Big Data Summit 2014: The way we present web search data could influence election outcomes, claims Yahoo

"How do you define interestingness?" asks director of research

Yahoo's grasp on search algorithms and its presentation of search results to users could theoretically influence the way people vote in elections, the company's director of research has claimed.

Speaking at Computing's 2014 Big Data Summit today, Alex Jaimes told the audience how the company is continually working to keep its web and image capabilities smart, and to ensure it is able to read the kind of unstructured, human behaviour that drives user search language.

"We had to build an algorithm," he said.

"Our first question was ‘How do you define interestingness?'"

"Context is key," said Jaimes, before revealing that Yahoo had run research on how US citizens used search engines to look up phrases such as "Obama birth certificate" and "Why Romney will win".

Jaimes said that Yahoo "realised that showing this data was a big responsibility, because it could potentially impact how people voted".

Starting to categorise users automatically into groups such as "casual" and "professional", and interaction states such as "searching" and "surfing" behaviour is just the tip of the iceberg, Jaimes explained, as he urged the big data industry to keep innovating.

"We need to innovate at this juncture, as we're still trying to work out what we're doing with big data," said Jaimes.

"Big data only has value if it brings insights."