Program: LOVELYHORSE 2 documents

Search documents matching “LOVELYHORSE” →

Codenames and programs recorded for each document.

LOVELYHORSE (a.k.a. LOVELY HORSE) - A GCHQ Cyber Defence Operations (CDO) project addressing a specific gap: cyber analysts were missing valuable open-source threat intelligence because reaching blogs and Twitter required using isolated JEDI terminals or reading from home, rather than on their normal classified ("high-side") systems. It aims to surface unstructured open-source information - blogs, forums, Twitter - directly on analysts' normal terminals, filtering for relevance rather than overwhelming them. Its initial prototype scraped content manually into GCDesk with personalized RSS feeds, monitoring 37 approved Twitter accounts (security researchers, hacktivist accounts, security-org accounts) and roughly 17 approved blog sources (SecureWorks, Mandiant, Trend Micro, etc.), with more pending approval. The long-term architecture stores full text as CLOBs, tags articles by entity extraction (IPs, domains, keywords) to link them to analysts' active investigations, integrates with Palantir for case enrichment, and includes a "like"-based feedback mechanism to build source-usefulness ratings - with the eventual goal of using a Google-like search front end, for which GCHQ specifically looked at NSA's LEXHOUND and its own MERA PEAK as models. glossary →
Refine (combine with another filter, via search)
Sort: Title
Show:

Documents