European Symposium Series on Societal Challenges

Call for Workshops and Tutorials

Workshops will give the opportunity to meet and discuss issues with a selected focus, providing an excellent forum for exploring emerging approaches and task areas and bridging the gaps between the social science and technology fields.

Tutorials will be an opportunity for cross-disciplinary engagement and a deeper understanding of new tools, techniques, and research methodologies. Tutorials should provide either an in-depth look at an emerging technique or software package or a broad summary of an important direction in the field.

Members of all segments of the social media research community are encouraged to submit proposals. To foster interaction and exchange of ideas, the workshops will be kept small, with 30 participants maximum.

Important Dates

Workshops and tutorials proposal submission deadline: August 1st, 2018

Workshops and tutorials acceptance notification: August 8th, 2018

Workshops and tutorials day: December 5th, 2018

Submission Guidelines

Please submit your workshop or tutorial proposal by sending a PDF file via email to css.eurosymposium@gmail.com

Proposals for workshops and tutorials should be no more than three (3) pages in length (10pt, single column, with reasonable margins), written in English, and should contain the following:

• A concise title
• The names, affiliations, and contact information of the organizers
• Planned duration of the event (half-day or full-day meeting)
• A short abstract describing the scope and main objective of the event
• A description of the proposed event format and a detailed list of proposed activities

Workshops and tutorials will be selected based on the following criteria:

• Timeliness of the topic
• Potential to attract the interest of researchers in computer science and social/organizational sciences
• Promotion of activities that are different from the classic mini-conference format; those include challenges, games, interactive sessions, brainstorming and networking.
• Involvement people of different backgrounds in the organizing committee
• Addressing topics at the intersection of different disciplines

SUBMIT YOUR PROPOSAL VIA EMAIL

Call for Doctoral Consortium

This year the symposium will also feature a doctoral consortium, which will take place on the pre-symposium day December 5th, 2018.
PhD candidates from all disciplines that work on topics related to computational social science are welcome. We are delighted to have two distinguished researchers acting as mentors during the event: Sandra González-Bailón and Jürgen Pfeffer.

If you are interested in connecting with other PhD candidates in the field and receiving feedback on your PhD topic, please consider applying for the doctoral consortium by submitting a description of your work. Based on the submissions we will select participants to present their work in the consortium meeting and to participate in discussions and mentoring sessions. We are interested in all topics that fit into computational social science, the topic does not have to be directly related to this year’s symposium theme “bias and discrimination”.

Submission Guidelines

Applications should be submitted via Easychair: https://easychair.org/conferences/?conf=csssymposium18
They should be PDF files of up to approx. 1000 words, including the following information:

• an abstract of the PhD thesis topic including information about the research question, applied methods and the current state of progress.
• a short paragraph about the applicant’s biography
• a paragraph on open questions that the applicant would want to discuss with experts in the field

Deadline for submissions is September 1st, 2018.

Travel Grants

Participants presenting in the doctoral consortium can also apply for travel grants. The number of grants as well as the amount per grant is limited. Please see more detailed information here.

SUBMIT YOUR ABSTRACT @ EASYCHAIR

Call for Dataset Challenge

For the first Dataset Challenge as part of our European Symposium on Societal Challenges in CSS, we invite researchers and practitioners from the spectrum of Social and Computational Sciences to approach a common dataset. The aim of this challenge is to encourage creative engagement with data from pluralistic perspectives in order to foster dialogue across disciplines. We envision that this challenge will lead to fruitful, in-depth discussions during the symposium, as participants will have a shared basis through interaction with the dataset(s), but varying ideas of where to go with it.

The research questions and methods applied to the dataset(s) can be drawn from the broad spectrum of computational social science. They should ideally be related to this year’s symposium topic “bias and discrimination” and approach new ways of learning about bias, be it by putting forward new research designs and questions, or by applying state-of-the-art methods to identify data patterns that can inform theory building. Topics of interest include but are not limited to:

• Comparisons of user/political actor behaviour
• Unequal structures and political action
• Biases in Web data
• Missing data
• Coverage bias on different platforms / through different collection methods

We ask participants of the dataset challenge to be creative in their approach to the data.

