Background

In previous quarters, we’ve made steady progress improving how Wikifunctions accesses content from Wikidata. This quarter, we’re building on that foundation to enable the creation of actual sentences in multiple languages on Wikifunctions using Wikidata.

Approach

• We provide a pre-defined & built-in function that will fetch (if available) a list of Lexeme IDs based on a Wikidata Item ID, Wikidata Property ID, and Language, based on the new search API built in Nov/Dec 2024

• Create a new UX component for Wikidata Property (similar to existing ones for Wikidata Item?)

Acceptance Criteria/Success Metrics

• Users can use functions to create the sentence “She is a [country] [professional]” in 3–5 of our target languages (it does not time out for typical uses)

Bengali [bn], Malayalam [ml], Hausa [ha], Igbo [ig], Dagbani [dag] + (demo purposes) en

• Performance – e.g. each Lexeme lookup for a QID takes at most 1s total. Covered in the Q3 tech debt goal
• Types exploration – document end user outcomes to give framing to types exploration

Success metrics:

• Organic increase of at least 5 multilingual functions that generate natural language sentences
• Will continue to monitor this
• Feature is performant enough for majority of functions (no long hang times, does not time out)

Stretch Goal:

• [UX] Create compressed views for Wikidata entities in the collapsed function call component view, to identify the Wikidata entity (by label or id) and skip the verbose Wikidata fetch function call labels, etc.