Annotating a driving experience corpus with behavior and subjectivity

Personality is the important internalframework that we use when wecommunicate with the others; thus, it iscritical for vehicle-driver communication.Evaluating an individual's personalityrequires associating behavior andsubjectivity. To develop the vehicle-drivercommunication system that will be mostapplicable to actual driving settings, acorpus is necessary that includes colloquialand daily experiential expressions alongwith their emotions, polarity, andsentiments. In addition, the corpus mustinclude a wide range of subjectivemeasures, such as human judgements,perceptions, and cognitions during cardriving. Thus, we construct a drivingexperience corpus (DEC) that constitutes253 blog articles (7,831 sentences) with thefollowing four manually annotated tags: (1)driving experience (DE), (2) other'sbehavior (OB), (3) self-behavior (SB) and(4) subjectivity (SJ). In this paper, wedescribe the guidelines, corpusspecification and agreement analysisbetween annotators. We identified threedifficulties: the extended self, importantinformation, and voice in mind. Weconducted automatic annotation experimenton the corpus using Conditional RandomFields (CRF). The results indicated FScores of .768, .478, .534, and .749 for DE,OB, SB, and SJ, respectively, on the testset. Our error analysis reveals difficulties ininterpreting nominatives and recognizingfacts.