OK here's my take on the second presentation here (responding to this mail because it has the link to the presentation).
At first I was surprised by the huge gap between the number of founders on enwiki (44,000) and only 2,000 on Wikidata, but then I recalled that many COI entries on living people eventually get merged into their companies. Though founder-company relationships are interesting, I prefer looking at the work on the family relationships, which is something I have worked on quite a bit for 17th-century biographies. It's a real pain to update these in Wikidata and so it would be great to get the edits these guys prepared loaded into some Wikigame. I also really liked their "factor graph model" and we could use something like this as a set up for Wikigames involving not just family relationships, but also employer-employee, prize lists-prize recipient, alumni lists, lists of mayors, abbots and pretty much anything that could add quality info to person items. I think it would be really useful to followup this work with looking into ways to get their prepared edits actually into Wikidata (with the Wikipedia reference statements) but also ways to expand their model.

One problem we have with Wikidata are inverse properties, so e.g. famous victims and their killers. We have a killed property but no killed by property and so on. It would be nice to implement an easy factor graph model for such properties so that inverse properties become unecessary.

On Wed, Jul 29, 2015 at 8:07 PM, Leila Zia <leila@wikimedia.org> wrote:
A friendly reminder that this is happening in 23 min. :-)

YouTube stream: https://www.youtube.com/watch?v=vGyrVg_qKSM
IRC: #wikimedia-research


On Mon, Jul 27, 2015 at 2:47 PM, Leila Zia <leila@wikimedia.org> wrote:
Hi everyone,

The next Research showcase will be live-streamed this Wednesday, July 29 at 11.30 PT. The streaming link will be posted on the lists a few minutes before the showcase starts (sorry, we haven't been able to solve this, yet. :-() and as usual, you can join the conversation on IRC at #wikimedia-research.

We look forward to seeing you!


This month:
VisualEditor's effect on newly registered users
By Aaron Halfaker
It's been nearly two years since we ran an initial study of VisualEditor's effect on newly registered editors. While most of the results of this study were positive (e.g. workload on Wikipedians did not increase), we still saw a significant decrease in the newcomer productivity. In the meantime, the Editing team has made substantial improvements to performance and functionality. In this presentation, I'll report on the results of a new experiment designed to test the effects of enabling this improved VisualEditor software for newly registered users by default. I'll show what we learned from the experiment and discuss some results have opened larger questions about what, exactly, is difficult about being a newcomer to English Wikipedia.

Wikipedia knowledge graph with DeepDive

By Juhana Kangaspunta and Thomas Palomares (10-week student project)
Despite the tremendous amount of information present on Wikipedia, only a very little amount is structured. Most of the information is embedded in text and extracting it is a non-trivial challenge. In this project, we try to populate Wikidata, a structured component of Wikipedia, using DeepDive tool to extract relations embedded in the text. We finally extracted more than 140,000 relations with more than 90% average precision. We will present DeepDive and the data that we use for this project, we explain the relations we focused on so far and explain the implementation and pipeline, including our model, features and extractors. Finally, we detail our results with a thorough precision and recall analysis.

Wiki-research-l mailing list