[Published in the OSALL newsletter of May 2011]
The Internet is a really strange environment in which to find myself not only working but positively revelling when I have time to spare. I say this because, for as long as I can remember, there has been little that makes my heart drop as much as going into a library or shop with no particular mission in mind. So much of my time is spent looking for specific information for either myself or on request for others that the thought of just ‘browsing’ is overwhelming.
As so often happens in life, and an aspect that thoroughly fascinates me, a number of recent threads came together in the last week relating to how and why the wide array of information that bombards us constantly is brought to our attention.
The first of these was a Twitter topic relating to how celebrities and Gareth Cliff in particular may or may not be producing sponsored tweets. While I used to enjoy some of Cliff’s humour, I haven’t followed him for a long time and can safely say I am not influenced if he does make use of this practise. There will be plenty of information available online but, in brief, companies are paying well-known personalities to post anything from a 140-character Tweet to blog posts endorsing a particular brand or product and, needless to say, the amounts they pay are not mean. The ire of many online users was raised because having unsolicited commercial information posted by an apparently unrelated source could be perceived as an abuse of one’s trust.
The second thread came in the form of a brief blog post entitled What comes after keyword search? 1. It drew my attention to the field of recommendation engines. Although most of us will be familiar with the practicalities, I hadn’t given much thought to the variety of predictive mechanisms that are used to try and meet the expectations of visitors and, where commercial interests reign, to tempt potential shoppers into further purchases.
We have become used to the way Google’s Page Rank works, relying on the number and to an extent identity of links to a webpage to display search results in a useful way. Google has also become known for using personal browsing habits to formulate profiles and provide ‘personalised’ results ; if the relevant settings are utilised, one will also get results based on one’s location. More recent approaches to identifying useful links have brought a number of other factors to the table.
While there are a number of ways of classifying how websites try to make sense of the statistics available to them, The art, science and business of recommendation engines2has identified four categories : personalised, social and item recommendations, and a combination of these approaches. The first relies on statistics relating to the individual visitor’s usage of the site ; the second is based on the way other visitors have used the site, and the third relates to individual items.
Every article I read agrees that Amazon leads the way in providing useful suggestions about products that may be of interest. Not surprising considering the resources at their disposal and the potential to attract more sales if they get the algorithms right.
In 2000 genetics was brought into the equation with the development of a service that could offer music based on as little as one’s choice of a single piece3. Of course many man hours have been put into breaking down common features to find ‘similar’ music, not as straightforward as it sounds. ”The natural question is can this genes-based approach be applied to other areas – like books, movies, wines, restaurants or travel destinations? What constitutes genes for each category?”
“So if the genes are the attributes of the object that make it unique in our mind, we should have no problem coming up with genes for various things. In the past few years we have been doing this a lot online. It’s called tagging!”2
Tagging is of course one of Del.icio.us8’s primary features. The same article2 makes an interesting observation : “the del.icio.us approach holds intriguing possibilities of self-organizing classification and recommendation systems. With enough users and more tweaking, social tagging can result in a system that works equally well for books, wine and music . . . these approaches hold the promise to provide instant gratification, without asking the user to reveal her preferences and past history”.
Visiting sites with the express purpose of identifying how they try to anticipate my requirements and interests is proving to be a fascinating exercise. Examples I’ve referenced below include Huffington Post’s experiment in bringing relevant news articles to the attention of visitors7, Goodreads Book Club3 (infinite browsing potential here based on any number of recent statistics), and Etsy’s Taste Test5. In the case of the latter, the site lists about 8 million items so there has been an urgent need to find ways of catching the interest of visitors by drawing their attention to specific items that seem to be to their taste.
Now the downside.
“Perhaps the biggest issue facing recommender systems is that they need a lot of data to effectively make recommendations” 6
This is where we come in. As with anything in life, and technology especially, what comes out is only as good as what goes in. We can all benefit from relevant suggestions by others, so do we take the time to make some input when we visit sites that ask for feedback? If we have enjoyed a book or product, why not say so and allow others to benefit from our experience?
At the end of last year I found myself urgently needing to find a reputable kennel for our old lady for a few days. One of Twitter’s claims to fame is the potential it has for spot surveys. I must admit that those surveys I’ve investigated don’t generally attract a huge number of voters, but they do provide an interesting cross-section of opinions. So I turned to Twitter and asked fellow Pietermaritzburg residents for suggestions. Fortunately Maritzburg is quite well represented on Twitter and we are a fairly vocal bunch. More than one person suggested a particular kennel and cattery just out of town that I hadn’t even heard of. It turned out to offer superlative service at a reasonable rate and I couldn’t have been happier. Neither could our Staffie who was even offered heated premises during a particularly cold spell. Fortunately she didn’t come home with higher expectations but she certainly won’t object if she needs to board there again.
These are exciting times as we watch the melding of Web 2.0 and 3.0. Let’s be a part of it.
1 What comes after keyword search? Legal Talk Network. March 2011
2 The art, science and business of recommendation engines. 16 January 2007
3 The Music Genome Project
4 Good Reads
5 Etsy Taste Test
6 5 problems of recommender systems / Richard MacManus. 28 January 2009
7 Stories you might like : join our beta program to test HuffPost recommendations. 6 January 2011
9 Strands Recommender
10 Aggregate Knowledge
Opinions expressed in this column are my own and not necessarily those of my employer.