Loading...
Starting point
From fragmentation to semantic interoperability
Integrating and consolidating audiovisual production data — across both video and audio — is a complex process. It requires input from third parties, the merging of multiple data sources, and ultimately metadata enrichment to give content the depth of information it needs.
The technical platform behind RTVE Play faced the challenge of integrating and consolidating information from multiple heterogeneous sources, both internal and external — ranging from historical content in the RTVE Archive to the latest productions. This technical complexity resulted in a fragmented user experience and made it difficult to access the vast audiovisual heritage of Spain's public broadcaster efficiently.
- Fragmented systems: The need to integrate multiple heterogeneous systems spanning historical content and current production.
- Limited interoperability: Difficulty exchanging information effectively between internal systems and with third parties.
- Insufficient metadata: Content lacked the precise, detailed tagging needed to support search and discovery.
- Suboptimal user experience: The need for more intuitive navigation with genuine semantic awareness of its resources.
Solution overview
RTVE Grafo: a semantic interoperability project
RTVE Grafo is designed to be genuinely useful — responding to the knowledge needs people express through their searches. To achieve this, the project has transformed the vast data and knowledge held within RTVE Play into fast, precise answers to complex questions. It does so within a framework that demands explainability, drawing on artificial intelligence that emulates human-like reasoning — semantic reasoning — powered by high-performance knowledge graph technology.
Building on the RTVE Grafo knowledge graph, RTVE has implemented a faceted semantic search engine that supports full-text search, contextual search, and advanced entity-based search, as well as the generation of enriched information pages. The result is a more intuitive, straightforward, natural and contextual system for querying and retrieving content from RTVE Play.
In practice, this means searches are no longer driven purely by keywords — they are driven by the meaning and context behind the terms, making them significantly more efficient and relevant.
The RTVE knowledge graph brings together approximately 2,000,000 digital resources, 26 million entities, around 85 million relationships between those objects and entities, and 167 million triples. These are used not only to understand the meaning behind a user's search query, but also to offer a faceted browsing system for exploring the collection and all its resources — providing every possible mode of navigation across that body of entities. The ontology and RTVE Grafo will enable content to be represented with greater precision, detail, depth and expressiveness, and will open up more natural, conversational ways for users to engage with it.
The RTVE Grafo project aims to deliver the best possible experience for its digital visitors, offering a search engine that queries a knowledge graph in which audiovisual resources are interconnected — enabling results to be presented in a well-organised, enriched and contextualised way by entity. Ultimately, one of the project's core goals has been to give audiences a browsing and search experience that is more intuitive, intelligent, personalised, semantically meaningful and effective.
And all of this while ensuring that this new knowledge discovery and content browsing experience works equally well across any device — so that all users can access what interests them and engage with RTVE Grafo however and whenever they choose.
Technology overview
Perhaps the most significant and far-reaching outcome of this digital project has been the consolidation of content from Spain's public radio and television into a single, unified knowledge graph — extensible, expressive, and queryable by both machines and people — enabling users to retrieve resources based on any interest or intent.
RTVE's ontological model
Consolidating all RTVE Play data into a unified knowledge graph required the design and development of a synchronisation process that collects data from RTVE's systems in real time, semantically annotates it according to an EBUCorePlus-based ontology and adopted term vocabularies (such as ESCORT 2007 — EBU System of Classification Of Radio and Television Programs), represents it as triples (predicative sentences in subject+predicate+object form), and deposits it in the semantic store — the graph database at the heart of RTVE's new semantic AI platform.
The ontological project developed for RTVE Play's Knowledge Graph has extended the EBUCorePlus domain ontology and hybridised it with general-purpose metadata schemas and vocabularies — including Dublin Core (dc) and schema — integrating them into a common ontological framework that represents the full range of activities in the audiovisual domain, understood here as the set of techniques, practices and processes involved in the operation of a broadcasting organisation.
Accurate semantic annotation required aligning existing content data from RTVE's databases with the classes and attributes defined in the RTVE ontology, and in some cases improving the metadata of that content at source. A crucial objective of the RTVE Grafo project was to enhance that metadata — that is, the way RTVE's content is tagged and described. More precise and detailed metadata makes information easier to search for and access. From an internal perspective, the project therefore set out to develop a refined system of semantic annotation and content representation that would close the gap between RTVE and the diverse range of audiences a public institution must serve and speak to. To that end, and beyond its public-facing use, RTVE Grafo is used to annotate, organise and present information in a meaningful way — gathering, for example, all relevant related information into each content record.
The RTVE Grafo search, query and information retrieval system operates within the technological landscape of semantically interpreted AI — an approach grounded in the ability of humans and machines to exploit the inherent possibilities of linked data within a knowledge graph. This is not only the condition that enables systems to interpret the knowledge generated by RTVE, but also to interconnect it and, in the future, link it to third-party sources that can further enrich and contextualise that content — thanks to the Contextual AI framework that RTVE Grafo provides.
Semantic standards and linked data
The RTVE Play Knowledge Graph has been built on semantic web standards and in accordance with the principles of the Linked Data Web, which has made it possible to:
- Connect Spanish Television's audiovisual resource management and documentation systems with the RTVE Grafo digital space.
- Optimise the use of those documentation systems, adding value to the work carried out across all areas of the organisation.
- Transform RTVE's information system into a Knowledge Graph expressed through a Linked Data Web.
- Develop querying and visualisation modes for that graph tailored to different audiences, designed to maximise the satisfaction of their interests by surfacing data explicitly related to the results that answer their questions.
- Build thematic web pages based on a dataset or subgraph meeting specific criteria.
- Create a semantically aware experience for exploring, discovering, querying and searching RTVE's content — enabling users to explore any topic related to the digital resources that make up the world of television in depth and in context.
All content on this site is represented and published in accordance with W3C standards for the semantic web and with the principles promoted by the Linking Open Data Project, with the aim of encouraging and facilitating the publication and interlinking of data on the web. As noted above, these semantic metadata generate a unified knowledge graph that is exploited primarily — though not exclusively — within the site itself, through its querying and recommendation systems, delivering a richer experience for users.
Impact
RTVE is conversations
RTVE is conversations. Personal conversations — because every user who visits is different, with different desires, different preferences, different aspirations. RTVE must be able to speak to all of them. Speaking with purpose, relevance and usefulness to each person who comes to the platform is the goal of the RTVE Play knowledge graph: a platform with a strong inclusive ethos that welcomes everyone — citizens, regular users, teachers, researchers, students and documentary makers — embracing their inherent diversity.
The RTVE Grafo digital project makes it possible to generate that narrative, that personal conversation with diverse audiences — delivering an exchange that is useful, contextual and rich, but above all personal.
The RTVE Grafo project has fundamentally transformed the way RTVE manages and presents its audiovisual heritage, setting a new standard for the Spanish audiovisual industry and creating tangible value for both the organisation and its users:
- Enhanced interoperability: The ability to exchange data whilst preserving clear and unambiguous meaning across heterogeneous systems.
- Unified access: All RTVE content accessible from a single query point.
- More effective search: A system that understands context and meaning, delivering more relevant and precise results.
- Personalised experience: The ability to generate useful, contextual conversations tailored to different audiences.
- Standards leadership: As a member of the EBU, RTVE is raising the bar for interoperability standards across the Spanish audiovisual industry.
- Foundation for advanced AI: An infrastructure ready to integrate with other artificial intelligence technologies and support the development of future services.