Features

Research faster, better. AI charged.

The Iris.ai products are process tools aimed specifically at researchers in the early phase of a new project. They are especially suitable for interdisciplinary projects where the combination of knowledge from across a range of research fields will be vital to the project’s success.

EXPLORE

FOCUS

ROADMAP

WHAT’S NEXT

THE EXPLORE TOOL

Exploration of interdisciplinary research

Consistently outperforming old school search tools, Iris.ai starts from a paper of your choice, “fingerprints” it based on machine extracted keywords, contextual synonyms and hypernyms, and matches the fingerprint against >83M Open Access papers.

Bypass Keywords

Traditional keyword searches limit you to what you already know. Great when you know what you’re looking for, but a big problem when you don’t.

Bypass Citations

The citation system is great for learning about researcher networks. For finding new solutions they can be unhelpful at best, or introduce popularity bias at worst.

Navigate papers visually

Endless result lists are terrible for quickly getting an overview. The Iris.ai maps give you a visual overview of the topics for a much faster distilling of the content.

Bookmark papers

Find something interesting? Bookmark it into your reading list to dig into it later.

Download full text directly

Found exactly what you’re looking for? Download the full-text Open Access directly.

Access full history

Connecting the dots with something you saw ten papers ago? Your full history is saved!

THE FOCUS TOOL

Full research landscape mapping

Use the Exploration tool starting from your free text description of the problem, and a result editor, to build a large corpus of documents related to your problem statement. Use the iterative Focus tool to narrow down to a precise reading list.

Incorporate additional databases

The Iris.ai tools can be connected to most scientific text-based content, including from third-party providers as well as internal company documents.

Exploration starting from full-text

Using a self-written problem statement as a starting point allows far more flexibility in the Exploration phase.

Edit exploration results

Premium users are given the option to iteratively edit the results in the Exploration phase, offering far more flexibility and customized results.

Iteratively narrow down results list

The Focus feature allows the user to iteratively narrow down a corpus of up to 20,000 documents, to a short comprehensive reading list.

Corpus based criteria suggestions

In the Focus process, the tool suggests topics for inclusion and exclusion based on the initial corpus, facilitating a rapid and precise iterative filtering.

Manual precision/recall validation

With optional manual verification, the user can measure the precision and recall of the filtered corpus – building confidence with the results.

Ready to get started?

Learn more about premium or start with free account

Premium Free

FEATURES ROADMAP

The road towards an accessible science world

The Iris.ai products semi-automate the existing research process – currently for the first two steps of the process: Broadly exploring the topic for an overview, and then narrowing down to a precise reading list.

Today this is a 2-4 week process in industry, with a self-reported 70% accuracy: researchers know they are missing out on important knowledge, and waste a lot of time just trying to find it.

Using the Iris.ai tools you can now do this in a systematic manner, modelled after the rigorous academic systematic mapping study – and you can achieve up to 85% precision and recall (“accuracy”) in just a couple of days of dedicated work.

STEP 1
Exploration phase

STEP 2
Focus phase

STEP 1
Exploration phase

STEP 2
Focus phase

Write out the problem you are trying to solve in your own words.

Use 300-500 words, and make sure to cover the problem from multiple angles.

Iris.ai uses this text to build a fingerprint of the problem, matches it against >120M research papers, and distributes them into topics.

Use the visual maps to gain an overview of your problem.

Build new maps from other articles or new text where more content or deeper exploration is needed.

“Jump into the rabbit hole of science” as a user once called it.

Use the hierarchy editor to fine-tune your map results.

Allowing you a bit more control over the exploration, the hierarchy editor lets you explore the fingerprint; merging words, selecting synonyms and deleting concepts that are currently irrelevant to you.

Bookmark papers and maps for later visits.

Build a reading list of specific papers you wish to read later.

Bookmark full maps for use in the next phase with the Focus tool.

Pull together your corpus

Import the corpus you built in the exploration phase, or use an existing set of documents from elsewhere.

As you select to import it, the machine reads every paper and builds a fingerprint of each one and models the topics of the combined corpus.

Select include/exclude concepts

Presented with a list of machine identified concepts, select what to include and exclude to narrow down your initial corpus. You can choose from a general list of concepts, highly relevant words as well as rare concepts.

The tool errs on the side of caution, with ‘include’ criteria overriding ‘exclude’ criteria in the case of overlap.

Select include/exclude topics

Iris.ai has identified topics, described using a string of 10 words, that the papers can be grouped into. Each paper can fall into multiple topics – now you use the topics to make include/exclude criteria of what is and isn’t relevant to you.

The topics are modeled both from a general list of 5M papers, from your specific corpus and from the now filtered-down corpus, for increasing specificity.

Manually verify the results

Through the process, you can choose to manually verify both the included and excluded list of papers, and this input is used to measure the precision/recall metric of your results.

Once you are happy with the results, you can export them as a .csv file or as a visual Iris.ai map.

WHAT’S NEXT?

Currently in the Iris.ai R&D pipeline

We’re currently working on the next steps of the technology, which is what we call hypothesis extraction: giving the Iris.ai machine the ability to extract the core information of a paper in the form of a hypothesis structure (problem – solution – evaluation – result). This will create a number of interesting features such as:

Highlighting which section of the paper is most relevant to your problem statement, and why.
Allowing you to build your process only focused on e.g. method sections, or only problem areas.
Show you a structural summary of the hypothesis of a paper.
… and stay tuned for more!