Computational Personality Recognition (Shared Task)
The Workshop on Computational Personality Recognition allowed participants to compare the results of their systems on a common benchmark. Unlike competitive shared tasks, the workshop did not focus just on performance, but rather on discovering which feature sets, resources, and learning techniques are useful in the extraction of personality from text. Organizers provided two gold-standard labeled data sets (released 1 February 2013): essays.zip, a large data set of 2400 stream-of-consciousness texts labeled with personality; and mypersonality.zip, consisting of approximately 10,000 Facebook status updates of 250 users, plus Facebook network properties (including network size, betweenness centrality, density, and transitivity) labeled with personality. Participants were required to use at least one of the data sets provided by the organizers for their experiments; provide the files used for the experiments; and submit a short paper reporting all the information about features, resources, and techniques used in the experiments, and discussing results.
The workshop consisted of eight participant teams who exploited a wide range of lexical resources and learning techniques. One team achieved a very high and promising result.
Francois Mairesse underlined the fact that, until now, performances of the available personality-recognition systems were not good enought to be useful for practical tasks. He listed possible fields of application for personality recognition: virtual assistants, human-computer interaction, health care (that is, mood detection, personality disorders), recommender systems and market segments (customer profiling). He suggested that we learn computational models mapping the big five personality domains to the dimensions of interest. Daniel Gatica-Perez presented a work on personality recognition from video. He used observer ratings in place of self-assessments to retrieve personality labels. He found connections among verbal and nonverbal cues and big five impressions.
After the poster session, Dejan Markovikj, who obtained the best result with his system, presented his work through Skype. The workshop concluded with a summary of the results and a discussion of the next challenges in personality-recognition research, including running personality recognition in languages other than English, comparing performances that can be achieved exploiting different personality-assessment tests; and finding new applications for computational personality recognition (such as sentiment analysis, deception detection, and mood detection).
Fabio Celli. Fabio Pianesi, David Stillwell, and Michael Kosinski served as cochairs of this workshop. This report was written by Fabio Celli. The papers of the workshop were published as AAAI Technical Report WS-13-01.
Social Computing for Workforce 2.0
The Social Computing for Workforce 2.0 workshop brought together researchers and practitioners to introduce new tools or methods that address workforce management issues, advance our understanding of workforce issues through qualitative and quantitative empirical studies, and discuss the impact and opportunities that recent trends such as social, mobile, and crowdsourcing have on the workforce. Social technologies used in the workplace are making it easier for workers to perform activities such as find information, share information, connect to others, find experts, find answers to questions, and infer workplace affect and emotion.
Social computing research in this area has advanced our understanding of technology in the workplace, but there remain several areas ripe for further investigation. This workshop sought to address the issues and challenges of new ways of working, by bringing in insights from researchers from multiple disciplines, as well as encouraging diverse methodological approaches to tackling them. …