What is WeGov? User Guide
4/18/2012
2 What is WeGov?
WeGov – Where eGovernment meets the
eSociety - is an EU research project in the 7th
Research Framework Program (“ICT for
Governance and Policy Modeling”). For more
information, please visit our project website
http://wegov-project.eu.
Introduction The WeGov project addresses the networking of citizens about politics, and with policy makers,
through social networks like Twitter and Facebook. It is not about investing in another citizen
participation platform. WeGov sees itself rather as a feasibility study to exploit the potential of social
networks for policy making, by having citizen opinions indirectly feed the decision-making processes.
The approach chosen consists in developing a site, including tools that support the political decision-
makers in the analysis of social networks. In terms of methodology, WeGov relies on the
participation of potential users (e.g. policy makers, communities, organizations) in the development
process of the software. The challenge is to reconcile as much as possible the requirements of these
user groups in terms of social media analysis with the technical feasibility of the WeGov analysis
models. WeGov has developed three alternative analytical approaches that are currently tested and
improved, as a basis for a later integration in the policy maker’s daily workflow.
Presentation of the WeGov analysis possibilities
Topic analysis
The topic analysis identifies groups of words that represent several areas of discussions that arise
within a wider debate.
This analysis is used for sorting of comments and users in the different concept groups and
can currently be used for Twitter and Facebook.
For each concept group approximately 3 user (key users) and about 3 comments (key posts)
are displayed.
Characteristic of key users and key posts is that they strongly refer to these topics, by
showing the highest overlap.
The analysis considers content related factors such as word frequency and the use of hash
tags (#). Not content related factors such as the number of retweets or likes are ignored in
this analysis.
Other social networks can also be analyzed using this model, but are currently not included.
In general, the quality of the concept groups should increase with the number and length of
comments.
The figures below show the topic analysis running on Twitter for the search term “European citizens’
initiative” on April 16. The analysis is also available for Facebook data.
Figure 1: Results of topic analysis
Basic properties for the analysis of the discussion activity and user behavior
WeGov provides two main analyses in its Toolbox (explained below). The results of our research
indicate that different social networks behave in different ways, and the factors that make a post or a
user to be “important or relevant” in one social networks are probably not the same ones in another
social networks. The models that are currently integrated in the WeGov toolkit are those ones
trained with Twitter data.
To generate these models we collect a big amount of data (representative sample of the social
networks) and perform feature engineering over the data to extract those features that represent
the user and the content.
The model is optimized for English text. As a further development, the model will in addition be
adapted Facebook and will be optimized for the German language. In general, the following
properties are used for the behavior analysis, based on User features and on Content features:
User behavior on social networks
within social networks different levels of use can be distinguished. Consider, for example, a single
Twitter user, who usually follows other Twitter users and is also followed by Twitter users. These
dependencies and other properties are taken into account to determine the influence power of users
and their posts within Social Networks.
In-degree refers to the direction of the followers, i.e. the circle of persons who potentially
read a post of that particular user
Out-degree, on the other hand, refers to the Twitter users that a person will follow and
potentially read their posts and react by responding or re-tweeting.
User Participation
4 What is WeGov?
Number of posts
Since when using social media
Frequency of Posts
General characteristics of the Posts
Length of Posts
Recommendation of the post by a third party (such as re-tweet)
Time of publication
Content characteristics of posts
Complexity is a measure for determining the level of content with respect to the word
accumulation - the higher the value, the greater the information content.
Readability is an index which describes to which extent a message is easy to understand – if
a message is understood from reading for the first time it is assumed to have good
readability.
Novelty value is a measure to determine the average number of times terms are occurring in
Posts - terms which appear for the first time increase the novelty value
Polarity measures the "mood" of the post and makes a statement how strongly the post deviates from the average - this determines whether a post is particularly negatively motivated
Timescale of the discussion activity
Figure 2: Graphical presentation of the
analyzed tweets in a timeframe
For the time being, WeGov performs the analysis on a download of the 99 last tweets, or all tweets
over a maximum period of one week if the total number of collected tweets is lower than 99. WeGov
shows how the analyzed tweets are concentrated in the concerned time period. In figure 2 above,
this means for instance that the search term “European citizens initiative” entered on April 16
generated 62 posts over the last week, with activity peaking on the first day.
