The jinkō platform offers a powerful set of features allowing for the human-curation of scientific knowledge behind a computational model and/or a research project.

Use them to perform a collaborative knowledge review that will be easily accessible for your R&D teams, trial managers, external experts and auditors, so they can fully understand on which scientific foundations your models and trials are based on.

Add references to your project

• From the menu 'create new', select 'Reference' and upload a PDF version of a scientific article of your choice. The meta-information related to the article, including the DOI, title, authors and date of publication, will be automatically extracted from the PDF and populated during the upload.
• You can add several references at once before uploading them in batch.
• Shoud you not have a PDF available, you can also create a reference manually, by entering its DOI or other information depending on the type of the article. You will always be able to attach the corresponding PDF later. 

Extracts: create pinpointed pieces of knowledge for documentation

You are now able to review your references and create what is called in jinkō 'extracts', which are pinpointed pieces of knowledge relevant for a given research, that can then be references throughout your project, in documentations and models notably.

Going back to the PDF of the reference we just uploaded, there are two ways of creating extracts in jinkō: 
 

     -> When wanting to select textual information: use the highlighter

 

     -> When wanting to select images, tables or equations: use a rectangular selection 

As you can see on the screen recordings above, your extracts are displayed on the right side bar. They are stored as objects that can now be used throughout your projects, as described in the following section. 

Adding extracts to documents and / or models 

Extracts can now be referenced to document your research and create transparent and traceable models, where equations are linked to verifiable knowlege foundations. 

      -> In your reports and research documentation, drag and drop extracts from the right-side panel: 

     -> In your models, copy the extract link in the parameter metadata: 

Classifying and adding scores to your scientific extracts

Nova's approach to the collective curation of scientific knowledge is also reflected in jinkō's throught its extracts scoring capabilities: an (optional) evaluation of the quality called 'Strength of Evidence' (SoE) of a piece of knowledge, or of a proposed explanation, extracted from a scientific publication. This evaluation is used to inform how reliabily it is possible to use a piece of knowledge in the construction of a model, and to make it easier to scan extracts that need more attention, notably to ask review to external experts. 

When scoring an extract, you are performing two actions at the same time:

• 1. Classify the extract, depending on its nature:

  • Statements are pieces of knowledge carrying scientific information
  • Hypothesis are proposed explanations made on the basis of limited evidence
  • Data are experimental data (parameter, variable, values..), figures or tables

• 2. Score the extract: a qualitative evaluation of the Strength of Evidence of the extract, which criterias vary depending on its nature. Follow the grid (see screen capture above) as a helper for the evaluation. 

A few last tips

• Name your extracts to help their identification, you can give a name to your extracts from the right side panel.
• Reformulate the original text extraction: you can reformulate the extraction so it better fits your documentation(s). The link to the original extract is of course always kept nonetheless. Note that you can also add a formulation to rectangular extracts. 

• Enrich extracts with multiple pieces of knowledge (from the same reference)

It is possible to compose extracts from multiple pieces of knowledge (called 'highlights' in this context in jinkō) from a same reference. When doing so, once an extract inserted into a Document, the associated highlights are immediatly visible in the sidebar: