Extraggo

— distill knowledge from text and organize key concepts, domains and sentiment

Multilingual text analytics to get the gist of any text

Machine reading is all about enabling computers to read on a large scale. In order to achieve this ambitious objective, we need software that is able to get the gist of a text. But what is in a text? Wouldn’t it be fantastic if a system could:

  • Extract the key concepts and entities involved in a text
  • Interconnect them in an intelligent way
  • Obtain the key domain(s), i.e., fields of knowledge, the text belongs to
  • Assign a sentiment to the whole text, but also to individual paragraphs and sentences, and even concepts and entities as they are used in the text?

Thanks to its tight integration with WordAtlas and Comprehendo, Extraggo can process text and extract key insights from unstructured data, making it easy to perform semantically-boosted text analytics independently of the languages used in the input data.

Statistical and semantic extraction

Extraggo provides a seamless combination of statistical and semantic techniques. Statistical machine learning is applied to identify the key terms in the input text, whereas semantic text understanding, provided by Comprehendo, enables linking terms to concepts and entities as available in the WordAtlas knowledge graph. Terms, concepts and entities are ranked by importance, and the domain and sentiment information conveyed in the text are provided. Thanks to its tight interaction with Comprehendo and WordAtlas, Extraggo makes it possible to distill knowledge from text written in multiple languages, including non-standard text, such as bags of words or search queries.

Features

Extraggo comes with the following features:

  • Extraction and ranking of key terms mentioned in the text
  • Generalization of terms to concepts and named entities (e.g., different mentions to the same entity are grouped into a single item)
  • Concepts and entities are linked to WordAtlas, which enables language independence
  • Identification of the domains (i.e., fields of knowledge) of the text
  • Sentiment analysis performed at all levels: whole text, paragraphs, sentences and individual terms, concepts and entities
  • Cross-lingual semantic similarity is enabled

How it works

Extraggo takes as input a text (or a collection of texts) and ranks the key concepts and entities, as well as domain and sentiment information, as shown below:

  • Europe LOCATION
  • Schiphol LOCATION
  • Brusseles LOCATION
  • airport CONCEPT
  • flight CONCEPT
  • transport CONCEPT
Snow in Europe triggers transport chaos
Heavy snow blanketing northern Europe has caused many flight cancellations and delays at Schiphol airport in the Netherlands and Brussels airport. Nearly 300 KLM flights were cancelled at Schiphol, while Brussels airport scrapped at least 50. Travellers have been advised to check flight updates at home, rather than set off for the airport in bad weather. In Germany the heavy snow has caused many car crashes and traffic jams, as well as train delays. More than 300 flights were cancelled on Sunday at Frankfurt airport, the busiest in Germany. In France 32 regions were put on an emergency footing, as snowstorms battered coastal areas and cut power to thousands of homes. About 80,000 homes lacked electricity in the Loire Valley on Monday, the daily Le Parisien reported. Not what you expect in Venice: snowflakes on the gondolas. Snowstorms have also spread southwards to Italy, causing some travel chaos in northern regions. The snow caused the closure of schools in Liguria, Piedmont and Tuscany, Italy's La Stampa daily reported. Ferry services to the islands off Naples were suspended because of strong winds. After heavy snow fell in the UK on Sunday thousands of homes were left without electricity and hundreds of schools were shut on Monday. Road conditions were described as treacherous in many areas.
Document domains Transport and travel Meteorology