SocInfo2014 CityLabs Workshop
Download as PPTX, PDF2 likes826 views
The document explores how policing organizations utilize social media to communicate with the public on various issues, highlighting the engagement dynamics on Twitter, particularly through the @dorsetpolice account. It identifies key characteristics of tweets that generate higher engagement, such as length, positive sentiment, and timing, using machine learning and semantic techniques for analysis. The conclusions emphasize the need to understand effective content features to enhance public interaction with policing messages.
![Approach
• @Dorsetpolice Twitter account
– 3,200 posts [2011-12-23 / 2014-06-12]
• Announcements, appeals, crime reports, etc.
• Retweet as engagement indicator
– 2,270 (86%) seed vs. 430 non-seed posts
• Seed post = received at least one retweet
• Balance dataset for the analysis
• ML & Semantic techniques to analyse engagement](https://image.slidesharecdn.com/socinfo2014v2-141114093308-conversion-gate01/85/SocInfo2014-CityLabs-Workshop-4-320.jpg)
![ML Analysis
• Content Analysis [Rowe & Alani 2014]
– Top discriminative features of retweeted vs. non
retweeted posts
– Top discriminative features of highly retweeted
posts
• Tweets that generate higher attention
– Are longer
– Have positive sentiment
– Mention other users
– Are posted between
8:00 a.m 16:00 p.m
[Rowe & Alani 2014]. Mining and comparing engagement dynamics across multiple
social media platforms. WebScience 2014](https://image.slidesharecdn.com/socinfo2014v2-141114093308-conversion-gate01/85/SocInfo2014-CityLabs-Workshop-5-320.jpg)
SocInfo2014 CityLabs Workshop
- 1. Policing Engagement via Social Media Date: November 2014, SocInfo Author: Miriam Fernandez, Amparo Elizabeth Cano, Harith Alani
- 2. Context • Policing organisations use social media to spread the word on crime, severe weather, missing people, … • Many forces have staff dedicated to this purpose and to improve the spreading of key messages to wider social media communities • Research shows that exchanges between police and citizens are infrequent
- 3. Goal • Understand what attracts citizen’s to social media policing content – What are the characteristics of the content that generate higher attention levels • Writing style • Time of posting • Topics – Help police forces to identify actions and recommendations to increase public engagement
- 4. Approach • @Dorsetpolice Twitter account – 3,200 posts [2011-12-23 / 2014-06-12] • Announcements, appeals, crime reports, etc. • Retweet as engagement indicator – 2,270 (86%) seed vs. 430 non-seed posts • Seed post = received at least one retweet • Balance dataset for the analysis • ML & Semantic techniques to analyse engagement
- 5. ML Analysis • Content Analysis [Rowe & Alani 2014] – Top discriminative features of retweeted vs. non retweeted posts – Top discriminative features of highly retweeted posts • Tweets that generate higher attention – Are longer – Have positive sentiment – Mention other users – Are posted between 8:00 a.m 16:00 p.m [Rowe & Alani 2014]. Mining and comparing engagement dynamics across multiple social media platforms. WebScience 2014
- 6. Semantic Analysis • Annotate tweets with entities & concepts using TextRazor – DBPedia & Freebase as knowledge bases • Seed posts talk about – weather, roads and infrastructures & mention locations • Non seed posts talk about – crimes such as burglary, assault or driving under the influence of alcohol.
- 7. Conclusions • Use a combination of ML and semantic techniques to understand the most discriminative language, time features and topics of those tweets generating higher attention levels towards policing content • Preliminary study – One social media platform (Twitter) – One police force(@dorsetpolice) – One engagement indicator (retweets)
- 8. Questions?