Ambiverse: Text to Knowledge

Ambiverse is a spin-off from the Max Planck Institute for Informatics, creating solutions based on the AIDA technology. Ambiverse provides solutions for automatic text understanding and intelligent text production and is supported by the German Federal Ministry for Economic Affairs and Energy with an EXIST Transfer of Research grant.

If you are interested in using the AIDA Web Service in a commercial scenario, have a look at the Ambiverse Natural Language Understanding API. The Natural Language Understanding API provides a faster, more accurate, and commercially supported service based on the AIDA technology.

AIDA JSON Web Service

AIDA is available as a web service that can recognize names and disambiguate them to Wikipedia (and YAGO) in any text you send. The resulting annotations will be returned as JSON.

Please do not use it for comparison in scientific papers or for running time experiments, as the service changes continuously. If you want to compare AIDA for research, please download it and set it up on your own machines.

Accessing the Web Service

To access the web service, call the following URL:

https://gate.d5.mpi-inf.mpg.de/aida/service/disambiguate

It expects a single POST parameter, text. You can call this for example with curl

curl --data text="Dylan was born in Duluth." https://gate.d5.mpi-inf.mpg.de/aida/service/disambiguate

Additional Web Service Parameters

If you want to configure the kind of disambiguation that is done on the web service, you can pass additional parameters to the web service:

text
The input text to disambiguate
type
The type of the input text (TEXT, TABLE, XML - default is TEXT)
tag_mode
Set to 'manual' to give AIDA pre-defined mentions as part of the text input (marked with [[..]])
doc_id
Specify the document id.
tech
The technique to use (LOCAL or GRAPH - default is LOCAL)
algo
Algorithm to use when GRAPH is set as tech. Default is CPSC (size constrained).
alpha
alpha is multiplied to ME edges, 1-alpha to EE edges (in [0.0, 1.0])
ppWeight
Weight to balance the prior probability of mention-entity pairs and contextual similarity.
entities_per_mention
Number of candidates to use in the CPSC setting.
coherence_threshold
Threshold to use for the coherence robustness test (in [0.0, 2.0])
exhaustive_search
Set to false to not do a exhaustive post-processing after the graph algorithm for selecting the correct entity. Default is true.
fast_mode
Set to true to cut down on the number of keyphrases used per entity candidate, using only the most specific ones. Speeds up processing with little impact on quality.
filtering_types
Semantic (YAGO) types to restrict the entity candidates to. Format: KB:typename,KB:typename,... . The most common types are YAGO:wordnet_person_100007846 for person,yagoGeoEntity, wordnet_organization_108008335 for organisations, wordnet_event_100029378 for events, and wordnet_artifact_100021939 for artifacts
maxResults
Number of entity candidates per mention to include in the returned JSON object.

Description of the returned JSON

This call returns a JSON formatted document containing the annotatedText:

[[http:\/\/en.wikipedia.org\/wiki\/Bob%20Dylan|Dylan]] was born in [[http:\/\/en.wikipedia.org\/wiki\/Duluth%2C%20Minnesota|Duluth]].

The annotations are formatted as Wikipedia style links: [[wikipedia-link|name]]. The JSON also contains a mentions array mentions pointing to the character offsets in the originalText.

The full output looks like this:

{
  "allEntities": [
    "YAGO:Bob_Dylan",
    "YAGO:Duluth\\u002c_Minnesota"
  ],
  "formatVersion": "2.1",
  "originalText": "Dylan was born in Duluth.",
  "jsTypeInfo": "",
  "cleanedText": "Dylan was born in Duluth.",
  "originalFileName": null,
  "docID": "1406646355825",
  "annotatedText": "[[http:\/\/en.wikipedia.org\/wiki\/Bob%20Dylan|Dylan]] was born in [[http:\/\/en.wikipedia.org\/wiki\/Duluth%2C%20Minnesota|Duluth]].",
  "overallTime": "33",
  "mentions": [
    {
      "allEntities": [
        {
          "kbIdentifier": "YAGO:Bob_Dylan",
          "disambiguationScore": "0.09367"
        }
      ],
      "bestEntity": {
        "kbIdentifier": "YAGO:Bob_Dylan",
        "disambiguationScore": "0.09367"
      },
      "name": "Dylan",
      "length": 5,
      "offset": 0
    },
    {
      "allEntities": [
        {
          "kbIdentifier": "YAGO:Duluth\\u002c_Minnesota",
          "disambiguationScore": "0.1375"
        }
      ],
      "bestEntity": {
        "kbIdentifier": "YAGO:Duluth\\u002c_Minnesota",
        "disambiguationScore": "0.1375"
      },
      "name": "Duluth",
      "length": 6,
      "offset": 18
    }
  ]
}