When Skype used to be a thing my friend and I would write the literal string keyword when we wanted to "bookmark" something in our chat. Then to find it we'd search for keyword and thered be at...
When Skype used to be a thing my friend and I would write the literal string keyword when we wanted to "bookmark" something in our chat. Then to find it we'd search for keyword and thered be at most a couple dozen results, no need to remember what the actual keyword was
it was honestly one of the best tagging systems ive used in my life
I found this post in /r/orgmode. The context here is the use of tags for personal stuff, not the role of tags in systems or websites that are accessed by more than one person. I sometimes read...
I found this post in /r/orgmode. The context here is the use of tags for personal stuff, not the role of tags in systems or websites that are accessed by more than one person. I sometimes read about that stuff given that I briefly used org-roam and obsidian in the past.
The author has studied tags in an academic setting. He seems to know his stuff. Other than that, I know nothing about him. I'm curious to see if my impressions align with those of people with actual knowledge.
I tried using tags for my personal wiki, and actually intuitively followed most of the author's recommendations, but that just doesn't work for me. In the end IMO tags are just barely less...
I tried using tags for my personal wiki, and actually intuitively followed most of the author's recommendations, but that just doesn't work for me. In the end IMO tags are just barely less restrictive than folders.
In Obsidian, instead of tags, I sometimes add a keywords frontmatter field and fill it with words that should help me find the note back in the future. I "boost" that field in the Omnisearch plugin settings and it works well enough.
While it kinda sounds like tags, it definitely works differently:
Those keywords aren't indexed anywhere in the app, they're not clickable, they don't work and I don't think of them as a kind of category. When I write them, I only think "what kind of search will I do in 6 months to find this note back?"
It works for me because I 100% rely on search to find back my notes. I don't have "maps of contents", most of my notes are orphans, and most of them aren't even in folders.
Interesting read. My experience is pretty aligned with the author's conclusions. My major tagging effort is the category assignments in my quote database . Having too many tags is definitely a...
Having too many tags is definitely a challenge. Right now there are 277 categories for 1063 quotes. Tagging categories is by far the biggest effort when adding quotes.
I have a simple suggestion feature in the admin interface that does word matching on the quote, but this is just beginning. I basically have to surf the list every time, unless I'm adding a batch and I start to get the list in my head.
For a particular quote, putting it in a category for a word contained in the quote feels redundant, but the category may include quotes that pertain to the idea without using the word, so I end up tagging ones that could also be found through search.
Many words are homographs, so the match may be for a different sense of the word. This is why I don't automatically include the search results in the category displays.
I try not to have quotes with no categories, which means can be a struggle for quotes that have a narrow meaning.
Adding a category is tricky because then I have to go through the existing quotes by search or related categories to see if they should go in the new category.
I am curious about using one of the LLMs to ingest the corpus and suggest categories for new or existing quotes, but I haven't figured out how to tackle that problem.
I have toyed with adding a category analytics admin view where I could do things like see quotes that are missing a word matched tag, find categories where all the quotes are word matched, find categories that are strict subsets of other categories, etc. but it's not immediately clear that this would have value. It's more of a gut feel. Since I am the primary user of the site, I haven't had too many complaints from the user base so far :)
When Skype used to be a thing my friend and I would write the literal string
keywordwhen we wanted to "bookmark" something in our chat. Then to find it we'd search forkeywordand thered be at most a couple dozen results, no need to remember what the actual keyword wasit was honestly one of the best tagging systems ive used in my life
I found this post in /r/orgmode. The context here is the use of tags for personal stuff, not the role of tags in systems or websites that are accessed by more than one person. I sometimes read about that stuff given that I briefly used org-roam and obsidian in the past.
The author has studied tags in an academic setting. He seems to know his stuff. Other than that, I know nothing about him. I'm curious to see if my impressions align with those of people with actual knowledge.
I tried using tags for my personal wiki, and actually intuitively followed most of the author's recommendations, but that just doesn't work for me. In the end IMO tags are just barely less restrictive than folders.
In Obsidian, instead of tags, I sometimes add a
keywordsfrontmatter field and fill it with words that should help me find the note back in the future. I "boost" that field in the Omnisearch plugin settings and it works well enough.While it kinda sounds like tags, it definitely works differently:
Those keywords aren't indexed anywhere in the app, they're not clickable, they don't work and I don't think of them as a kind of category. When I write them, I only think "what kind of search will I do in 6 months to find this note back?"
It works for me because I 100% rely on search to find back my notes. I don't have "maps of contents", most of my notes are orphans, and most of them aren't even in folders.
Interesting read. My experience is pretty aligned with the author's conclusions. My major tagging effort is the category assignments in my quote database .
Having too many tags is definitely a challenge. Right now there are 277 categories for 1063 quotes. Tagging categories is by far the biggest effort when adding quotes.
I have a simple suggestion feature in the admin interface that does word matching on the quote, but this is just beginning. I basically have to surf the list every time, unless I'm adding a batch and I start to get the list in my head.
For a particular quote, putting it in a category for a word contained in the quote feels redundant, but the category may include quotes that pertain to the idea without using the word, so I end up tagging ones that could also be found through search.
Many words are homographs, so the match may be for a different sense of the word. This is why I don't automatically include the search results in the category displays.
I try not to have quotes with no categories, which means can be a struggle for quotes that have a narrow meaning.
Adding a category is tricky because then I have to go through the existing quotes by search or related categories to see if they should go in the new category.
Sometimes it's hard to concisely describe an idea, leading to categories like Accumulative Weight
I have emergent tag hierarchies as mentioned in the article, such as animals being the super set of dogs and cats, plus other animals that don't have enough mentions to merit their own category . They are manually managed and have no explicit relationships defined.
Future work
I am curious about using one of the LLMs to ingest the corpus and suggest categories for new or existing quotes, but I haven't figured out how to tackle that problem.
I have toyed with adding a category analytics admin view where I could do things like see quotes that are missing a word matched tag, find categories where all the quotes are word matched, find categories that are strict subsets of other categories, etc. but it's not immediately clear that this would have value. It's more of a gut feel. Since I am the primary user of the site, I haven't had too many complaints from the user base so far :)