Case2: Resources Extraction

[djarecka]

his is the use case from the list in #23

This place will be to discuss ideas around the use case, especially around the 3 main components: ingestion, public view and feedback. Below are notes from our initial discussion on August 27.

Ingestion

• should be public
• we should be able to submit papers and an additional processing function (structsense based) should be run on a PDF
• perhaps feedback during the ingestion process

Public View

• we can group the resources, e.g., datasets, software

Feedback

• feedback can be on the quality of the resource

[tekrajchhetri]

Continuing the discussion:

1. We want the public view displaying the resources information as in https://brain-bbqs.org/bbqs-kg/
2. Ingestion: Currently, we plan to support the following ingestion modes (doi, text, JSON and PDF).

Upon extraction, user will be able to review before saving.

[satra]

there also needs to be an interface for post integration review/edit. it would also be useful to describe how schemas will be generated. for example the models/datasets mentioned are themselves resources, and thus should have a way for enriching those as well.

[tekrajchhetri]

Cross-posting from Slack — here’s a version I’m considering for the public-facing view. This should address our earlier discussion (with @djarecka and @puja-trivedi) on how to display provenance information and what to do if someone wants to review prior versions. Moreover, I believe we can adopt this view for all of our use-cases.

Note: The data you see is a dummy data.
Initial view

Detailed view: this view will appear when click on open button. By default, we will load the latest version. We will also provide opportunity to load different versions and also compare it. When you click on the related card, it will be loaded.


For public feedback: This is the simple one but we could also make each fields editable and capture it. We can decide which way we want to go.

[tekrajchhetri]

Small update on display of the information. Now we provide the detailed information. Note: All data shown is placeholder content for design purposes only

• tab view so that information can be displayed in clean manner without cognitive overload.
• when you load different version, color highlight to show what's updated
• link to publication and clinical significance if any

when no information is available we maintain the consistent design

cc: @satra @kabilar

[tekrajchhetri]

Moving @satra's comment from slack on human-in-the-loop process. Our workflow should have the following human-in-loop process: Create → Review → View → [Update ↔ Review → Publish].

@djarecka @puja-trivedi

[tekrajchhetri]

Phase 1 of entity extraction is completed. StructSense has been integrated + user would be able to view the extracted data.

Current flow - Extract -> Review -> Submit -> View. Todo: Feedback -> Review -> Update.