الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 124 |  |  |
| | | from | 124 | | |
Abstract
Personality detection attracts researchers nowadays. There are several domains that benefit from this area. In social media, personality knowledge enhances recommendations to users. The consequences are more advertisements and users.
The Enneagram is a personality model that illustrates desires, motivations, fears, features and problems. Psychiatrists use the Enneagram to understand the patient’s personality. Enneagram’s knowledge helps them to give the right support, thus they use the Enneagram assessment test to know the patient’s Enneagram. People are not eager to do a test because it takes time and effort. Enneagram personality detection is required as it does not consume time and effort. Twitter provides a huge amount of text that can be utilized. The Enneagram can be useful in education, human development and dating applications.
In this thesis, the Enneagram personality detection system is proposed. Enneagram personality detection utilizes Twitter text to detect the Enneagram personality. The system uses a combination of Enneagram ontology, English lexicon and a statistical approach. The Enneagram ontology is developed using METHONTOLOGY design principles and application-based evaluation.
The Enneagram personality detection system contains multiple phases: text preprocessing, word- based feature extraction, word-based feature selection and personality detection. Text preprocessing prepares text by removing unnecessary information, normalizing text, stemming and lemmatization. Word-based feature extraction converts text to words and represents it as a bag of words. Word- based feature selection chooses important features using Enneagram ontology and English lexicon. The personality detection phase applies a statistical approach. This approach computes the probability distribution of words related to each personality. The highest probability distribution across personalities is the predicted personality.
Two case studies are presented. The first case study uses Bill Gates Twitter account. The result of the case study is the Investigator personality which is equivalent to the Enneagram experts’ analysis. The second case study is applied on Morgan Freeman Twitter profile. The outcome of this case is the Peacemaker personality which is identical to the Enneagram experts’ analysis.
The Enneagram personality detection system uses the Enneagram collected dataset. The Enneagram dataset is composed of two sources. The first source is the official Enneagram Institute that mentions the celebrities’ Enneagram. The Twitter accounts of these celebrities are part of the dataset. The second source is the public Twitter users who posted their Enneagram assessment test results.
Different evaluations are computed from the output. These measures are precision, recall, f1 score and accuracy. The result varies across personalities. The highest three personalities’ recall are Enthusiast, Reformer and Investigator which are 95%, 83% and 81%. The medium personalities’ recall are Peacemaker 42%, Individualist 22%, Achiever 20% and Challenger 20%. The lowest two personalities’ recall are Helper 7% and Loyalist 4%. This system is the first to recognize the Enneagram from text. Enneagram is far more complex than other past-used personality models. The results are promising for further research on this point.