An “icon” is an ideogram that conveys its meaning through its pictorial resemblance. The term “icon aggregator” refers to an icon library that is created through uploads from multiple creators. My thesis has involved redesigning an icon aggregator search engine with the help of machine learning, so that it can effectively organize icons by their visual similarities. Its goal is to help users of an icon aggregator service, by reducing the time and effort spent searching for icons, and thereby let them more fully utilize systems of an icon aggregator like Noun Project.
Icons vary in their level of detail, scale, weight, and style depending on their themes and purposes. In order to create coherent, digital environments for end users of devices like mobile phones, interface designers strive to ensure visual consistency in the use of icons. Without such consistency, a set of icons falls apart as a visual language and becomes more like a random jumble of doodles. The key to effectively transforming a set of icons into a universal language is consistency.
Current icon aggregator systems have some pros and cons. The pros are the high number of collections and the diversity in style. The Noun Project, which is the largest example of an icon aggregator, has over 2 million curated icons created by a global community. The downside to using Noun Project is its lack of consistency throughout different collections and the fact that its tagging-based search system doesn’t help solve the problem. Due to this structure, users have to rely on infinite scrolling and bit of luck, in order to find what they’re looking for. Icon aggregators like Noun Project not only cost users a lot of time and effort, but also fail to effectively allow users to utilize its amazing number of collections.
My thesis, which involves the redesign of an icon aggregator search engine, approaches the problem with the use of machine learning. Image feature extraction has been widely used for other types of search engines and has the potential to be applied to icon sorting as well. Organizing icons by their visual similarities, rather than only by tags, makes the icon searching experience easier and quicker, and encourages collaboration within the community of users by effective cross-collection search that is not limited to an individual collection.