Big Five personality traits Analysis in Turkish tweets is implemented with Bert
The dataset was created by collecting Tweets from Twitter users in line with a survey we conducted for Big Five personality traits.
Notebooks are self-explanatory. Separate models were trained for each of Big Five personality traits. In addition, models were trained for 5 different traits such as "Satisfaction with Life" and "Income Level". You can take a look at the details of the training by reviewing the "training_example.ipynb" notebook.
We obtained separate results for each model.
For "Agreeableness" :
The analysis is generated as follows:
| name | perc(%) | count |
|---|---|---|
| muhalefet | 74.6% | 1838 |
| iktidar | 25.3% | 624 |
| total | 100% | 2462 |
| name | perc(%) | count |
|---|---|---|
| karamsar | 27.7% | 683 |
| memnun | 72.2% | 1779 |
| total | 100% | 2462 |
| name | perc(%) | count |
|---|---|---|
| alkol-olumlu | 40.8% | 1005 |
| alkol-olumsuz | 59.1% | 1457 |
| total | 100% | 2462 |
| name | perc(%) | count |
|---|---|---|
| din-olumlu | 26.4% | 652 |
| din-olumsuz | 73.5% | 1810 |
| total | 100% | 2462 |
| name | perc(%) | count |
|---|---|---|
| öğrenci | %15.6 | 385 |
| tüccar | %22.0 | 543 |
| işçi | %30.1 | 742 |
| memur | %32.1 | 792 |
| total | %100. | 2462 |
| name | perc(%) | count |
|---|---|---|
| agree | 39.6% | 975 |
| cons | 83.1% | 2046 |
| extra | 88.2% | 2173 |
| nevro | 11.9% | 294 |
| open | 37.2% | 918 |
| (from:Twitter Username) lang:tr -filter:replies - 2462 |