides that I've taken during the talk.

Doing it properly

Choosing good syntax is important. It needs to be natural, so the text file makes sense when read directly, but unambiguous, so the parser can make sense of it. Here are some of my choices: HTTP header syntax is good at expressing metadata, so I'm using something similar. Link syntax (square brackets, starting with a keyword followed by a colon) is taken from various forms of wiki markup. I'm not too keen on the "~~" separator; I just needed something quick and unambiguous to say "metadata ends here". Blank lines are kept for section separators, because the meaning is obvious to a human reader; Markdown and others use blank lines to separate paragraphs, which is similar. Same for titles enclosed by "==" marks: it stands out visually and it's easily parsable.

Why bother with markup?

The syntax outlined above is generic, one could use it for describing lots of things: I use it to store contact information, todo lists, calendar events, to organize projects, etc. I want this data to be structured so I can parse it with code and then display it in various ways. For example, when describing a project, I could add markup to specify deadlines; afterwards, I could write code to make an overview, showing all deadlines for the following two weeks. Once the raw text file is parsed, it's easy to select particular data from it.

Going back to my conference notes, I would like a view with an index of talks (only titles and speaker names); the talk's content should be one click away, same for pictures. And this should all tie in with other information: relevant photos are pulled from my photo archive, contact information comes from my address book, and the conference itself shows up in my timeline.