A Visual Map of Public Mementos and Conjectures

Pepe, Alberto

Permalink

https://hdl.handle.net/2142/15150 Copy

Description

Title

A Visual Map of Public Mementos and Conjectures

Author(s)

Pepe, Alberto

Issue Date

2008-02-28

Keyword(s)

• information visualization
• mood analysis
• data mining

Abstract

"How do we collectively feel about our future? Do we look forward to it with anxiety or vigor? Are we apprehensive or optimistic of what the future will bring? Since mood affects performance and well-being, the answers to these questions matter greatly to anyone concerned with public policy. The web is awash with material indicative of public mood, collective forecasting and personal relics. Several efforts have been undertaken to assess emotional status from online sources such as blogs, emails, web sites (Balog & De Rijke 2006) and search engine queries (see, for example, Google Trends). However, these efforts are limited, by the nature of their source material, to hindsight and near-present observations. The work presented here is concerned with collective speculations about the future. We present a visual analysis of publicly available textual content from futureme.org, a popular web service that allows its users to send themselves emails to be delivered at a later date, up to 30 years in the future. Many of these emails resemble ""confessional"" time capsules: their content is intended to project the user’s present emotional state at the origination date towards the indicated delivery date. These emails fall into two broad categories of content: a) conjectures about the future and b) mementos regarding the present or the past. By aggregating mood indicators extracted from messages directed to future dates, we can thus assess short and long term shifts in the collective emotional perception toward a particular point in the future. This principle is related to ""wisdom of crowd"" phenomena as observed in finance and prediction markets (Surowiecki 2004). Numerous psychometric instruments to assess individual mood states and monitor their fluctuations over time exist, the most prominent of which is the 65 item Profile of Mood States (POMS) questionnaire (McNair, Loor, & Droppleman 1971). The 6 dimensional factor analytical structure of the POMS (tension, depression, anger, vigor, fatigue and confusion) has been validated repeatedly (Norcross, Guadagnoli, & Prochaska 2006) and applied in hundreds of studies since its inception (McNair, Heuchert, & Shilony 2003). To make the POMS questionnaire applicable to the open-ended nature of email content, we extended the POMS set of 65 adjectives by nearly 793 synonyms using WordNet and Roget’s Thesaurus. We calculated the occurrences of extended POMS terms in the content of 30,000 publicly available ""future"" emails and mapped them to a normalized six-dimensional mood vector representing levels of tension, depression, anger, vigor, fatigue and confusion. These mood vectors were grouped according to the delivery date of the original email, resulting in a set of mood state vectors. Statistically significant mood changes were detected especially for depression and vigor indicators. The computation of mood levels was then implemented with a more specific textual analysis of the entire email corpus, aimed at identifying manifestations of conjectures and mementos. The results, presented in this poster, blend two different visual representation of the content analyzed: an ""emotional timeline"" - a cumulative depiction of mood levels between 2007 and 2036 - and a superimposed topic map of mementos and conjectures - an ontological model of commonly used terms and adjectives illustrating the chains of word association."

Type of Resource

text

Permalink

http://hdl.handle.net/2142/15150

Owning Collections