Search Results for "semantic"

rag-search is a lightweight Retrieval-Augmented Generation API service designed to provide structured semantic search and answer generation through a simple FastAPI backend. The project integrates web search, vector embeddings, and reranking logic to retrieve relevant context before passing it to a language model for response generation. It is built to be easily deployable, requiring only environment configuration and dependency installation to run a functional RAG service.

...It is designed to help an AI agent retain project context, decisions, structured facts, reasoning patterns, and prior conversations across sessions. The system uses seven memory layers that combine flat files, SQLite, full-text search, structured facts, semantic recall, Qdrant vector storage, and a self-curating wiki pipeline. It injects only relevant context back into the agent so memory remains useful without wasting tokens. The project is provider-agnostic and can work with services such as OpenRouter, OpenAI, Anthropic, Ollama, or local models. It is useful for power users who want a self-hosted, long-term memory layer for AI coding and research workflows.

...It avoids a separate training phase for the quantizer, which can simplify setup compared with systems that require codebook learning. TurboVec is useful for developers building retrieval, ranking, semantic search, recommendation, or AI memory systems. Its main value is combining Rust performance with a Python-facing workflow for modern vector search experiments and applications.

...It provides a drag-and-drop canvas with reusable components representing servers, databases, APIs, message queues, and user clients, enabling stakeholders to create diagrams that describe data flows, dependencies, and interactions clearly and intuitively. Beyond drawing diagrams, the tool supports semantic relationships — meaning elements can have structured metadata, links, and annotations that explain behavior, constraints, and performance assumptions directly within the design. It’s particularly useful for collaborative design sessions, documentation, and architecture reviews, bringing visual clarity to topics that are often buried in text or code comments.

Paul Graham GPT is a specialized AI-powered search and chat app built on a corpus of essays from Paul Graham, giving users the ability to query and discuss his writings in a conversational way. The repo stores the full text of his essays (chunked), uses embeddings (e.g. via OpenAI embeddings) to allow semantic search over that corpus, and hosts a chat interface that combines retrieval results with LLM-based answering — enabling RAG (retrieval-augmented generation) over a fixed dataset. The app uses a Postgres database (with pgvector) hosted on Supabase for its embedding store, making the backend relatively simple and accessible, and the frontend is again built with Next.js/TypeScript for a modern responsive UI. ...

...In each round, features produced by the network are clustered (e.g. k-means), and the cluster IDs become supervision targets in the next epoch, encouraging the model to refine its representation to better separate semantic groups. This alternating “cluster & train” scheme helps the model gradually discover meaningful structure without labels. DeepCluster was one of the early successes in unsupervised visual feature learning, demonstrating that clustering-based reformulation can rival supervised baselines for many downstream tasks. The repository includes code for feature extraction, clustering, training loops, and evaluation benchmarks like linear probes. ...

...Using the web user interface, the crawlers (web, file, database, etc.) and the client libraries (REST/API , Ruby, Rails, Node.js, PHP, Perl) you will be able to integrate quickly and easily advanced full-text search capabilities in your application: Full-text with basic semantic, join queries, boolean queries, facet and filter, document (PDF, Office, etc.) indexation, web scrapping,etc. OpenSearchServer runs on Windows and Linux/Unix/BSD.

This project is no longer actively maintained. It is succeeded by the Eclipse RDF4J project, which can be found at GitHub and at http://www.rdf4j.org/. Sesame is a de-facto standard framework for processing RDF data. This includes parsing, scalable storage, reasoning and full SPARQL 1.1 query/update support. Sesame offers a fully modular toolkit and an easy-to-use Java API that can be connected to all leading RDF storage solutions.

The goal of this project is to provide a Content Provider plugin to RSS Feeds for iQser GIN Semantic Middleware (www.iqser.com). It is an easy way to integrate news sources for smart mashups.

HERAKLES is a reasoning broker framework for OWL (Web Ontology Language) reasoning systems.

itemis CREATE - formerly known as Yakindu Statechart Tools (SCT) - is a tool for the specification and development of reactive, event-driven systems with the help of state machines. It consists of an easy-to-use tool for graphical editing and provides validation, simulation and code generators for different target platforms. Visit http://www.statecharts.org for more information! !! YAKINDU SCT HAS MOVED !! DOWNLOAD FROM https://info.itemis.com/download-yakindu-statechart-tools

SCAN (Smart Content Aggregation and Navigation) is a universal semantic content aggregator. It combines search, text analysis, tagging and metadata functions to provide new user experience of desktop navigation and document management.

a Solr Based Semantic Mediawiki Store

MapPSO is a tool for Ontology Alignment, which uses Discrete Particle Swarm Optimisation. A particle swarm is used to search for the optimal alignment. The algorithm is massively parallel and adapts naturally on parallel architectures.

A universal platform for resource discovery and description that shares XML meta-data over existing peer-to-peer (P2P) networks such as Gnutella and JXTA.

Self-aware dynamic, reflective, and declarative semantic space representation, transformation, and exploration system.

MACOW is a formal and scalable mandatory access implementation suitable on open worlds such as the provided on Semantic Web, Autonomic Computing and Coaltions and Federations scenarios. It is able to access control on distributed systems.

Semantic Peer to Peer Recommender Systems -A simulation framework for semantic peer to peer recommender systems -P2P prototype

WOOKI is a peer-to-peer wiki. It based on unstructured p2P network with data replication. WOOT framework synchronizes data. SWOOKI is a semantic extenion of WOOKI. It is implemented as a plugin of WOOKI. SWOOKI is peer-to-peer semantic wiki.

WYMIWYG KnoBot is a content management system for the semantic web designed to promote the exchange of content with a P2P and reputation based system. The aim is to create a socio-topical space, a perspective dependent space of information and knowledge.

Self Organization of an adaptive learning framework for pervasive graph structures from grids, utilities to semantic governance. Business, performance and provisioning assurance are converged with EGA, ITIL, eTOM,SOX, HIPPA, CobiT and AC

The PeerTrust system uses policies to describe trust and security requirements. Such policies are a component of Peertrust run-time system that can negotiate to establish trust among systems on the Semantic Web

CRML - Conflict Resolution Markup Language - http://www.equiforum.org - is an open source innitiative to develop an XML based Semantic Web ontology (RDF, OWL) for researchers interested in peace to create a standard way of collecting and sharing informa

The Dataisland Framework is a ReST compliant content management and semantic web service framework.

A semantic memory software tool “reflects” one's “imagination”. In it, arbitrary concepts may be described, stored, arranged, transformed, and shared. One may verbalize thoughts as they arise in the mind to externalize and digitize “mental obj