01
Safe data homeOWN
Data sovereignty
Your listening and viewing data lives in your personal Solid pod — not on a platform's server. You decide who can read it.
COMBINE-LD
Reveal the profile behind the stream.
Safe data homeCONSUME
One feedOBEY
Watch nextWATCH
Always onPROFILE
Made for youSLEEP
ConnectedCONFORM
YoursBUY
ControlSUBMITscroll →
What you get
One profile.
Every platform.
All yours.
COMBINE-LD captures what you listen to and watch, stores it in a pod you control, and runs recommendations on your device. Take back control.
02
One feedUNIFY
Cross-service profile
Music from SoundCloud, video from YouTube — one unified RDF profile. One pod, every platform.
03
Connected dataGRAPH
Open standards
Every event is stored as RDF triples using W3C vocabularies. Portable, interoperable, yours forever.
04
Smart picksPREDICT
Local AI, no cloud
A model running on your device reads your profile and surfaces recommendations. Nothing leaves your machine.
Media behaviour data
Everything you track becomes data you can manage.
Platform language frames tracking as convenience. COMBINE-LD reframes it as a user-managed profile: listen events, watch events, charts, and local recommendations stored in a Solid pod.
Your soundtrackLISTEN
Music history
Every track logged via the MPRIS bridge lands in your pod as an RDF listen event. Your full history — because it's yours, not Spotify's.
Watch nextWATCH
Watch history
Videos captured by the Firefox extension are stored as schema.org watch actions in your pod. Your viewing data stays with you.
Taste connectedMERGE
Aggregated charts
Cross-service aggregation produces your personal top artists, albums, tracks, and channels — all derived from pod data.
Made for youMODEL
AI recommendations
A local Ollama model queries your aggregated profile and generates suggestions entirely on your machine. No API call leaves your network.
Open standards
Built on Solid. Managed by you.
Media behaviour events are stored as RDF triples in a Kvasir Solid pod: no proprietary format, no vendor lock-in. Access is authenticated through Keycloak, and local AI recommendations can run without sending your profile to third-party infrastructure.
Solid / RDFKvasirKeycloakOllamaGraphQLFirefox extensionMPRIS
Why it matters
Platforms already profile you. Now you can too.
Every stream, every watch, every skip is already being recorded — just not for you. COMBINE-LD gives you the same profile back, stored where only you hold the key.
PersonalizedCONTROL
70%
of users feel they have little control over their personal data
Always onSPLIT
3+
streaming platforms the average user is active on — with no unified view
Data homeLOCAL
0 bytes
sent to third-party servers for AI recommendations — all processing is local
Connect allOPEN
100%
of your data stored in open RDF — portable, interoperable, yours forever
About the project
A proof-of-concept for user-centric media profiles
This is the practical implementation for a bachelor's thesis conducted at imec-IDLab, Ghent University & Odisee. The research question asks which key design decisions are required when building media behaviour profiles across multiple media services in a user-centric way.
The prototype validates those decisions through a working stack: SoundCloud via MPRIS/D-Bus, YouTube via a Firefox extension, RDF events in a Kvasir Solid pod, and local Ollama-based recommendations.
Open, portable, user-controlled media data.
OWN
THE
PROFILE
THE
PROFILE
Get started
Connect your pod
The live demo is under construction. In the meantime, read the thesis or check back when the source code goes public.
Read the thesis →
Source code
GitLab repository
Open source release coming soon