透過腦機介面技術擷取資料進行認知負荷分類: 應用於服務設計的初步努力

The Classification of Mental Effort with BCI: The Preliminary Effort on Service Design

杜長城(Chang-Cheng Tu)

指導教授 ： 林福仁

國立清華大學/科技管理學院/服務科學研究所/碩士(2014年)

摘要

隨著服務設計越來越重要，許多方法論也發展出來已做使用者研究。然而這些方法論並非適用于所有情形，例如研究者無法直接觀察或記錄受試者。為開發一套分法論用以偵測受試者的心里狀態，此研究採用消費性腦機裝置，受試者穿戴此裝置並且接受英文聽力測驗，同時標示題目的難易度。我們採用支持向量機器，以五秒的特徵集來分類，並以 F-measure、精准率以及查全率來衡量，結果呈現最高的 F-measure 為 0.353。此外，部分結果顯示特徵值越多可能導致表現變差。此言就呈現了可行的分類器以及特徵集，並且設計現實生活中可能的語言學習方式。

關鍵字

腦電圖 ； 支援向量機器 ； 心理狀態 ； 英文聽力學習 ； 分類器

並列摘要

As the service design becomes more important and highly related to customers’ needs, there are several methodologies could be used for user research. However, these existed methodologies may not be able to fit to all situations; for example, researchers cannot directly observe and record users’ mental states. In this research, we adopt a commercial EEG-based device with single electrode in the setting of listening to English to record their mental states. This research aims to develop a classifier that can predict users’ mental efforts via experiments in which 35 college students were asked to listen to English and respond their mental states, easy or difficult, before choosing the answer for each question. We chose Support Vector Machine (SVM) as the classifier, and valuated its performance in terms of recall, precision, and F-measure in five different feature models with different time windows of extracting EEG data. The results indicate that one of the proposed features gives the highest F-measure of 0.353 with precision rate of 65.8% and recall rate of 57.1% from EEG data extracted in the time window of five seconds. Besides, in some situations, the more features we include for classification, the lower F-measure scores the system obtains. This research has contributed to the literature that a classification of mental effort using SVM classifier is effective to capture users’ mental states in the process of listening English and comprehension. The embedded SVM classifier could be used for detecting users’ mental state in service design process in the real world application on language learning activities. Additional efforts can be made to extend its applications on different service contexts.

並列關鍵字

Electroencephalography (EEG) ； Support Vector Machine (SVM) ； Mental state ； English Listening and comprehension ； classification

參考文獻

Mostow, J., Chang, K.-M., & Nelson, J. (2011). Toward exploiting EEG input in a reading tutor. Paper presented at the Proceedings of the 15th international conference on Artificial intelligence in education, Auckland, New Zealand.

Allen, J. J. B., Harmon-Jones, E., & CavenderJ.H. (2001). Manipulation of frontal EEG asymmetry through biofeedback alters self-reported emotional responses and facial EMG. Psychophysiology, 38, 685-693.

Bergen, D. (1988). ENcyclopedia of neuroscience. JAMA, 260(1), 104-104. doi: 10.1001/jama.1988.03410010112050

Bos, D. O. (2006). EEG-based Emotion Recognition - The Influence of Visual and Auditory Stimuli. Emotion, 57(7), 1798-1806. doi: 10.1109/TBME.2010.2048568

Bradley, M. M., & Lang, P. J. (1994). Measuring Emotion: The Self-assessment Manikin and the Semantic Differential Journal of Behavior Therapy and Experimental Psychiatry, 25, 49-59.