The LineUp software package enables Internet users to check out the information behind published rankings and also to create and visualise tailored rankings based on their own personal preferences.
Rare indeed nowadays is the person who has never seen and made use of online rankings. Whether we are talking about the best French films or the ‘car of the year’, these rankings, which proliferate on the Internet, can prove extremely helpful. But who exactly is behind a given set of rankings? While some high-profile people have sufficient credibility to put forward their own rankings, most of the assessments to be found on the Internet are posted without any explanation. And while they may be useful for gaining an overall view of a product or service, it is nevertheless difficult to get hold of a ranking list that corresponds to particular features or factors that you are especially looking for. In response to these two basic drawbacks, graduate students at the Harvard School of Engineering and Applied Sciences (SEAS) have developed a piece of open source software called LineUp, which enables you to quickly build your own rankings using information search algorithms and lets you assign weightings to different parameters.
Dynamic visualisation software
LineUp is the first dynamic visualisation software of its kind. It is part of a larger software package called Caleydo, an open-source framework designed to help visualise genetic data and biological pathways. LineUp’s original purpose was, for example, to analyse and characterise cancer subtypes. “LineUp really was developed to address our need to understand the ranking of genes by mutation frequency and other clinical parameters in a group of patients. It’s an ideal tool to create and visualise complex combined scores of bioinformatics algorithms,” explains Hanspeter Pfister, An Wang Professor of Computer Science at SEAS and co-founder of the project. The Harvard research team decided to apply LineUp to rankings, with the aim of extracting any inherent subjectivity. The programme draws solely on raw information available on the Internet, and offers an evaluation system which assigns weightings to the various parameters. So with a few clicks the user can build an entirely unique, personalised scale of values corresponding to his/her own subjective views. One example of its use might be to find the ‘best’ hotel or restaurant in town. The algorithm incorporates and analyses the various parameters to which the user has assigned weights, and then in a few seconds comes up with a corresponding list with rankings, presented in a highly visual way. “Essentially, it’s a tool to allow people to explore the complexity of reality,” underlines Alexander Lex, a post-doctoral visualisation researcher at SEAS who is another of the project’s co-creators.
How objective can you be?
There remains the question of just how objective rankings of this kind can ever be. For instance, if the algorithm lets you rank cars in terms of their mechanical features, what would be the intrinsic value of such an assessment? And when it comes to rankings of restaurants, hotels, films, books etc., the information used by the algorithm could well be useful but it is still based on very subjective data. When an audience grades a film, this might at first glance appear to be objective data, but it is in fact ‘aggregated subjectivity’ rather than an objective view. This is why an assessment put forward by a specific person whom you trust and whose tastes you know is often more meaningful than a mathematical calculation. Nor should we forget that subjective rankings can enable consumers to discover new things which at first sight they might not have thought they would like.