Abstract
The processing of textual data to obtain psychological measures is becoming prevalent. However, the problem is that in most cases, tagged corpora are unavailable and therefore conventional classification methods cannot be easily applied. We present a novel solution for measuring psychological dimensions in textual data. The solution is based on (i) the construction of patterns from validated psychological questionnaires measuring the target dimension and (ii) similarity measurements of some of the text's sentences to the pre-defined patterns. Our proposed methodology was tested on (i) the identification of the five personality dimensions in a gold standard corpus, (ii) the measurement of depression and (iii) the screening of civilian mass murderers, gaining preliminary support for its validity.
| Original language | English |
|---|---|
| Pages (from-to) | 1408-1414 |
| Number of pages | 7 |
| Journal | Computer Journal |
| Volume | 59 |
| Issue number | 9 |
| DOIs | |
| State | Published - 1 Sep 2016 |
Keywords
- NLP
- computational personality
- computational psychology
ASJC Scopus subject areas
- General Computer Science
