Skip to content

mikezucc/questlog

Repository files navigation

Omen

Vision

Empower the prefrontal cortex https://en.wikipedia.org/wiki/Prefrontal_cortex This brain region has been implicated in planning complex cognitive behavior, personality expression, decision making, and moderating social behavior.

Mission

Organize professional relationships.

Solvable Unsolved Challenges

  1. Recalling conversations
  2. Fragmentation over how things are remembered between different parties
  3. Context switching between different relationships (i.e. work vs side_project_1 vs side_project_2)
  4. Passive note taking
  5. Self improve ability to communicate

Solutions

  1. Annotation of recorded conversations
  2. Partitioning conversation into chunks (highlighting)
  3. Tagging/ indexing of annotations for recall
  4. Association of conversations to longstanding "identities"
  5. Annotation and markup of images
  6. Store any resulting entities as homogenous set of later described frames

Random thoughts

https://github.com/mikezucc/questlog/blob/master/relationships.txt

Entities within the tool

  1. Conversations
  2. Snippets
  3. Tags (i.e. "computer vision", "q4 prospects", "ipad redesign", "magazine print review")
  4. People
  5. Contexts ("zine", "crypto clothing", "omen", "SFCC")

Discussion about Entities

Conversations are part of a superclass of entity which is denoted as "Frame". Frame signifies any kind of captured bits from the owner's sensors. Audio is the primary method of capture since it is passive. However, owner's may be more proactive during a session whereby they capture notes via other unlinked medium, like a photo or PDF document. Conversations can be corrected manually by the user.

Conversations can be subdivided into "Snippets".

Tags are phrases or words that can form relationships to Frames/Snippets. Some tags are auto generated from the output of the transcription API's. Tags created by the user will also automatically register matched phrases coming out of the transcription API's

People are user defined. They can have associated Tags, Contexts, Frames, and Snippets. They are one of the primary methods of organization.

Contexts are user defined. They are the most encompassing organization. These are meant to be be containers for Frames, Snippets. Examples of a Context include "Magazine Print Project", "Alexa Partnership Meetings", "Coffee Chats"

"""

About

Index anything using Google OCR Cloud API's

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages