Iris.ai

Iris.ai is a developer of artificial intelligence (AI)-based research assistance tools. These tools work on open-access research papers and use a combination of keyword extraction, word embeddings, neural topic modeling, and artificial intelligence to review research papers, find documents, extract key data, and identify pieces of knowledge. The company's platform uses natural language processing and machine learning to review collections of research papers or patents and to help research and development professionals or students to spend less time researching.

Iris.ai was founded at Singularity University in the summer of 2015 by Anita Schjøll Brede, Victor Botev, Jacobo Elosua, and Maria Ritola. The company is headquartered in Stabekk, Norway, with additional offices in Sofia, Odessa, and Denmark.

Iris.ai has developed its AI engine for scientific text understanding based on algorithms for text similarity, tabular data extraction, domain-specific entity representation learning, entity disambiguation, and linking. These technologies work together to build a comprehensive knowledge graph, which contains appropriate entities and links to allow the user to learn from the knowledge graph and offer feedback to the system to help the system to continue to develop.

Iris.ai's approach, in this case, is to blend algorithms to create "document fingerprints" capable of reflecting word-usage frequency, and using this, rank the papers according to relevance. The system is capable of summarizing abstracts and relevant table information from these papers, while also developing technology to check aspects of a research paper against related documents to better validate the use case of a given paper or hypothesis.

This technology goes into Iris.ai's tool to offer a workspace that offers content-based search, allowing users to explore research papers with a content-based recommendation engine; context filtering; data filtering, to allow users to filter based on information extraction from documents; extraction and systematizing of data; document set analysis, allowing users to analyze a set of documents; summarization, in which single or multiple documents can be quickly overviewed; and monitoring and alerts, for when a system reruns searches, filters, and extractions.

The workspace offers users a single place to find their research, where they can upload research documents or connect to proxy data sets. Once these contents are added, users have access to the tools that can be applied to the research to extract the necessary information, based on users' needs. The machine learning engine behind the workspace is able to adapt to the research domains of the user, through uploading ten to twenty representative documents relevant to the field of research. The Iris.ai workspace also offers a variety of datasets that users can load into their personal workspace, including most open-access content from PubMed or the US Patent Office, among others.