TwitterExplorer-Energy: Social Conversation

Click here for "Value Networks and Social Media Conversations" presentation at BECC2010.

Preliminary results of this research were presented at BECC2010, November 19, 2010, by Martha Russell and Camilla Yu.

A Tweetonomy-based Investigation of Energy-related Conversations

This research program is focused on ecolinguistics-based social media analytics aimed at understanding the social context and impact of the Stanford University ARPA-e research projects which individually and in concert aim to reduce energy use. This Department of Energy (DOE) funded research is focused on human behavior change via feedback from an array of energy monitoring sensors (both high and low resolution sensors) on online coverage of targeted behavioral descriptions, technologies, and polices.

Our broad research question asks, “How does the online conversation about energy efficiency change overtime?” We operationalize conversations to be overtime mentions in social media such as Twitter. Further, we aim to understand the “drivers” of shifts in the conversation and to leverage that understanding for initiatives that seek to change consumers' energy behavior. Our time series design includes monthly monitoring of conversation as well as collection of “online conversation stimulating events” such as national policy, new technology launches, and media events.

We examine both the content of a conversation (content is broadly defined as in a “tweetonomy”) and the social network of a conversation. We begin with a “Tweetonomy, ” a term created by Wagner and Stromeir (2010) to describe social awareness streams of twitter that include a message and its content, URLs, and other user based syntax such as hashtags, slashtags or @ replies. The contemporary communication conventions yield a large number of ways of considering communication: author(s) of messages, recipients of messages, the retweets or diffusion of information, the content of a message such as links embedded in messages, keywords or hashtags, and the time of messages.

This special version of TwitterExplorer-Energy is based on a dataset of tweets containing any of the 121 ecolinguistic terms being studied by the Stanford ARPAe  that were collected between 2010.09.03 and 2011.01.04.  Hashtags selected into this sample occurred in at least 3 of the 2.47 million tweets.  All 18,338 hastags are explorable on this special version of TwitterExplorer.

The eco-linguistic terminology used to crawl and collect the tweets was developed at Stanford University by Drs. June Flora, Carrie Armel, and Martha Russell, under sponsorship from the US Advanced Research Projects Agency for Energy, and Media X at Stanford University. A preliminary analysis of this data was presented at BECC2010, November 19, 2010, by Martha Russell and Camilla Yu.  For information about Understanding the Role of Social Media in Changing Consumer’s Energy Behavior, contact Martha.Russell@stanford.edu.

This instance of Twitter Explorer is populated with data collected and analyzed by the Active Intelligence Laboratory, University of Electro-Communication, Tokyo Japan, directed by Dr. Neil Rubens, assisted by Rafael Perez. For information about collection of tweets, please refer to ActiveIntelligence.org.

TwitterExplorer is a tool for exploring hashtags on Twitter, created by Jan Pöschko. The initial project was partly motivated by the Language Processing course (“Språkteknologi”) at KTH Stockholm, but will also be the basis for future research in the group of Markus Strohmaier at Graz University of Technology. Therefore, it is still subject to active research and development. You can read the report, see the slides of the short presentation at KTH, and the initial clustered graph of the 1000 most frequent hashtags and their co-occurrences.

For information about Understanding the Role of Social Media in Changing Consumer’s Energy Behavior, contact Martha.Russell@stanford.edu.