1 of 9

​

​

​

Wikidata Powered

Language Keyboards

Andrew Tavis McAllister

Initiator and lead developer

DATA RE–

USE DAYS

14 - 24 March 2022

2 of 9

Schedule

​

  • Presentation (30 min)
    • Short Introduction
    • iOS Demo
    • Data Update Process
    • Android Status and Demo
    • Future Plans
  • Discussion (rest of session)

* Potential tasks will be discussed along the way

* * Questions always welcome!

3 of 9

Overview

  • Keyboards for second language learners
  • “Never leave your keyboard”
  • Noun genders, verb conjugations, translations, preposition cases
  • French, German, Italian, Portuguese, Russian, Spanish, Swedish
  • Open-source + open-data

4 of 9

​

​

​

iOS Demo

Scribe’s German Keyboard

DATA RE–

USE DAYS

14 - 24 March 2022

5 of 9

Data Update Process

​

  • WDQS queries for each word type of each language
  • Ran via SPARQLWrapper
  • JSON outputs saved
  • Data formatting
    • Outputs to Scribe apps
  • Takes ~10min

Scribe-Data/src/scribe_data

└─── load

│ update_data.py

│

└─── extract_transform

│ └─── German

│ └─── nouns

│ │ query_nouns.sparql

│ │ nouns_queried.json

│ │ format_nouns.py -> Scribe-iOS, etc.

│ └─── prepositions

│ └─── translations

│ └─── verbs

│

│ ... (other languages)

scribe_data/load [main] » python update_data.py

​

Data updated: 0%| | 0/16 Querying German nouns

6 of 9

Android Progress

​

During Hackathon

​

  • Using open-source as basis
    • Simple Keyboard (GPL-3.0)
    • Simple Commons (GPL-3.0)
  • Combined above repos
  • Italian, Portuguese, Swedish
  • General improvements
  • All iOS Swift codes → Kotlin
    • Conversion via SwiftKotlin
    • Ready to connect data JSON files to keyboards

7 of 9

​

​

​

Android WIP Demo

Current Status

DATA RE–

USE DAYS

14 - 24 March 2022

8 of 9

Future Plans

​

  • Finish Scribe-Android 1.0.0
    • See Scribe-Android issues
    • See Phabricator issues
  • Scribe-Desktop GUI in Python
  • Wikidata based translations
  • Autosuggest, complete and correct
    • Downloadable language packs
    • Packs include NLP models
  • More languages
  • Language learning :)

Büro

Scribe im Büro

N

9 of 9

Thank you for your interest and efforts!

To stay in touch or contribute:

Andrew Tavis McAllister

andrew.t.mcallister@gmail.com

https://github.com/scribe-org

DATA RE–

USE DAYS

14 - 24 March 2022