Hastings Janna, Brass Andy, Caine Colin, Jay Caroline, Stevens Robert
Cheminformatics and Metabolism, EMBL - European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, CB10 1SD UK ; Swiss Center for Affective Sciences, University of Geneva, Geneva, Switzerland.
School of Computer Science, University of Manchester, Oxford Road, Manchester, M13 9PL UK ; Faculty of Life Sciences, University of Manchester, Oxford Road, Manchester, M13 9PL UK.
J Biomed Semantics. 2014 Sep 3;5:38. doi: 10.1186/2041-1480-5-38. eCollection 2014.
We evaluate the application of the Emotion Ontology (EM) to the task of self-reporting of emotional experience in the context of audience response to academic presentations at the International Conference on Biomedical Ontology (ICBO). Ontology evaluation is regarded as a difficult task. Types of ontology evaluation range from gauging adherence to some philosophical principles, following some engineering method, to assessing fitness for purpose. The Emotion Ontology (EM) represents emotions and all related affective phenomena, and should enable self-reporting or articulation of emotional states and responses; how do we know if this is the case? Here we use the EM 'in the wild' in order to evaluate the EM's ability to capture people's self-reported emotional responses to a situation through use of the vocabulary provided by the EM.
To achieve this evaluation we developed a tool, EmOntoTag, in which audience members were able to capture their self-reported emotional responses to scientific presentations using the vocabulary offered by the EM. We furthermore asked participants using the tool to rate the appropriateness of an EM vocabulary term for capturing their self-assessed emotional response. Participants were also able to suggest improvements to the EM using a free-text feedback facility. Here, we present the data captured and analyse the EM's fitness for purpose in reporting emotional responses to conference talks.
Based on our analysis of this data set, our primary finding is that the audience are able to articulate their emotional response to a talk via the EM, and reporting via the EM ontology is able to draw distinctions between the audience's response to a speaker and between the speakers (or talks) themselves. Thus we can conclude that the vocabulary provided at the leaves of the EM are fit for purpose in this setting. We additionally obtained interesting observations from the experiment as a whole, such as that the majority of emotions captured had positive valence, and the free-form feedback supplied new terms for the EM.
EmOntoTag can be seen at http://www.bioontology.ch/emontotag; source code can be downloaded from http://emotion-ontology.googlecode.com/svn/trunk/apps/emontotag/and the ontology is available at http://purl.obolibrary.org/obo/MFOEM.owl.
我们评估情感本体（EM）在国际生物医学本体会议（ICBO）上听众对学术演讲的反应背景下，在情感体验自我报告任务中的应用。本体评估被视为一项艰巨的任务。本体评估的类型范围从衡量对某些哲学原则的遵循程度、遵循某些工程方法，到评估是否适合目的。情感本体（EM）表示情感及所有相关的情感现象，并且应该能够实现情感状态和反应的自我报告或表达；我们如何知道情况是否如此呢？在这里，我们在“实际应用”中使用EM，以便通过使用EM提供的词汇来评估EM捕捉人们对某种情况的自我报告情感反应的能力。
为了实现这种评估，我们开发了一个工具EmOntoTag，在这个工具中，听众能够使用EM提供的词汇来捕捉他们对科学演讲的自我报告情感反应。我们还要求使用该工具的参与者对EM词汇术语用于捕捉他们自我评估情感反应的适当性进行评分。参与者还能够使用自由文本反馈功能对EM提出改进建议。在这里，我们展示所捕获的数据，并分析EM在报告对会议演讲的情感反应方面是否适合目的。
基于我们对这个数据集的分析，我们的主要发现是，听众能够通过EM表达他们对演讲的情感反应，并且通过EM本体进行报告能够区分听众对演讲者的反应以及演讲者（或演讲）本身之间的差异。因此我们可以得出结论，在这种情况下，EM底层提供的词汇适合目的。我们还从整个实验中获得了有趣的观察结果，例如所捕捉到的大多数情感具有正效价，并且自由形式的反馈为EM提供了新的术语。
