Replies: 2 comments 8 replies
-
Thank you for your suggestions! On automatic titling, how would you see that actually implemented? I think this could be implemented with an in-editor LLM interface which would let you highlight parts of the note and tell the llm something like "Generate a title and subtitles". On linking, we have two big things coming in the next release (i.e. this week): Obsidian-like square bracket linking & a feature that let's you highlight any block of text and show the semantic links corresponding to only that block. We think that'll be quite a powerful workflow, giving you the ability to leverage vector search to then generate Obsidian like backlinks...What do you think? |
Beta Was this translation helpful? Give feedback.
-
This would be an awesome idea! |
Beta Was this translation helpful? Give feedback.
-
as Related notes are connected automatically via vector similarity so if we can use LLM-powered RAG on the corpus of notes to create/suggest Automatic titles/subtitles for notes saves time also keeps everything organized, especially when you have a lots of notes if we allow the AskFile Feature to rewrite the current file or suggest Automatic titles/subtitles for notes. The titles/subtitles also give a quick idea of what each part of your notes is about, helping you understand better and easy to connect related topics/subtopics in your notes. | Graph view - In Reor, notes are divided into chunks and stored in an internal vector database. the graphic view function in Obsidian which connect by [[]] to surround the phrase or text: [[link text]]. When typing the brackets, Obsidian offers autocomplete suggestions, which are based on the already existing note title or keywords but have to link manually by [[]] and visually in graph view section https://help.obsidian.md/Plugins/Graph+view. Check out this Obsidian plugin (https://github.com/AlexW00/obsidian-note-linker) for a similar concept, where it suggests note references based on existing titles or keywords. i think notes can be automatically connected based on their similarity, by enabling an LLM RAG system we can suggest links between points or notes (which auto heading might also help to connect). This eliminates the need for manual searching through keywords for linking.
Beta Was this translation helpful? Give feedback.
All reactions