The work presented at the symposium can be early stage, but should be methodically sound and provide either preliminary insights into an interesting research question or first inductively-reached conclusions of systematic patterns of bias in the dataset. These can also be on a meta-level or may be designated to positioning the work in broader theoretical contexts, e.g., about biases in the data collection. We especially invite inventive combinations with other datasets (e.g., digital trace data, survey data, multimedia data, multilingual data, data from other case studies).

Submissions selected by our review committee will be presented in a special dataset challenge session at the pre-symposium day (December 5th, 2018). After the presentations, participants (and audience members) are invited to engage in discourse about the presented approaches and possible future work and collaborations.

A jury will select the best presentation (including any created resources, such as secondary datasets or analysis code) to be awarded the very first Euro CSS Dataset Challenge Award of 500 Euros sponsored by EPJ Data Science.

Submission Guidelines

Extended abstracts of work in progress or completed projects based on the suggested dataset(s) should be submitted in English in PDF format via the EasyChair submission system: https://easychair.org/conferences/?conf=csssymposium18

Submissions should be abstracts of approx. 2-3 pages (maximum up to 1000 words plus references and figures).

The abstract submissions will be evaluated based on originality and the potential to stimulate interesting discussions, but should also consider the feasibility of the proposed idea. Please consider that reviewers will be from an interdisciplinary community, therefore describe your ideas, approaches and potential complementary datasets in sufficient detail.

Deadline for submissions to the dataset challenge is September 1st, 2018. The final work will be presented at the symposium.

Please see the additional information for how to apply to travel grants here.

The Datasets

The following datasets all relate to the topic of political communication in Germany, particularly around the German federal election which took place in September 2017. Participants of the challenge should make use of one or more of the main datasets listed below and are encouraged to combine them or compare them among each other. Additional datasets listed further below may also be used, as well as any other datasets of the participants’ choice (qualitative, quantitative, textual, multimedia, etc.).

Main Datasets

(participants must choose at least one of these – or combine them with other datasets)

1. Algorithmwatch’s donated search results dataset for the German Federal Elections 2017

How similar are the results of Google searches for the names of political parties and candidates for different users? What role does personalisation play in these differences?

The Datenspende: BTW17 is a crowdsourced project which is a cooperation of AlgorithmWatch with media regulatory authorities of the German state and other partners.

Through a browser plugin, users donate their search result data to the project. The plugin conducts searches (on Google Web search and Google News search) for a fixed set of search terms (described below) through the browser of each participant and returns the first page of search results back to AlgorithmWatch’s servers. In addition to the search results, the approximate location of the user (city-level), the time of the query and whether the user is logged into their Google account or not (but no personally identifiable information) is also logged.

More than 4000 participants have donated more than 5 million search results in the period around the German federal elections of 2017. Anonymized search results data has been made publicly available.

Search terms

The search terms comprise of candidate names or party names / abbreviations. The exact list of search terms (two types):

Important politicians:
‘Angela Merkel’, ‘Martin Schulz’, ‘Christian Lindner’, ‘Katrin Goering-Eckardt’, ‘Cem Özdemir’, ‘Sahra Wagenknecht’, ‘Dietmar Bartsch’, ‘Alice Weidel’, ‘Alexander Gauland’

Major parties:
‘CDU’, ‘CSU’, ‘SPD’, ‘FDP’, ‘Bündnis 90 / Die Grünen’, ‘Die Linke, ‘AFD’

Data and its format

The search data for each day can be downloaded as a 7z-packed file from https://datenspende.algorithmwatch.org/data.html.

It is formatted as json objects consisting of two parts:

(i) Search queries made at day X: Each query includes information about the type of search (web or news), location, login status, language, search term, timestamp and the result hash.

(ii) List of search results: Multiple queries could lead to the same list of search results being shown to the users. Therefore the result hash in the query (described above) identifies the exact list of search results shown to the users.

