Wikidata:WikidataCon 2017/Submissions/From unstructured to structured data using machine learning
This is an Open submission for WikidataCon 2017 that has not yet been reviewed by the members of the Program Committee.
- Submission no. 25
- Title of the submission
- From unstructured to structured data using machine learning
- Author(s) of the submission
- Daniel Ecer
- E-mail address
- d.ecer
elifesciences.org
- Country of origin
- United Kingdom
- Affiliation, if any (organisation, company etc.)
- eLife Sciences Publications Ltd
- Type of session
- Talk
- Length of session
- 45min
- Ideal number of attendees
- 50-100
- Abstract
A huge amount of unstructured data exists that is designed for human consumption but not machine readability. For example, in science, historic research is archived and the latest findings (preprints) are often distributed as PDF. Converting the unstructured PDF to a structured machine-readable format will substantially improve the reusability of this research.
eLife is a non-profit organisation that invests heavily in software development and collaboration so that the potential for improvements in the digital communication of new research can start to be realised. With this goal in mind, we recently started a project to convert the unstructured PDF to structured XML, with the aim to build on and improve the accuracy of existing conversion tools. This output could feed into Wikidata.
In this talk, we will provide an overview of the existing tools for unlocking data within unstructured scientific manuscripts and describe their limitations. We will then outline our promising computer vision approach to improving conversion accuracy, before calling on contributions and collaborations for this open-source project.
(This talk could adapt to different formats if preferred)
- What will attendees take away from this session?
- Challenges of extraction from unstructured data
- Overview of existing tools
- How computer vision can help
- How you can contribute
- Slides or further information
https://github.com/elifesciences/sciencebeam
- Special requests
Interested attendees[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.
- ...