Hi Everyone,
The next Wikimedia Research Showcase will be live-streamed Wednesday, July 11, 2018 at 11:30 AM (PDT) 18:30 UTC.
As usual, you can join the conversation on IRC at #wikimedia-
research. And, you can watch our past
research showcases here.
Hope to see you there!
This month's presentations:
- Mind the (Language) Gap: Neural Generation of Multilingual Wikipedia Summaries from Wikidata for ArticlePlaceholders
- By Lucie-Aimée Kaffee
- While Wikipedia exists in 287 languages, its content is unevenly distributed among them. It is therefore of the utmost social and cultural interests to address languages for which native speakers have only access to an impoverished Wikipedia. In this work, we investigate the generation of summaries for Wikipedia articles in underserved languages, given structured data as an input.
- In order to address the information bias towards widely spoken languages, we focus on an important support for such summaries: ArticlePlaceholders, which are dynamically generated content pages in underserved Wikipedia versions. They enable native speakers to access existing information in Wikidata, a structured Knowledge Base (KB). Our system provides a generative neural network architecture, which processes the triples of the KB as they are dynamically provided by the ArticlePlaceholder, and generate a comprehensible textual summary. This data-driven approach is tested with the goal of understanding how well it matches the communities' needs on two underserved languages on the Web: Arabic, a language with a big community with disproportionate access to knowledge online, and Esperanto.
- With the help of the Arabic and Esperanto Wikipedians, we conduct an extended evaluation which exhibits not only the quality of the generated text but also the applicability of our end-system to any underserved Wikipedia version.
- Token-level change tracking: data, tools and insights
- By Fabian Flöck
- This talk first gives an overview of the WikiWho infrastructure, which provides tracking of changes to single tokens (~words) in articles of different Wikipedia language versions. It exposes APIs for accessing this data in near-real time, and is complemented by a published static dataset. Several insights are presented regarding provenance, partial reverts, token-level conflict and other metrics that only become available with such data. Lastly, the talk will cover several tools and scripts that are already using the API and will discuss their application scenarios, such as investigation of authorship, conflicted content and editor productivity.