More information about the data formats can be found at: https://datenspende.algorithmwatch.org/data.html
EN: http://translate.google.com/translate?js=n&sl=de&tl=destination_language&u=https://datenspende.algorithmwatch.org/data.html

First publications on the data are listed here: https://datenspende.algorithmwatch.org/en/index.html#ergebnisse

2. Abgeordnetenwatch.de – All members of German parliaments and their votes + more

An API is provided to query several aspects of politicians’ and parties’ metadata in the German Bundestag (BT) and the German state parliaments. While the time range varies, BT votes are available from 2005 and for most state parliaments at least for the last couple of years. Information is available about all elected officials in these parliaments, their metadata and their voting behavior in all the polls.

The general description of the possible API calls can be found at:
http://translate.google.com/translate?js=n&sl=de&tl=destination_language&u=https://www.abgeordnetenwatch.de/api (EN w/ Gtranslate).

Below we outline some of the important details and options for the different API calls (some of which are not explained exhaustively on the website):

https://www.abgeordnetenwatch.de/api/parliaments.json (.xml also possible) provides a good entry point as it gets information about all parliaments – per legislature – and allows to get all that parliament’s deputies, candidates, constituencies, polls and committees in separate calls either by name or uuid (See individual “parliaments” and then “datasets” entries under the json root node). Importantly, note that all calls for these lists have also subset methods (defined in the “meta” header”) that lets you dive deeper into the data.

https://www.abgeordnetenwatch.de/api/parliament/bundestag/polls.json for instance retrieves all polls in the BT with metadata (such as a “topics” field, which comes from a fixed category vocabulary) and votes per party, which can be expanded to votes per individal member via the …polls.json?subsets=votes call.

Detailed information about individual politicans can also be retrieved, for instance via https://www.abgeordnetenwatch.de/api/parliament/bundestag/profile/jens-spahn/profile.json

Apart from official parliamentary process data there is also additional data on questions asked by users to politicians on the platform abgeordnetenwatch.de itself, which in many cases have been answered by the politicians themselves.

3. Social media content generated by German politicians during the German federal election 2017 (BTW17)

This dataset contains results from the social media monitoring for the German federal election campaign 2017. The project collected the tweets of political candidates, organizations and other gatekeepers, as well as the engagement of users with these contents in the form of retweets and @-mentions. In addition, all messages on Twitter containing at least one keyword denoting central political topics were also collected. All data was publicly available at the time of data collection. The dataset covers the time period between July and October 2017.

This dataset contains 22,123,230 tweet IDs linked to 2517 politician and 444 organization profiles. Full texts and metadata for these tweet IDs can be retrieved via the Twitter API. This can for example be done with this tool, requiring only moderate technical expertise: https://github.com/DocNow/hydrator. Additionally, we can provide the full (mostly German) text and metadata for a random subset of 50,000 tweets upon request. While there are also 434 Facebook profile links in the dataset, text content of posts are unfortunately not available from us at this time.

A detailed description of the collection process can be found here: https://dbk.gesis.org/dbksearch/download.asp?db=E&id=63766

And the dataset can be found here: https://dbk.gesis.org/dbksearch/sdesc2.asp?no=6926

Auxiliary datasets (may complement the main datasets)

• Everypolitician.org

For German politicians in the Bundestag, http://everypolitician.org/germany/bundestag/download.html offers CSV lists per legislature period plus metadata, including

– biographical data

– social media profile links (Twitter, FB, Youtube,…)

– external identifiers (Wikidata/Wikipedia, gnd, viaf, …) Wikidata for instance has more structured information about these politicians, including links to their respective Wikipedia pages.

There is even a python library you might want to use

• Offenesparlament.de

This initiative collects all talk topics and corresponding minutes from the Bundestag sessions 2013-2017 in easily accessible formats and even offers an API for easy access.

