Aktuelles Ziel

Recommender System - Entscheidungsbaum (Policy) (Decision tree ist abhängig von Institut, Universität, Domaine, Fördergeber) -> NFDI4Ing, Institut (WZL, und zwei weitere, Sahar/Stephen), Domaine (Januar), Fördergeber (DFG, BMWF, vllt 1 weiteres)

Idea

Create a recommender system that allows researchers to easily get information what tools/trainign materials/services etc. they should use.

For each Project, the following information is needed:

• Funding Organisation
• Institution
• University
• Discipline --> by DFG numeration and terminology. See Coscine!

(last two might be overkill). This information should not change when looking at the same person. So maybe include this info in the user info and fetch it from there. See Coscine

Note: This does not allow for multi-university or multi-institute policies... Funding organisation should chatch most of that.

Each of those has a individual RDM policy. Additionally, Jarves has its own default policy that is always there to act as a fallback.

Policies should come in two spices. Hard and soft.

• Hard: These rules must be obeyed, no matter what.
• Soft: Recommendations on how to improve RDM even if e.g. the Funding Institution does not require it.

Each Project has its very own hard and soft policy (in sum there are two) that will be created from the combination of guidelines.

The prioisation

We need an algorithm that creates a hard and a soft requirements policy for each project.

So basically the rules are as followed:

1. Jarves is default. This is soft. Write to soft policy.
2. Go to Institution/University hard policy and write hard requirements to hard policy.
3. Go to Institution/University soft policy and overwrite soft requirements by Jarves to soft policy.
4. Go to Institute hard policy and overwrite hard requirements by University to hard policy.
5. Go to Institute soft policy and overwrite soft requirements by University to soft policy.
6. Repeat 4. and 5. in the following order: Departement, Group, Funding Organisation

Database Structure

Scribble: So könnte unsere Struktur aussehen:

Siehe auch: Excel Tabelle

---

Just kiddin. Will be created automatically later down the line.

Following rules apply:

• Each Project has 1 soft policy
• Each Project has 1 hard policy
• Each Group has 1 Institution/University, 1 Institute, 1 Departement
• Each Departement has 1 Institution/University, 1 Institute
• Each Departement can have multiple Groups
• Each Institute has 1 Institution/University
• Each Institute can have multiple Departements and Groups
• Each Institution/University can have multiple Institutes, Departements and Groups
• Each of those has 1 soft policyand 1 hard policy

Models needed

For the policies: (Might be overkill)

• Projects_Hard_Policy --> This will be the info that jarves creates recommendations from
• Projects_Soft_Policy --> This will be the info that jarves creates recommendations from
• Institution_Hard_Policy
• Institution_Soft_Policy
• Institute_Hard_Policy
• Institute_Soft_Policy
• Departement_Hard_Policy
• Departement_Soft_Policy
• Group_Hard_Policy
• Group_Soft_Policy
• Funding_Organisation_Hard_Policy
• Funding_Organisation_Soft_Policy

For the Organisational structure

• Institution
• Institute --> Already exists!
• Departement
• Group
• Funding_Organisation
• User or correlating.

Approach

1. Add a sketch of how tables/models could look like.
2. Implement them in the backend, creating the UML automatically through the pipeline (already implemented).
3. Add Institution/University, Institute, Departement, Group as 1to1 fields in the user model.
4. Add frontend functionallity to add these informations: Dropdown to choose Institution/University next dropdown for Institute then only shows institutes from this university and so on. Institution/University is mandatory, the others are optional.
5. Write this info to the database.
6. ... Time's up i will continue later...