Cognitive Text Mining

As much as 90% of all data generated today is unstructured and text-heavy, including publications, images, and physician narratives to name a few. With over 40 zettabytes (40 trillion gigabytes) of unstructured data in the digital universe by 2020, it would be impossible for any group of humans to aggregate, curate, and structure this amount of data using traditional keyword-based approaches. iPhronesis overcomes traditional keyword-based approaches Cognitive Text Mining.

Cognitive Text Mining is the application of Artificial Intelligence (AI) technologies like Natural Language Understanding (NLU) and Machine Learning (ML) to retrieve, study, determine relevance, annotate, extract, recognize patterns, and identify associations in unstructured data; quickly structuring that data into an actionable and reproducible format ready for analysis.

Reduces the time required to identify, retrieve, and determine relevance of documents of interest.

Accelerates the consumption of knowledge by automatically identifying, reading, and analyzing documents of interest.

Reduces errors by automatically extracting and harmonizing unstructured data.

Understands complex scientific language and is able to identify related concepts – even if those concepts are expressed differently thanks to NLU and an extensive Ontological library.

Continually learns, improving output with each human interaction thanks to sophisticated ML techniques.

Transforms unstructured data into a structured format that is actionable, reproducible, and ready for further analysis.

What is Natural Language Understanding?

NLU is the ability for computers to understand human language as it is spoken or written. Once iPhronesis understands the context of the information it is searching, it is quickly able to identify and structure relevant data into an actionable and reproducible format.

Context is everything in natural language. For example, in math, the number "1" is clearly defined, but in natural language, "one" can mean a number, an abstracted person, or a scatological reference.  In the spoken word, "one" and "won" sound the same.  Clearly, context means something in natural language.
  • iPhronesis uses a large set of thesauri and ontologies that define over 65 million unique concepts within the life sciences domain. These concepts are used to infer meaning from unstructured, natural language sources, such as scientific literature, patents, electronic health records (EHRs), clinical trials data, news feeds, social media, and proprietary content.

What is Deep Machine Learning?

Identifying relevant data in context and structuring it, only solves half of the problem. How does the user validate the information retrieved is accurate? This problem is solved by deep Machine Learning. Machine Learning is the ability for a machine to learn about the data it retrieves from the regular input of human operators.

  • iPhronesis uses Machine Learning to build predictive algorithms, forming patterns discerned from large datasets, in order to produce answers to relevant questions.
  • iPhronesis uses an integrated feedback learning module to facilitate the Machine Learning functionality.
  • With each interaction, iPhronesis better understands what users are looking for, improving productivity across initiatives.
  • Organizations can expect to improve efficiencies and help reduce costs by implementing this type of machine learning technology.

What are Ontologies?

Ontologies are sets of continually-updated, standardized vocabularies that define terms and logical relationships between them, in each vocabulary. Ontologies can be used to capture relationships, "sometimes associated" relationships, and temporal relationships.

  • iPhronesis has an embedded ontological library of over 65 million unique concepts that allow the user to mine through complex relationships using everyday language and terminology.