TXTWRK.com — Just For You Feed
Friday Development Mini Report
Date: 23 October 2026
Technology: PHP · MySQL · JavaScript · HTML
Algorithm: Personalised Fuzzy Relevance
Overview
TXTWRK's new "Just For You" feed introduces personalised
content discovery based on what members actually write.
Rather than relying entirely on likes, clicks or
popularity, TXTWRK analyses the subjects appearing
in a member's recent posts and notes.
The goal is TRUE RELEVANCE:
Find content that shares the flavour of someone's
interests, without requiring exact text matches.
1. Understanding Your Interests
The algorithm examines:
•Your latest 10 visible posts
Source: post_table.post_message
•Your latest 10 visible notes
Source: business_notes.note
These 20 text samples create a temporary
personal interest profile.
2. Language Processing
PHP processes the text to:
1. Remove HTML and unnecessary characters.
2. Convert words to lowercase.
3. Remove common words such as "the" and "and".
4. Simplify selected word endings.
5. Identify frequently occurring topics.
6. Calculate weighted relevance scores.
The algorithm retains up to 35 significant terms.
Example interest profile:
Community 95%
Technology 83%
Investment 72%
Education 61%
These are illustrative scores.
3. Finding Relevant Content
TXTWRK examines up to 500 recent posts
from other members.
Each candidate is ranked using:
•Weighted keyword and topic overlap.
•Normalisation for text length.
•A small recency boost.
•A small discovery boost.
Posts do not need identical sentences to match.
For example:
Your writing:
"Investing in sustainable community gardens."
Potential recommendations:
•Community greenhouse projects.
•Local sustainability initiatives.
•Educational gardening programmes.
•Investment in neighbourhood development.
4. Relevance Meets Discovery
The feed aims for a mixture of:
80% — Higher-ranked relevant posts.
20% — Broader discovery posts.
It also attempts to limit repeated appearances
from the same author.
This helps avoid a feed filled with nearly
identical content.
5. Privacy and Technology
•Runs using TXTWRK's PHP/MySQL infrastructure.
•No external AI API is required.
•Uses the member's own recent writing.
•Respects existing post visibility filtering.
•Integrates with the existing feed interface.
6. Current Limitations
This is a first-generation lexical relevance engine.
It recognises related words and overlapping topics,
but does not yet fully understand synonyms or
sentences with completely different vocabulary.
The current candidate pool contains up to 500
recent posts, rather than searching the entire
database using SQL FULLTEXT.
Future improvements could include:
•SQL FULLTEXT candidate discovery.
•Better synonym and phrase recognition.
•More advanced semantic similarity.
•Member-controlled relevance preferences.
•Improved diversity and feedback signals.
Conclusion
TXTWRK's Just For You feed introduces a different
approach to personalised discovery:
"Let what you write help shape what you discover."
It is an early foundation for a more meaningful
recommendation system — one that prioritises
relevance, community interests and exploration.
TXTWRK.com
Friday, 23 October 2026
Building relevance around real conversations.
COMMUNITY
People