Identifying and Pruning Features for Classifying Translated and Post-edited Gaze Durations
| Autor: | Sivaji Bandyopadhayay, Tanik Saikh, Dipankar Das |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
business.industry
Computer science Pattern recognition 02 engineering and technology Translation (geometry) Machine learning computer.software_genre Gaze Cross-validation Set (abstract data type) Data set Support vector machine 03 medical and health sciences 0302 clinical medicine 0202 electrical engineering electronic engineering information engineering Eye tracking 020201 artificial intelligence & image processing 030212 general & internal medicine Artificial intelligence Pruning (decision trees) business computer |
| Zdroj: | Mining Intelligence and Knowledge Exploration ISBN: 9783319581293 MIKE |
| DOI: | 10.1007/978-3-319-58130-9_12 |
| Popis: | The present paper reports on various experiments carried out to classify the source and target gaze fixation durations on an eye tracking dataset, namely Translation Process Research (TPR). Different features were extracted from both the source and target parts of the TPR dataset, separately and different models were developed separately by employing such features using a machine learning framework. These models were trained using Support Vector Machine (SVM) and the best accuracy of 49.01% and 59.78% were obtained with respect to cross validation for source and target gaze fixation durations, respectively. The experiments were also carried out on the post edited data set using same experimental set up and the highest accuracy of 71.70% was obtained. Finally, Information Gain based pruning has been performed in order to select the best features that are useful for classifying the gaze durations. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|