A team from Syracuse University’s School of Information Studies took home the Best Full Research Paper Award at the annual iConference 2025 held from March 18-22 at Indiana University.

Alexander Owen Smith, a current iSchool PhD student, led the project as an extension of his completed dissertation. Jeff Hemsley, Interim Dean of the School of Information Studies and Associate Professor at the iSchool, and Una Joh, another current PhD student, were also on the team. The study was published March 11 in Information Research. 

Their study examined images of memes and how Google interprets the data behind those images, seeking to understand the algorithm’s ability to discern the underlying meaning of the image. 

The first meme they looked at was “Doge,” an image of a Shiba Inu dog; the second was “Hello There,” which is based on a scene in the film Star Wars; the third meme “Loss” reflects a pattern. 

The researchers wanted to see if Google API could predict a description of the meme’s cultural idea–or reverse engineer a web search for the meme–upon receiving a file of the image. In other words, the study tried to understand how the language of web search indicates an aspect of collective human memory of cultural ideas at a granular–or memetic–level, said Smith, who plans to study the topic more.

With regard to memetics–the study of memes–understanding how internet technology interprets the content can shape the way humans interpret culture with expressions based on our memories, the authors wrote.

The team used Google Cloud Vision API to look at the keywords and search terms that were attached to each image.

Then they looked at feature networks, which represent how the features (the label or web entity) are connected to each other based on common images containing both entities. The team also evaluated image networks, which show how individual images and files are connected based on common entities in both images.

“When Google fails to find a regular way to search for image files that indicate a similar meaning, it is because a ‘meme’ is not so clearly definable,” Smith said. “With lack of cultural definition in mind, this study was designed to understand how these gaps in web search occur.”

“Web search technologies provide an excellent way to explain this because when I use Google to search for something, I am performing the search by expressing what I remember using language as best as I can,” Smith explained. “Yet memetic content is often more based around senses like vision, sound, or even smell instead of language.”

Findings indicated that the more data can be put into a smaller number of labels, the less the internet has to “remember” to see a meme. When labels are not strongly connected, Google can’t understand some cultural information about a meme, the team wrote in its report.

“Certain parts of cultural memory are not reducible to computationally reducible visual data. More context is needed to understand what makes a meme important,” the team wrote. “We see this in each of our memes’ network data. However, we do see that Google can provide a partial understanding of what artefactual trace data frequently ties them together.”

“While our results suggest that Google is ‘good’ at understanding certain parts of 

memetic meaning, we find that it needs to be augmented through human interpretation to make sense of a lot of what it provides,” they added.

In addition to the Best Full Research Paper Award, iSchool researchers were also acknowledged in another category. PhD students Sarah Appedu and Yigang Qin were recognized as finalists in the Best Short Research Paper category for their study, “Discourses of Fear around AI and their Implications for Library and Information Science.”

The iConference is organized annually by the iSchools consortium, an international organization of over 130 universities on all continents that aims to unite university members’ research and teaching. The non-profit group was founded in 2005, and is based in Germany.