Search Result Priorization

[rsommerfeld]

Especially after introducing search results from #111, the number of results rose sharply and search result order got less intuitive. As somebody looking for search queries by their relation to other entities, I would like better search result ordering to find my results faster.

Pre-Context:

• There are a bunch of relational search results for a query.
• Maybe PageRank or other publicly known simple search algorithms might be helpful to look at.
• It might also be useful to look at frequencies or things like that.
• Could also spike using an external search engines, with strong focus on being able to deploy.

Post-Context:

• The acceptance criteria of this story are only about priorizing search results. It would also be useful to filter off bad search results and not even show them. We can cover that in another story and only focus on bringing the best results to the top of the list in this story

Acceptance Criteria:

• The query "chair representative Enterprise Platform and Integration Concepts" produces Perscheid as the first result
• The query "room perscheid" produces the office of Perscheid (or whatever room is assigned to him)
• The query "Professor Enterprise Platform" produces Hasso Plattner as the first result
• The query "H-2.3.2" shows Prof. Baudisch (it is his office) within the first few search results
• The query "Room for Working" still shows rooms with the working tag first.
• Matching entire words ranks higher than matching partial words. (Example: "Lounge For" shows Platform over a room with tag lounge, because for matches in Platform)

 

• The query "tutor Internet Security" shows a tutor of the course ISEC as the first result if courses and course-lecturer relations are implemented at the time of this user story.
• The query "main building printer room" shows a printer room in the main building, if search supports searching for building name by the time this user story is implemented.

If not all of these can be achieved simultaneously that is also okay, as long as the priorization is close to the desired effect 😅