Zettelkasten Forum


Keyword list

In Niklas Luhmann’s Card Index: The Fabrication of Serendipity Johannes F.K. Schmidt Published: July 26, 2018, lists the 4 characteristics of Luhmann's zettelkasten.

  1. specific system
  2. card numbering rules
  3. internal linking
  4. comprehensive keyword index

I've been thinking (a dubious prospect) about number 4 comprehensive keyword index. Luhmann was computerized pre-full-text-search. Recognizing that full-text search has both advantages and seldom talked about disadvantages.
Recognizing that tags have both advantages and disadvantages. And these play different roles.

I can see where in Luhmann's case a way into the zettelkasten was via his massive keyword index. This could be very useful as it would be the list of key terms actually used in the zettelkasten/archive accumulated prospectively and not as with a full-text search where one has to remember retrospectively. (Cognative load) Understanding how fallible memory is, having a list of actual terms, with used spelling and gramatical tence, might be the difference between a hit and a miss. Tags are no help here as they group notes by theme or context and not literal content.

Most note-taking software address 1-3 in the list as does The Archive. I unaware of any the address the 4th in the list, comprehensive keyword index.

Please school me in my errors of logic.

Will Simpson
I'm a futzing, second-guessing, backtracking, compulsive oversharing, ZK-maniac, in other words, your typical zettelnant.
Research areas: Attention Horizon, Productive Procrastination, Dzogchen, Non-fiction Creative Writing, Cognitive Workload, Python, Data Science
kestrelcreek.com

Comments

  • If you are using Windows then both MyInfo and Ultra Recall maintain a full index of all the words in every text article. Ultra Recall also has a 'stopword' list to cut down the size of the index you can nominate words which will not be indexed, usualy used for words like :- 'of', 'is', 'the', 'and' & 'a'. Connective words which it would make no sense to search on.

    I believe InfoQube also maintains a word index but I would have to verify that.

  • @Paul_J_Miller thanks for your input. I worry about a system that maintains a full index "of all the words in every text article". What is the difference between this and a full-text search mechanism? Luhmann's zettelkasten was 90K zettels and his keyword index was only a few thousand words.
    I also worry that the cognitive value of creating a keyword list and working with it is part of the benefits of this system.
    Here are a few more values I came up with:
    1. visually see topics that have multiple references
    2. help cross-linking
    3. a way into Zettelkasten without having to remember the exact reference for text search
    4. draws attention to maintenance

    Will Simpson
    I'm a futzing, second-guessing, backtracking, compulsive oversharing, ZK-maniac, in other words, your typical zettelnant.
    Research areas: Attention Horizon, Productive Procrastination, Dzogchen, Non-fiction Creative Writing, Cognitive Workload, Python, Data Science
    kestrelcreek.com

  • I think that overview/structure notes can help with important keywords. Think "manual tag index; instead of an unsorted set of notes matching your tag search, you create a note with manual order and hierarchy.

    It's the same reason structure notes are useful at all: because you can fix a hierarchy and thus add additional meaning.

    Author at Zettelkasten.de • https://christiantietze.de/

  • @Will said:
    @Paul_J_Miller thanks for your input. I worry about a system that maintains a full index "of all the words in every text article". What is the difference between this and a full-text search mechanism? Luhmann's zettelkasten was 90K zettels and his keyword index was only a few thousand words.

    Programs which maintain a full index of words in the articles can do a full text search very much faster than programs which don't.

    Luhmann only indexed what he thought of as 'significant' words. This is a much more sensible approach. Programs just index every word unless you tell them not to (stop words).

    I agree that actually being able to see and browse the index of words is useful and interesting.

    I use 'InfoQube' and 'ConnectedText'.

    InfoQube has the disadvantage that it is very complex and a steep learning curve but there are ways of stripping out a lot of the complexity. Also links between items are not as easy or quick to insert as in ConnectedText.

    My Zettel is a single grid within InfoQube with no hierarchy, just a linear list. All the other grids are for other purposes or are just clusters of cards from the Zettel grid arranged in arbitrary ways.

    I like the hierarchical tagging system in InfoQube, it is better than almost every other program except for ConnectedText.

    ConnectedText is a wiki but with a lot of really good features. The tagging system is also hierarchical and links between articles are quick and simple to insert. You can even insert a link to an article which doesn't exist yet without breaking your train of thought just by inseeting the [[Article Name]] between square brackets. When you follow that link the program creates a new blank article with the name in the link and opens it for editing.

    Of the two programs (InfoQube and ConnectedText) both have advantages and disadvantages. I find InfoQube easier to work with but ConnectedText gives better results.

    ConnectedText has no structure imposed by the program, the data dictates the structure, it is difficult to explain but this can lead to an emergent structure appearing which was not apparent in the original data and this can lead to new insights. Similar in a way to when Niklas Luhmann talked about having a conversation with his slip box.

    This characteristic is much less noticable in InfoQube and seems absent in two pane organisers like MyInfo and Ultra Recall.

    Sorry for the long post.

  • @Paul_J_Miller no need to apologize for a detailed (long) post. I'm currently a Mac user but would lean towards ConnectText if I switch to Windows. I follow Manfred Kuehn over on https://takingnotenow.blogspot.com.

    Will Simpson
    I'm a futzing, second-guessing, backtracking, compulsive oversharing, ZK-maniac, in other words, your typical zettelnant.
    Research areas: Attention Horizon, Productive Procrastination, Dzogchen, Non-fiction Creative Writing, Cognitive Workload, Python, Data Science
    kestrelcreek.com

  • I believe there is a ConnectedText rough equivalent for the Mac called 'Tinderbox', I haven't used it but I have heard good things about it.

  • I've played around with keywords, but found that 9/10s of the time my keywords were reflected in the notes themselves, which in Notion can be searched in the databases, and double as a filter to pair down notes to the word(s) searched. Of course, if your keywords are not based in the text of the notes themselves—if keywords are meant to come from and stimulate some other cognitive reference you may have to the note (the feeling you had when you took the note, / a sensation of some kind)—then this probs won't work. But, for me it has proven to be just enough to get me back into the ZK in roughly the right spot, among the right notes.

  • @Will said:
    I unaware of any the address the 4th in the list, comprehensive keyword index.

    DEVONthink 3 does this. It has a concordancer that operates at the group level or on a single note. (See screenshot attached.) While I don't use it very often, I wonder if it might be used to help surface words that appear frequently in my notes but do not yet have a tag.

    (I realize this is an old post, but adding a comment here in case someone might need this information in the future.)

    Slavic languages, natural language processing, etc.
    Writes at ojisanseuichi.com.

Sign In or Register to comment.