In Search of Design Inspiration: A Semantic-Based Approach
In Search of Design Inspiration: A Semantic-Based Approach
复制标题
寻找设计灵感：基于语义的方法
DOI：
--
复制
发表时间：
2010
影响因子：
3.1
通讯作者：
C. Bouchard
中科院分区：
文献类型：
--
作者：
R. Setchi;C. Bouchard
灵感来源帮助设计师定义其设计的背景，并反映其新产品的情感影响。通过观察和解释灵感来源，设计师形成术语词汇表，颜色调色板或带有图像的情绪板，表达他们的感受，激发他们的创造力并帮助他们传达设计概念。这些想法是欧盟资助的项目TRENDS背后的动机，该项目旨在开发一种软件工具，通过为概念汽车设计师提供各种灵感来源，支持设计的灵感阶段。本文集中在OntoTag，基于语义的图像检索算法的趋势项目内开发，和它的评价。OntoTag使用来自通用词汇本体OntoRo的概念，以及来自特定领域本体CTA的语义形容词，为设计师索引TRENDS数据库中的图像，为设计师提供一定程度的意外发现并激发他们的创造力。基于语义的算法涉及以下四个步骤：（i）创建从网络检索的文档和图像的集合，（ii）对于每个文档，识别图像周围文本中最常用的关键字和短语，（iii）识别每个文档中表示的最强大的概念，以及（iv）对识别的概念进行排名并将其与集合中的图像相关联。OntoTag与早期的方法有很大的不同，因为它不依赖于机器学习和标记语料库的可用性。它的主要创新之处在于用词的单义性和多义性来衡量词属于某个概念的概率。所提出的方法说明了基于软件工具的需求，参与趋势项目的两个工业合作者的例子。
Sources of inspiration help designers to define the context of their designs and reflect on the emotional impact of their new products. By observing and interpreting sources of inspiration, designers form vocabularies of terms, pallets of colors, or mood boards with images, which express their feelings, inspire their creativity and help them communicate design concepts. These ideas are the motivation behind the EU-funded project TRENDS, which aimed at developing a software tool that supports the inspirational stage of design by providing designers of concept cars with various sources of inspiration. This paper concentrates on OntoTag, the semantic-based image retrieval algorithm developed within the TRENDS project, and its evaluation. OntoTag uses concepts from a general-purpose lexical ontology called OntoRo, and semantic adjectives from a domain-specific ontology for designers called CTA, to index the images in the TRENDS database in a way which provides designers with a degree of serendipity and stimulates their creativity. The semantic-based algorithm involves the following four steps: (i) creating a collection of documents and images retrieved from the web, (ii) for each document, identifying the most frequently used keywords and phrases in the text around the image, (iii) identifying the most powerful concepts represented in each document, and (iv) ranking the concepts identified and linking them to the images in the collection. OntoTag differs significantly from earlier approaches as it does not rely on machine learning and the availability of tagged corpuses. Its main innovation is in the use of the words' monosemy and polysemy as a measure of their probability to belong to a certain concept. The proposed approach is illustrated with examples based on the software tool developed for the needs of two of the industrial collaborators involved in the TRENDS project.