metaSENSE Natural Search APIViewers search the way they talk. metaSENSE understands the meaning of every request and returns the right titles from your catalogue, each with a clear reason why.Request API accessSee it in action

metaSENSE Natural Search API: semantic search for TV and streaming

The metaSENSE Natural Search API lets viewers search in plain language. It understands the meaning of each request, matches it on metaDNA story profiles and returns ranked titles from your catalogue with the reason for every match, in any language.

Cineverse logoOoredoo logodu logoDeutsche Telekom logoAzerconnect logoBalticom logoCineverse logoOoredoo logodu logoDeutsche Telekom logoAzerconnect logoBalticom logo
See it in action

One sentence. The right titles.

A viewer types or simply says what they feel like watching. metaSENSE works with typed and voice search, reads the request, matches it on metaDNA story profiles from the Metaprofile Entertainment Database and returns ranked titles with the reasons behind every match.

01 · Viewer types or speaks
02 · metaSENSE understands
TypeFilmGenreThriller · GE-028MoodTense · MO-0001SettingJungle · ST-00075
+ survival thriller+ action thriller400+ titles matched
03 · Results with reasons
FILM · 2023 · tvID 14719414

Plane

Score 175 · coverage 3 of 3
ConceptCentrality
Thriller4
Tense5
Jungle3
FILM · 1985 · tvID 1160

Rambo: First Blood Part II

Score 163 · coverage 3 of 3
ConceptCentrality
Thriller4
Tense3
Jungle4
FILM · 2001 · tvID 8797

Jurassic Park III

Score 163 · coverage 3 of 3
ConceptCentrality
Thriller3
Tense4
Jungle4
FILM · 2025 · tvID 31666511

Anaconda

Score 139.5 · coverage 3 of 3
ConceptCentrality
Survival thriller3
Tense3
Jungle4
FILM · 2026 · tvID 34575328

The Bluff

Score 139 · coverage 3 of 3
ConceptCentrality
Thriller3
Tense4
Jungle2
FILM · 1997 · tvID 469710

The Lost World: Jurassic Park

Score 138 · coverage 3 of 3
ConceptCentrality
Action thriller3
Tense4
Jungle3
Example results from metaSENSE · Title data from the Entertainment Database
The problem

Viewers don't think in genres and keywords

Operators bring linear channels, catch-up, VOD and streaming apps into one interface. Yet most search boxes still match titles, actors and directors, and browse rows keep promoting the same popular titles. Good content stays hidden, the evening turns into scrolling, and frustrated viewers leave.

Keyword search

  • Matches words in a title or cast list
  • Built on schedule data, not on what a title feels like
  • Cannot answer "a warm, slow-burning film"

Natural Search API

  • Understands the meaning of a plain-language request, typed or spoken
  • Matches mood, theme, story and style through metaDNA
  • Shows its interpretation, so viewers can refine in one step
How it works

From a sentence to the right title

01

Understand the request

Large language models read the request, typed or spoken, the way a friend would, tolerate small typos and split it into content type, feelings and themes. "Thriller" also finds survival thrillers and action thrillers.

02

Match on metaDNA

Every title in the Metaprofile Entertainment Database has a metaDNA profile: themes, narrative arc, feelings and moods, intensity, visual and narrative style, scene-level elements, keywords, genre and sub-genre.

03

Rank by relevance

Titles that match more signals rank higher, especially when the requested theme drives the story. A film where the jungle drives the story beats one with a single jungle scene.

For developers

One request. Results with reasons.

Send the viewer's words as they typed them. The API returns its interpretation and a ranked list of titles from your catalogue, each with the metaDNA tags that explain the match.

Request
GET /v1/natural-search
  ?q=A tense thriller movie with jungle scenes
  &lang=en
  &catalogue=your-platform
Response
{
  "interpretation": {
    "type": "film",
    "concepts": ["GE-028 thriller", "MO-0001 tense", "ST-00075 jungle"]
  },
  "total_matched": "400+",
  "results": [{
    "tvID": 14719414,
    "title": "Plane",
    "year": 2023,
    "score": 175,
    "coverage": "3 of 3",
    "why": [
      "thriller · exact · centrality 4",
      "tense · exact · centrality 5",
      "jungle · exact · centrality 3"
    ]
  }]
}

Illustrative example. Endpoint and fields shown for presentation.

Language-independent

One concept. Every language.

Each metaDNA tag has a language-neutral ID. A request in English, Arabic or Croatian maps to the same concept, so new markets launch without re-tagging the catalogue country by country.

jungle EN
džungla HR
أدغال AR
metaDNA conceptST-00075 · jungle
Built to be trusted

AI that interprets. Data that decides.

The language model only reads the request. It never decides what a film is about and never picks titles from general knowledge. Results come from metaDNA profiles in the Metaprofile Entertainment Database, which Metaprofile normalises, enriches and checks.

LLM

Reads the request

Intent, feelings, themes, typos.

→

Checked profiles

Every reason traces back to a tag.

→
Catalogue

Your titles only

Every result exists. The same request returns the same answer.

Who it's for

Better discovery for every screen

Telcos and pay-TV

Discovery as a differentiator. More engagement and less churn, without building an in-house AI team.

Broadcasters and streamers

Back-catalogue, local and niche titles found on what they are, not on recent popularity.

Product teams

One integration, language-independent tags and faster launches in new markets.

Viewers

Less scrolling, results that fit the evening's mood and new favourites, in their own language.

Get started

Try Natural Search on your catalogue

Request API access and see results on your own titles.