information-filtering

This repository contains an ulta-fast Python implementation of the Triangulated Maximally filtered Graph (TMFG).

Bm25 Information retrieval model using python language

In this repository we present a novel unsupervised, graph-based filter feature selection technique which exploits the power of topologically constrained network representations.

An intelligent news aggregation platform that delivers personalised, real-time content through adaptive filtering and user-centric recommendation systems.

winnow — 把群聊噪声压缩为结构化知识：LangGraph 降噪管道 + L3 知识层 + MCP 接口 + 长期群记忆

破壳 Poke — campus info filter to actionable cards (WeChat mini program + Node backend)

Local-only signal extractor for noisy information streams. Telegram, RSS, Reddit sources. Deterministic keyword scoring, no ML. Terminal-first digest.

A minimal information platform designed to help people stay informed about important real-world events without depending on social media.

Agent Skill: design a research intake filter that admits only information likely to change your next action. Search widely, intake narrowly.

A Codex skill that filters noisy raw sources into task-ready core context with traceable references.

Hermes-compatible MCP and Skill for deterministic, fail-safe email triage with Feishu-ready events.

A human-first, read-only signal digest for Technocore

🧠 The world's first Engineering Cybernetics-driven personal news intelligence system. Not another aggregator — a value transformation engine.