Web of Linked Entities
Doing Good by Linking Entities
Developers Challenge at WoLE2013
The WoLE2013 Challenge is offering prizes for the best applications that show the impact of the Web of Linked Entities on problems/solutions affecting the local community, e.g. detecting corruption, tracking criminality, facilitating access to education or health services, helping to search for the cure for neglected diseases, promoting citizen participation on the government, improving tourism-related services, etc.
Although we have observed an explosion of the number of structured data sources shared on the “Web of Data”, the majority of the available Web content is still unstructured or semi-structured -- e.g. encoded in (hyper-) textual documents. Web content (structured, unstructured and semi-structured) often makes references to other documents or to entities described in structured data sources (e.g. Linked Data). Moreover, structured data sources provide entity to entity interconnections, resulting in a Web of Linked Entities spanning structured and unstructured data.
We believe that interconnecting and sharing explicit interconnections between documents and open data sources on the Web will increase the value of each data source and enable a number of innovative applications that leverage both the individual data sources and their interconnections.
Submissions will be judged based on originality, potential impact and pertinence to the subject of the workshop. Note to prospective participants: if you run a website that has (potential) social impact, but you are not an expert on WoLE research/technologies, or if you are a technical person that needs ideas for applications with high potential for the region, we have prepared a team for assisting you with questions/answers at our discussion group.
We will award up to 2 iPad2 16GB to the best application(s). The awards are generously sponsored by Globo.com.
How to participate
Submit your application to the challenge at: https://www.easychair.org/conferences/?conf=wole2013
( Please select the track "Doing Good by Linking Entities (Challenge)" )
April 10th, 2013 April 24th, 2013 (extended).
If you want to build a team, share an idea, or ask any questions, please don't hesitate to join our discussion group http://groups.google.com/group/wole2013challenge
The challenge is now also open to "ideas papers". You are invited to send a 1-2 page extended abstract describing your application idea, (optionally) along with a 5-min video to motivate others to build your idea. The best ideas will be shown at the workshop and, if you are there, you may present it yourself in a lightning talk!
Applications must use as an important component of the application, one or more of the following:
- URIs as identifiers for entities: suppose your application talks about Pelé, the Brazilian football legend. When outputting content for third parties to use, instead of (or in addition to) using plain old strings such as “Pelé” or “Edson Arantes do Nascimento”, or even internal database identifiers such as “123”, your application should point to globally unique identifiers on the Web. You may use identifiers for that entity (e.g. http://dbpedia.org/page/Pel%C3%A9), or identifiers for documents that describe that entity (e.g. http://pt.wikipedia.org/wiki/Pelé).
- Markup languages (such as Microdata or RDFa) for semantic typing of content according to a domain of knowledge, rather than describing the structure or presentation of a Web page, preferably in combination with Schema.org or other schemata that provide URIs for their types.
- Information Extraction (IE) techniques that allow one to automatically recognize names of entities (e.g. Named Entity Recognition), guess a unique identifier for a name given the context (e.g. disambiguation, entity linking) or guess relationships involving entities (e.g. relationship extraction).
- Reuse available Open Data or produce data that qualifies as 4 or 5-star Open Data (http://5stardata.info/).
Submissions must include a link to a functional application available on the Web, are encouraged to include a short paper (PDF document) with the description of the application of 2-4 pages in English language, and can optionally provide a screencast or other video presentation of the information. Make sure that your short paper and/or website contains all the information that a README file would contain, in order to make your app easier to use or your experiment easily reproducible: i.e. which datasets are used, how entities are collected, which information extraction system was used, with which settings, etc. The papers, video and link to the application will be available from the workshop website.
Winning entries must provide a video presentation of their application to be shown during the workshop, or register for the conference and attend the workshop to present their application in person.
We include below a (non-exhaustive) list of resources that may serve as building blocks for your entry to our challenge:
Applications are encouraged to use freely and openly available data on the Web, such as:
- DBpedia (http://dbpedia.org) and its localized version DBpedia Portuguese (http://pt.dbpedia.org).
- Brazilian Open Data Portal (http://dados.gov.br).
- Common Crawl (http://commoncrawl.org) and Web Data Commons (http://webdatacommons.org).
- Several Linked Open Data sets available at The Data Hub (http://thedatahub.io).
Open Source Software:
The (non-exhaustive) list below includes a few tools that fit well within the vision of the Web of Linked Entities, are free and can be customized for your needs:
- DBpedia Spotlight: Apache-licensed Open Source software that looks for ~3.5M things of unknown or ~320 known types in text and tries to link them to their global unique identifiers in DBpedia.
- Apache Stanbol: Apache-licensed Open Source software used to extend traditional content management systems with semantic services.
- LOD2 Stack: comprises a number of tools for managing the life-cycle of Linked Data, including Extraction of RDF from text, XML and SQL, Querying and Exploration using SPARQL, Authoring of Linked Data using a Semantic Wiki, Semi-automatic link discovery between Linked Data sources and Knowledge-base Enrichment and Repair.
Example Applications that Impact the Local Community:
The list below contains examples of applications focusing on problems/solutions that affect the local community (e.g. in Brazil and South America). Some of them already use techniques of interest to this workshop, and others are amenable to the application of such techniques to enhance, integrate and innovate. You should feel free to get inspired by these ideas, or come up with your own.
- Ligado Nos Políticos (http://ligadonospoliticos.com.br), is a website integrating data about Brazilian Politicians from several open data sources in one interface for exploration, including Linked Data and SPARQL endpoint access.
- Where does my money go? (http://wheredoesmymoneygo.org/), is a website that allows citizens to see how their daily taxes are divided between the different parts of government. Brazilian version (currently only for São Paulo state): http://www.paraondefoiomeudinheiro.com.br/
- Vote Na Web (http://votenaweb.com.br/) - In Brazilian Portuguese, is a website where citizens can vote on laws discussed in the Brazilian house of representatives, allowing a comparison between the opinions of citizens and representatives.
- Reclame Aqui (http://reclameaqui.com.br) - In Brazilian Portuguese, is a website where customers can post complaints about products and services, and allows prospective customers to search for online reviews of products/services before buying/hiring them.
- Multas Sociais (http://multassociais.net) - In Brazilian Portuguese, is a website where citizens can upload pictures of illegal or undesired actions performed by drivers in their region so that other citizens can “issue” virtual tickets to those drivers.
- Queremos Saber. http://queremossaber.org.br - In Brazilian Portuguese, is a website where citizens can demand their right to information access (warranted by the Brazilian legislation).
- Adote um Pedido. http://adoteumpedido.info/ - In Brazilian Portuguese, is a website where citizens can demand their right to information access, while keeping their identities anonymous by having another citizen adopting their question.
The best paper will be given a "best paper award" certificate and an iPad2 16GB.
Challenge winners will be awarded iPads2 16GB.
Feb 25th 2013 Mar 1st 2013: research paper submissions
Mar 13thMar 19th, 2013: research paper notifications
March 27thApril 1st, 2013: camera-ready research paper
April 10thApril 24th 2013: challenge submissions
May 13th 2013 : WoLE2013 workshop day
All deadlines are 23:59 Hawaii Time.