Adding content to my knowledge base

I wrote about it a while back but meanwhile some tools no longer exist and I stopped using some others. This post describes my current setup as well as my plans for the next evolution.

The intention and generic workflow remains unchanged but some tools have changed

• Once Omnivore disappeared I switched to Readwise reader for collection.
• When Logseq announced to move to a database version I had doubts how portable and tool independent it might be
• Independent of Logseq I decided to switch back to Emacs.
• Part of that switch away from VS Code is that I no longer use the Dendron software but I keep following its hierarchical note-taking format since it works well for me.
• On mobile I use native graphical Emacs but also the mobile Obsidian client which supports more “standard” markdown. The nice Dendron Tree plugin makes Obsidian feel like Dendron.
• I also consume PDF and EPUB documents in Moon+ Reader for Android and the tolino epos 3 e-book reader.
• calibre is my tool for e-book management.

Changes to my workflow

Capture

I tend to collect lots of links to interesting articles, PDF and EPUB files and I regularly buy e-books.

In GTD words these end in four different “inboxes”:

• Web articles go into Readwise reader to read them later.
• PDFs go into Moon Reader on my tablet.
• EPUB files go either in Moon Reader both on my tablet and my mobile.
• Larger EPUB e-books go to my tolino.

Clarify

I try to regularly process those inboxes in the spirit of “work‑in‑progress limit” to keep my reading list under control.

As before I may simply delete or archive items:

• Web links go to the archive in Readwise reader or to nb,
• PDF and EPUB files go to calibre.

The remaining parts I read and highlight the interesting parts.

Organize

This involves exporting the highlights and importing them into my knowledge base.

Readwise has a nice integration both with Moon Reader and Obsidian.

Highlights from Moon Reader are automatically imported into Readwise and then exported exactly like web article highlights.

I have configured a highlight export to Obsidian with the following settings:

The highlights go into a file with a custom title that is in the Dendron hierarchy “readwise”. The setting is readwise.{{last_highlighted_date | date('Y.m.d')}}.{{title | replace(' ','-')}} which creates a title like readwise.2021.09.06.Example-Title.md.

Each file receives an YAML header

created: {{published_date|date('c')}} 00:00:00
updated: {{last_highlighted_date |date('c')}}
title: {{title}}

Page metadata use

## Metadata
- Author: {% if author %}[[{{author}}]]{% endif %}
- Category: #{{category}}
{% if document_note -%}
- Document Note: {{document_note}}
{% endif -%}
{% if document_tags -%}
- Document Tags: {% for tag in document_tags %}[[{{tag}}]] {% endfor %}
{% endif -%}
{% if url -%}
- URL: {{url}}
{% endif -%}

and the highlights header is

 {% if is_new_page %}
{% elif has_new_highlights -%}
## New highlights added {{date|date('F j, Y')}} at {{time}}
{% endif -%}

and for each highlight I only add the text with {{ highlight_text }}.

The Obsidian Readwise plugin stores all these files in a subdirectory of Obsidian, which forms yet another inbox for processing.

Every now and then I look at all those entries and either convert and move them into the knowledge base properly or add parts to existing entries. Afterwards I delete the entry in the Readwise folder.

tolino stores all highlights in a single file notes.txt. The Notulator website lets me convert tolino highlights to text and markdown files. Once I finished an e-book I convert them and add process them into the knowledge base.

Planned changes

This workflow works reasonably well but the Readwise subscription is somewhat expensive so I looked for alternatives.

Currently I experiment with Raindrop instead. The free tier includes everything I need. One of attractive features is its API which allows me to export highlights there as well.

Raindrop is less comfortable than Readwise since the API provides the highlights in the sequence of their creation not in the sequence they appear in the text. Also headings are not marked as such, everything seems just to be text.

My current workaround is to be cautious to highlight everything immediately and use different colors for headings.

I wrote a simple Python script to import the highlights into the knowledge base.

This for now replaces the web part.

Moon Reader has a simple export feature for a file into a text file which I can process manually like the tolino highlights.

We’ll see how it goes and if I decide to trade cost for the Readwise subscription to more manual work. Should I stay with it I will buy the Raindrop subscription to support the developer and help to keep the service alive.