5 What is WeGov?
Results of analysis of the discussion activity
The purpose of this analysis is to predict which posts are going to generate more attention. The
results of our analysis indicate that in order to generate attention the content of the post is more
important than the reputation of the user within the SNS. In particular, those posts that generate
high levels of attention generally fit the following characteristics:
They were not written in the afternoon
They are written in a familiar language (the readability is high and the information content is
rather low)
They were written by people who follow many users and even read the news (high out-
degree)
The statement tends to be rather negative (stronger negative polarity)
In the WeGov toolkit the output of this analysis is translated in top posts to watch. The top users to
watch are computed by adding the scores of the top posts for each user. I.e., the top users are those
who generate more top posts (post that are likely to generate higher levels of attention).
For the search term “ACTA” this generated the following result on April 13.
Figure 3: Top 5 posts and Top 5 users to
watch
Results of analysis of user behavior
The purpose of this analysis is to classify users according to their behavior and interactions within the
SNS. For this analysis we only use user features, in-degree, out-degree and the properties of user
participation in the information (number of posts, length of use, and frequency of posts).
Within WeGov the following groups are considered:
Broadcaster is someone who posts with
high daily rate and has a very high
following (in-degree). However he
follows very few people (out-degree).
Information Source is someone who
posts a lot, is followed a lot but follows
more people than the Broadcaster His
involvement in social networks is
generally much higher than the
broadcaster.
Daily User is an average user in relation
to the number of posts, followers and
the people he follows himself.
Information Seeker is someone who
posts very rarely but follows a lot of
people. An information seeker is
generally interested in getting
information and not in the possibility to
discuss
Rare Poster is a user with very low post
rate.
Figure 4: User roles distribution
According to this behavioral characteristics, the
most influential and engaged role is the
Information Source, which is probably the people
that PM should pay more attention to.
Figure 5: Identification of users
7 What is WeGov?
Presentation of the WeGov Website The different analysis possibilities explained above can be used on the search page and can be
restricted to geographical areas of social networks.
On the home page, the analysis capabilities also function within small windows (widgets). The
widgets offer the advantage that they are immediately visible from login to the site and they display
updates of the analysis results of previously set search criteria.
WeGov – Application
The WeGov Website can be reached at
https://wegov.it-innovation.soton.ac.uk/. The website
is currently optimized for Firefox and Chrome. The
URL leads you to the screen shown under Figure 6.
Here you can login with your personal user name and
password. The site is currently available with an
English and German interface. We ask you to
apologize for editorial gaps, since this project is still
under development.
Figure 6: Login
WeGov – Home page
After successful registration, you will see the WeGov Home page (Home). The small windows
(widgets) show different results for Twitter and Facebook requests. Currently, a total of 100 Twitter
tweets and of 1700 Facebook posts can be polled. This figure is a technical limitation of the social
networks; we are working on improving this. To query Twitter data, no registration is required. For
Facebook, this does not apply. If you want to use the Facebook analysis, it is necessary that you are
logged in with a Facebook account. It is enough if you have opened Facebook within your browser
and you have logged in. WeGov will automatically detect the connection to Facebook and use this to
query data from Facebook. WeGov guarantees at this point that no personal data are stored or
retrieved.
You will find a total of 4 different types of widgets on the front page. Using the symbol "I" in white
on a gray circle, you can adjust the settings of the window, e.g. change the search word or duplicate
the window in order to compare results. If you delete the window, it cannot be recovered. But you
have the option to "hide" the selected window on the home page (consult the tab WeGov - Personal
Settings).
8 What is WeGov?
Widget: Google Maps
In this widget (Figure 7) your current position is determined automatically, so you can automatically
display search results relative to your current location. When you are in Brussels, Brussels will be
pointed here.
Figure 7: Current location
If you are interested in analysis results related to other locations than your current location, you can
use the following widget (Figure 8) to enter and select other locations on the search page. To do this,
click on "Add new" and enter a location or region.
Figure 8: Store locations
9 What is WeGov?
Widget: Topic Analysis
The following window (Figure 9) shows topics that are currently discussed on Twitter on Klaus
Wowereit. For each group, the term "key user" is displayed.