Description: http://translate.google.com/translate?js=n&sl=de&tl=destination_language&u=https://offenesparlament.de/daten/ (EN w/ Gtranslate)

Data: http://translate.google.com/translate?js=n&sl=de&tl=destination_language&u=https://github.com/Datenschule/offenesparlament-data (EN w/ Gtranslate)

For example, https://github.com/Datenschule/offenesparlament-data/blob/master/tops/tops.csv

has all the discussion points for different days in the Bundestag, with predefined categories for most points and these can be linked to the actual utterances of individual politicians via https://github.com/Datenschule/offenesparlament-data/tree/master/sessions. Politicians are also explicitly linked to Abgeordnetenwatch.de profiles. An API for most up-to-date data is also available and a Jupyter Notebook explaining the access can be found at: https://github.com/Datenschule/offenesparlament-data/blob/master/load_data.ipynb

Miscellaneous

Other potentially relevant datasets:

• German Longitudinal Election Study – http://gles.eu/wordpress/english/, surveys and a categorized long term media content analysis relating to German elections
• Politicians on Wikipedia and DBpedia – https://datorium.gesis.org/xmlui/handle/10.7802/1515

Any other relevant dataset that you can find

You are encouraged to use other datasets that you can combine with the main datasets provided. Additional datasets should optimally be available under an open license at the time of presenting at the event – or they should at least be recreatable with reasonable effort.

How to join these datasets

Linking the provided datasets can be most easily done via

• Parties: these names are mostly constant (either abbreviations or full names), apart from changes in the name of “Die Linke” from a merger of two parties in 2007; as well as ‘Bündnis 90 / Die Grünen’, which in some cases is written as only ‘Die Grünen’ (Abgeordnetenwatch.de)  or ‘Grüne’ (see BTW17 Codebook). Note that “fraktionslos” (”independent” / “without party affiliation”) is also contained in some datasets.
• Politicians: first name, last name, birth year, gender are contained in 2 of the 3 main datasets (with the nine names in the Algorithmwatch dataset easily mappable by hand) and constitute composite unique keys in almost all cases (if not, this data provides some disambiguation for changed names/ writing styles).

Furthermore, other fields like “constituency” etc. offer rather straightforward linking opportunities as well.

FAQ / assistance

For any questions related to technical aspects, data structure and especially German language that might come up with these datasets, please do not hesitate to contact us! We chose to put the focus on German political data because we think it is underexplored with CSS methods, at the same time holding very interesting potential insights into biases in political processes and communication. We are aware of the language barrier but are confident that most of the structured and even textual data can still be useful even to non-German-speakers.

Please feel free to use our Google Group / mailing list “CSSnet” for discussions about the dataset challenge, for looking for potential collaborations, or to ask questions to the community: https://groups.google.com/forum/#!forum/cssnet

Specific questions can also be sent to us directly at css.eurosymposium@gmail.com.

SUBMIT YOUR ABSTRACT @ EASYCHAIR

Science Slam

A Science Slam is an epic scientific event where scientists compete with short talks on their research. It’s just like a poetry slam, but with science instead of poems. Slammers are completely free to do whatever they want on stage, everything is allowed including slides, games, the more creative, the better!

The only two rules are:

1) The topic of the slam has to be related to data science / computational social science or to social media, online data or digital behavioral data, and

2) Presentations should not take more than 8 minutes.

When? –> December 5th, 8pm
Where? –> “die wohngemeinschaft” (Richard-Wagner-Straße 39)
How much? –> For free!

On stage:

Lisa Sivak: “Gender Bias in Sharenting and Gender Inequality”
Komal Florio: “10 Things I Hate about You: Immigration and Hate Speech on Twitter in Italy”
Mykola Makhortykh: “Algorithms for Peace: How News Recommender Systems can facilitate constructive Conflict Reporting”
Brett Buttliere: “Conflict as a Necessary but not Sufficient Condition of Being Extra-ordinary”
Samin Aref: “Is the Enemy of MyEnemy, a Friend? What about the Enemy of MyFriend or the Friend of MyEnemy?”
Lixue Lin: “Mapping the Landscape of Science outside of Web of Science”
Christo Wilson: “Control Your Experiments, Donald Trump”

Presentation: Aniko Hannak and Fabian Flöck

We will have a limited number of 70 seats. The seats will be given according to the principle “first come first serve”.

The Science Slam is supported by HumTec (Human Technology Center), RWTH Aachen University.

Get direction to Science Slam Venue