As discussed, we will be developing some sort of visual query builder. To shield from the complexity and time needed to build a fully fledged interface, this should be a collaborative effort. This milestone covers the basic requirements of a useful query builder.

Documentation items are a must-have, stubs are based on time available and ranking by "bang-for-buck" and collaboration with @timrepke .

Documentation

We need some documentation for users (pro users—those who could write SQL; and other users—those who would normally use WoS query builder).

• easy to understand description of basics
  • basic logic of websearch_totsquery (|,&,<->,",...)
  • behaviour of quotes
  • ...
• Brief introduction to the basic SQL statements and gotchas, ideally as part of https://gitlab.pik-potsdam.de/mcc-apsis/nacsos/nacsos-docs (ask @mhansen)

Helpful stubs for query builder

@timrepke will build an interface for NACSOS, any help towards that is greatly appreciated, e.g.:

• define schema to represent query logic to abstract SQL
• query builder (from json schema [to be defined] to prepared SQL statement, maybe including virtual fields like title&abstract that convert to a join)
• translate publication in OpenAlex to a dict following the AcademicItemModel schema definition (see https://apsis.mcc-berlin.net/nacsos-core/docs)
• on-the-fly query statistics to (heuristically) estimate the scope of a query (e.g. by hinting the user of how common terms contained in the query are or how many papers are published in the selected years; whatever postgres data profiling gives us)
• query overlaps (only left set, inner, only right set)
  • on-the-fly statistics, such as relative time histogram difference, overlap of authors, overlap of venues, ...; whatever is fast and easy;
  • sample of query results for the sets

Interface sketch

The interface will probably be somewhat similar to this: https://developer.twitter.com/apitools/query

Roughly:

• add search parameter
  • select field
  • based on type, add conditions / value
• add group
• for group/parameters, select logical relation (AND/OR)