Figure 9: Topics discussed on a search term
The next window (Figure 10) uses the same analysis as Figure 4. All tweets are analyzed on content
to identify the topics currently under discussion on Klaus Wowereit. The difference is that here is a
geographical restriction of tweets has been on the current location (here Berlin)
Figure 10: Topics discussed locally on a search term
10 What is WeGov?
The widget Facebook Post for: Angela Merkel1 (Figure 11) shows the most recent posts on the
Facebook Fan Page of Angela Merkel. In WeGov a total of last 25 posts is available. By choosing the
title detail page all 25 posts will be displayed in full length. In addition, the number of Facebook likes,
and the number of comments are visible. Below the post the unique Facebook-key (a combination) is
displayed. This can be entered in a separate window to start an analysis of the comments of this
post. We are currently working on a solution in which posts can be directly selected for analysis.
Using the icon (bottom right corner) another fan page to be entered for analysis.
Figure 11: Recent posts of the selected Facebook fan page
The widget in Figure 12 shows some of the topics discussed on the Facebook post 60 years
Protestant work... For each concept group, the term "key user" is displayed.
Figure 12: Topics discussed on a Facebook post
1 URL: https://www.facebook.com/AngelaMerkel (Searched in March 2012)
11 What is WeGov?
The window in Figure 13 also refers to the post 60 years Protestant work ... Relevant comments are
displayed here. The detail view of the list is available by clicking on the title bar.
Figure 13: Comments on a selected Facebook post
Figure 14 shows the analysis result of the recent posts on the fan side, Angela Merkel. Topics and
"key users" are displayed.
Figure 14: Topics discussed on a selected Facebook fan page
12 What is WeGov?
Widget: Behavior analysis
The following window (Figure 15) organizes tweets about the theme minimum wage according to the
user roles of the authors who publish the tweets. The group Information Source is considered to be
particularly interesting, because these users tend to have greater visibility in social networks (Twitter
in this case) than others.
Figure 15: user roles on a topic
Figure 16 shows a total of two Twitter user profiles with the role of Information Source. These users
are displayed as they were identified as a user with great influence from the set of tweets on the
subject of the theme minimum wage. This view displays users with other roles.
Figure 16: Users seldom posting on a topic
13 What is WeGov?
Widget: Analysis results of external websites
The following windows are examples of the integration of non-WeGov services on the site. The idea
behind this is that there many good analytical capabilities exist in social media. WeGov wants to
integrate these analysis capabilities to its product range and provide them to its users and users. The
following window (Figure 17) integrates Twitter and displays the latest tweets about Klaus Wowereit
released near your current location (in this case Berlin).
Figure 17: Comments on a search term
Figure 18 also integrates analytical results from Twitter. In this case, Twitter-generated topics that
Twitter users are currently discussing in Germany are displayed.
Figure 18: Top 10 themes on Twitter
14 What is WeGov?
WeGov - Personal settings
By clicking on the username in the upper right corner you can change personal settings as shown in
Figure 19. You can for instance set up a new password for your account. The list shows all My
widgets that are available on the homepage. If the box is checked, the window is displayed. If not, it
remains invisible.
Figure 19: User preferences
15 What is WeGov?
WeGov - Search Figure 20 shows the WeGov search page with detailed analysis of results. Currently 100 tweets from
Twitter are queried. This number will soon be expanded, as well as the ability to search on other
social networks. Enter a search word to start and possibly narrow down your search geographically
by selecting one location. The available locations are displayed on the map and must be pre-selected
in the window My Saved locations on the homepage. As results, you will see three lists. The first list
contains the hit list, which is generated through Twitter. The other two navigation points Topic
Analysis and Behavior Analysis show the previously described thematic sorting as well as the sorting
of the user and comments in relation to user behavior.
Figure 20: Detailed search
16 What is WeGov?
Your contact persons for questions and constructive input Dipl.-Inf. Timo Wandhöfer
GESIS – Leibniz-Institut für Sozialwissenschaften
Abteilung: Wissenstechnologien
Unter Sachsenhausen 6-8, 50667 Köln
Tel. +49 (0) 221 – 47694 – 544
E-Mail: [email protected]
Catherine van Eeckhaute
Deputy Director Gov2u
Tel + 32 (0)475-243522
E-Mail: [email protected]