WikiArabia 2019/Program/A look into the future: Supporting Arabic Wikipedia with Wikidata and ArticlePlaceholder
- Submission no. مداخلة رقم
- Title of the submission عنوان المداخلة
A look into the future: Supporting Arabic Wikipedia with Wikidata and ArticlePlaceholder
- Author(s) of the submission صاحب المداخلة
Lucie-Aimée Kaffee and Hady Elsahar
- Username اسم المستخدم
User:Frimelle and User:Hadyelsahar
- E-mail address البريد الإلكتروني
kaffeesoton.ac.uk
hadyelsahar
gmail.com
- Country of origin بلد الأصلي
Germany and Egypt
- Affiliation, if any (organisation, company etc. .الانتماء، إن كان (مجموعة، مؤسسة الخ
University of Southampton and Université de Lyon
- Type of session نوع الجلسة
Talk عرض
- Language of session لغة الجلسة
English
- Length of session مدة الجلسة
30 min
- Theme of presentation موضوع الجلسة
A look into the future: Supporting Arabic Wikipedia with Wikidata and ArticlePlaceholder
- Abstract ملخص

Arabic is one of the most spoken languages in the world. However, at the web at large and Wikipedia in specific, there is little information available for native speakers. With little information on the Wikipedia, it is hard to win editors, and with few editors, it is hard to increase the number of high-quality articles. However, we have a central storage for knowledge, that is already connected to Wikipedia and is completely language-independent: Wikidata!
We want to tackle the problem from the readers’ side: We introduced ArticlePlaceholder, an extension deployed on over 14 Wikipedias to date, that displays triples from Wikidata on Wikipedia. But rather than just displaying just information as facts, we want to make the site easy to access and read for the community. Therefore, we worked on natural language generation of text from Wikidata triples for Wikipedia introductory sentences.
Today, most of the natural language generation focuses on English. We adapt existing neural network models and adapt them to a low-resource setting and the challenges that come with working on Arabic. We tested our approach with Wikipedia readers and editors and could show promising results. Further, we conducted a set of interviews to gain a deeper understanding of the needs of the community. We want to invite you to see our progress and discuss further ways from here, that can serve the Arabic speaking community best.
- Important Links
- Our research has been featured in Wikimedia Research Showcase July 2018 https://www.mediawiki.org/wiki/Wikimedia_Research/Showcase#July_2018
- Our published research can be found here:
- Mind the (Language) Gap: Generation of Multilingual Wikipedia Summaries from Wikidata for ArticlePlaceholders https://eprints.soton.ac.uk/419727/
- Learning to Generate Wikipedia Summaries for Underserved Languages https://aclanthology.info/papers/N18-2101/n18-2101
- An example of the ArticlePlaceholder in Arabic with an automatically generated summary can be found here: Example for Marrakesh in Arabic
- Page of the ArticlePlaceholder extension
- What will attendees take away from this session? ماذا سيستفيد الحضور من هذه الجلسة؟
- Participants will learn about the integration of linked data from Wikidata in Wikipedia in the form of the ArticlePlaceholder extension,
- get an overview of natural language generation for Wikipedia and our research in the field,
- and finally, you will get a good insight into the studies we conducted with members of Arabic and other underserved Wikipedia communities and the conclusions we could draw for our project.
- Will you attend WikiArabia if your submission is not accepted? هل ستحضر لويكي عربية إذا لم يتم قبول مداخلتك؟
Yes
- Special requests طلبات خاصة
- Is this Submission a Draft or Final? هل هذا الطلب مسودة أم نهائي؟
Final
Interested attendees مهتمون بالحضور
[edit]If you are interested in attending this session, please sign with your username below. This will help reviewers to decide which sessions are of high interest.
إذا كنت مهتمًا بحضور هذه الجلسة، فرجاءا قم بإضافة اسم المستخدم أدناه. سيساعد هذا المراجعين على تحديد ما هي الجلسات ذات الاهتمام الكبير
- --Reem Al-Kashif (talk) 18:13, 11 November 2018 (UTC)
- --Zack (talk) 18:58, 11 November 2018 (UTC)
- --Sky xe (talk) 07:11, 13 November 2018 (UTC)
- It's different from what we do actually on Arabic Wikipedia (discussing a frame text about a subject then use a bot to generate articles using Wikidata). This presentation will give us the opportunity to compare the 2 methods. --Helmoony (talk) 02:06, 14 November 2018 (UTC)
- --New Zeus (talk) 21:59, 27 November 2018 (UTC)
- --Dew (talk) 20:37, 29 November 2018 (UTC)