Academia.eduAcademia.edu

Figure 1 – uploaded by desi wulandari

See full PDFdownloadDownload figure
Fig. 1. Research overview diagram A. Data Collection

Figure 1 Research overview diagram A. Data Collection

Related Figures (5)

TABLE 1. Threshold Values for each Classifier in English Dataset The predicted result for each method can be different from one another. To avoid confusion in the conclusions, we propose a combined result which taken from three existing method’s results. The combined result uses majority vote of the three methods. For example, if on a label there are two or more methods with the “Yes” prediction, then the final prediction is “Yes”. Table 3 shows combination possibilities for combined method. ‘ABLE 3. Combination Possibilities for Combined Method
TABLE 1. Threshold Values for each Classifier in English Dataset The predicted result for each method can be different from one another. To avoid confusion in the conclusions, we propose a combined result which taken from three existing method’s results. The combined result uses majority vote of the three methods. For example, if on a label there are two or more methods with the “Yes” prediction, then the final prediction is “Yes”. Table 3 shows combination possibilities for combined method. ‘ABLE 3. Combination Possibilities for Combined Method
Support Vector Machine (SVM) is a supervised learning algorithm that analyzes the data and recognizing patterns used for classification [17]. SVM take the set of training data and marking it as part of a category then predicts whether the test document is a member of an existing class. SVM models represent the data as a point in space divided by a line/hyperplane. Optimum hyperplane search function shown in (4) subject to (5). Scoring in SVM to find the test document class use (6).
Support Vector Machine (SVM) is a supervised learning algorithm that analyzes the data and recognizing patterns used for classification [17]. SVM take the set of training data and marking it as part of a category then predicts whether the test document is a member of an existing class. SVM models represent the data as a point in space divided by a line/hyperplane. Optimum hyperplane search function shown in (4) subject to (5). Scoring in SVM to find the test document class use (6).
TABLE 2. Threshold Values for each Classifier in Indonesia Dataset
TABLE 2. Threshold Values for each Classifier in Indonesia Dataset
Combined method that is the final prediction of the application, got the best results on respondent testing with an accuracy of 65%. The combined method can produce better accuracy because there were improvements in the classification results. In the case where one of the methods fails at classifying, it will be covered by the two other methods if the predictions of other methods are correct. Overall accuracy is not quite good, but it can show that automatic text personality recognition can be alternative for questionnaire based test.
Combined method that is the final prediction of the application, got the best results on respondent testing with an accuracy of 65%. The combined method can produce better accuracy because there were improvements in the classification results. In the case where one of the methods fails at classifying, it will be covered by the two other methods if the predictions of other methods are correct. Overall accuracy is not quite good, but it can show that automatic text personality recognition can be alternative for questionnaire based test.
TABLE 5. Accuracy of Indonesia dataset The next scenario is respondent testing. This test intended to determine how this automatic personality prediction system fares against more traditional personality prediction. Right now one of most popular personality prediction test is online questionnaire based test. Therefore, we compare the system’s result (Fig. 3) with the predicted result from the IPIP 50-Item Set of IPIP Big-Five Factor Markers questionnaires [18]. System will retrieve text data from users Twitter account and classified using three methods and the combined method. The users then complete questionnaire test and report the results. Respondents consisted of 40 people. Respondents chosen must have a Twitter account with a minimum number of 1.000 tweets. Classification language selection is based on user’s primary language. The test results are shown in Table 6.
TABLE 5. Accuracy of Indonesia dataset The next scenario is respondent testing. This test intended to determine how this automatic personality prediction system fares against more traditional personality prediction. Right now one of most popular personality prediction test is online questionnaire based test. Therefore, we compare the system’s result (Fig. 3) with the predicted result from the IPIP 50-Item Set of IPIP Big-Five Factor Markers questionnaires [18]. System will retrieve text data from users Twitter account and classified using three methods and the combined method. The users then complete questionnaire test and report the results. Respondents consisted of 40 people. Respondents chosen must have a Twitter account with a minimum number of 1.000 tweets. Classification language selection is based on user’s primary language. The test results are shown in Table 6.
Related topics:
Machine Learning
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved