Versions
version 1.5 “Meaning”
- Context scoring – Tags scored, staticized & ranked by how many posts they appear in, different authors use them & how many branches they appear in. The last is context, therefore, “Meaning”.
- Sensitivity keyed to how many posts are processed. The more granular the collection of posts, the more sensitive, the larger the collection, the more generalized.
- Better trees, branches look more like narratives, memes better describe a category, branch, tree or site
- Coming next, Humans get to assign multiple associations for tags, meaning, more tags added to each post resulting in more narratives & context, hence, meaning.
version 1.4 “Honeybee”
- RSS Feed2posts. Given a set of RSS feed URLs, provide a button on the Admin control panel to grab the latest feeds, convert them to posts to be given the Athena treatment. THis is not used in the Athena Demo but is used to feature Bsky posts about the Alt-Right & Trump.
- Canonical Tags. A basic problem with tags is w/people names. My story features Nick, short for Nicholas-walk-like-the-wind. I would also like to link him to the Trumbull family. This app creates a new taxonomy with parents, meanig it is hierarchical. This allows me to manually add full names & link the short names to them. This works great for Acronyms. In Honeybee, I also use it to link people to Trump-appointed positions. My code then adds these parent tags to posts when necessary, so posts with “Nick”, “Nicholas Walk-like-the-wind” would appear together.
- Consensus Scoring. My script score tags by how many posts in a collection they appear in, as well as how many authors use them. Taken together, each tag is given a meaningful weight, used when ordering clusters of tags in storylines. This means the first tag in a branch is first for a good reason, and is truly a top concept. This is a big innovation and so simple.
- Tiling. Rather than a stack of bricks, excerpts will show up as columns of Portrait tiles. Very pretty, but also easier to make sense of. Excerpts look like pages in a book.
- Knowledge. All the elements combine to create a meaningful quantum of knowledge, with Trees, Categories giving clear pathways into the content, with easy to grasp content tiles to speed up cognition of the content subject matter.
version 1.3 Athena13. No ETA
- More code cleanup. Changing entity concepts like Flavor to Boughs, making code more semantically consistent
- Introducing View skins, nice styling tiling for Apples
- Introducing recommended trees using Meme Matching
- Custom simili Feed or REST endpoint to store data/content ready for the MemeWeb Search engine to access when it comes crawling
- MemeWeb 1.0 prototype
- Adding Authorship, part of the greater plan
- Massive code clean-up & reorganizing. Trees built in 3 stages: Corpus for content. Nexus for trees. Lexus for Apples, relevance scores & rendered elements
- streamlined generation process, creating Classic tree & Athena Trees in 2 stages rather than multiple, refining stages previously due to awkward transition from classic tree to Athena tree, simplifying code & improving performance
- Found numerous, well-hidden bugs in the code, one-in-a-thousand rock bugs
- The Athena12 tree, based on 2+ shared tags across all available posts in each sectioned portion of a website, using a Storyline context
- Deprecating the concept of flavors. Athena is a style of tree, with multiple trees identified by their root concepts. Internally, I name these mini-trees Boughs
- A streamlined ASSTRA tool, 2 rows instead of 3, with a Clean & detailed view mode available at the press of a button
- stars renamed as Relevance scores, available in Detailed view.
- Hits renamed as Relevant Bits, showing sentences containing matched tags. Available in all view modes
- Tags available in detailed mode only, cleaning up default view
- Memes used for Post relevance scoring & Relevant bits
version 1.2 Athena12. December, 2024
- Introducing the Advanced semantic Tree Retrieval Agent (ASTRA) in 3-rows: branding, settings & telemetry & the biewing/browsing area
- Introducing hits & stars, assessing post relevance to the site, branch & categories. But hidden for now