Skip to main content

Guide

Our Media API offers the following core functionalities:

  • Article Analysis
    We perform a detailed linguistic and semantic analysis of the article text, returning:

    • Entities (details) – These include proper names (e.g., people, locations, organizations, products, events) and general concepts (also called keywords), such as flu season or income tax, mentioned in the article.

    • Derived entities – These are not explicitly mentioned in the text but can be inferred using our knowledge base (see below).
      For example:

      • If Berlin is mentioned, we also return Germany.
      • If Toyota is mentioned, we may return automotive industry
        See the section on derived entities for details.

    • IPTC Media Topics – An industry-standard taxonomy for categorizing articles.

      • Contains 1200+ categories (e.g., sport, basketball, music, classical music)
      • Organized in a 5-level hierarchy
        For more information, see this article.

    • Semantic tags (details) – The most important entities and categories describing the article's content.

    • Sentiment analysis (details) – We detect sentiment at different levels:

      • Whole document
      • Individual sentences
      • Individual entities

      This helps identify how the author feels about specific subjects mentioned in the article.

  • Recommendation of Related Content
    Based on article content, we recommend:

    • Relevant photos (details) from a variety of sources:
      • Our partners
      • Your licensed providers
      • Third-party photobanks
      • Public images (e.g., Wikipedia)
    • Related articles – Other articles about similar topics, based on semantic similarity and relevance

  • Knowledge Base (details)
    All tags and entities are linked to the Geneea Knowledge Base (GKB, Geneea KB).

    • GKB integrates:
      • Open data (Wikidata, DBpedia, OpenStreetMap, company registries, etc.)
      • Geneea's proprietatry data
    • It supports:
      • Custom properties (e.g., your internal identifiers)
      • Private/custom items only visible to you

  • Localized Output (details)

    • Entities and tags can be localized (i.e., returned in your desired language)
    • Useful when working with multilingual content or publishing in different locales

  • Feedback-Based Learning (details)

    • The system can learn from feedback to improve quality and adapt to your preferences
    • Feedback can be automatic or manual (e.g., from editors in your CMS)

  • Integration

    • Commonly integrated into a publisher's CMS
    • Editors can:
      • Review and confirm suggestions
      • Provide feedback
    • The API can also run in fully automated pipelines, e.g., for news aggregators or archives

  • Customization
    The API is typically tailored to your use case:

    • Custom entity IDs, labels, tag types, or tag communicates
    • Preference tuning
    • Creation of new tags or types
    • Integration with internal data or taxonomies