BROADCASTS.com

b"

When Google launched in 1998, a prime ingredient in their not-so-secret sauce was the question: if a user randomly clicked links where on the web might they end up?

They called the answer PageRank. This involved treating the web as a network rather than a bunch of isolated documents containing keywords. The outcome was a new verb and the near destruction of their competitors. Could repeating and refining 'the Google trick' help cultural bodies with research, collection care or digitisation?

One limitation to overcome is the assumption that all users behave in the same way. Users are individuals within fuzzy communities. So, can we personalise PageRank and treat people more like individuals than averages?

Matthew Pearce, from The National Archives, works on public sector information - in particular, its economics. His research is on the statistics and algorithms needed for personalised predictions.