We wrote about Abdur Chowdhury in our recent
coverage of the Open Data 2007 Workshop. Until recently he was running AOL Search,
but left to start a new venture. Abdur is an impressive thinker and articulate speaker,
with a great track record in the industry. So we decided to take a look at what he is up
to! As it turns out, he is developing an innovative vertical search engine for
shopping called Summize. This search engine
uses a technique called heatmaps, which we'll explore in this post.
A heatmap is a visualization technique for
displaying complex information in two dimensions, using colors. For example, heatmaps are
used in weather maps to display temperature or amount of rain. The coloring works like
this: you choose a range of values (for temperature on the chart to the right, the range
is from -60 to 120 F) and then map colors onto this range. There various ways of doing
this mapping, depending on how fine grained you need the map to be. In the case of just
two colors, say red and green, you can do it using RGB encoding - starting with red
moving towards black, and then passing black towards green.
Heatmaps are also actively used in biology, particularly in DNA microarrays. Recently, stats service CrazyEgg applied heatmaps to tracking what visitors do on a user's website. Their software captures user clicks on each page and then presents a summary in the form of a heatmap. The visual effect is stunning - you get instant insight as to what your visitors are doing on the site.

Just as CrazyEgg's application of heatmaps to click analysis is clever and useful, so is Summize's way of applying heatmaps to shopping. As with any search engine, you key into the search box what you are looking for. Summize searches its database and comes back with the results. It is the presentation of the results that makes this search engine very different. We see a summary of all product ratings presented as a heatmap, plus a digg-like voting mechanism and product thumb previews presented - all in one clean and simple interface.

Summize's infrastructure crawls user reviews all over the web and accumulates them together, into a single normalized rating database. The search results show the votes from this database using the heatmap. For example, in the figure above we see that the sum total of opinions of the Canon Powershot is 60% - very positive. Only 9% are quite negative and the rest are in-between. So this is essentially an instant recommendation for the family of products.
Beyond that, we get details for each matching product - including a product-specific heatmap. To get reviews of each product, you can click the product link.

Beyond informative heatmaps, Summize does a good job with other search result essentials. Firstly, it has three sorting knobs - by relevance, by red (bad) and by green (good). It also uses Ajax to show product previews on rollover - which is elegant, unobtrusive and useful. Another interesting feature is the voting up or down, which allows Summize to collect its own voting information. This looks like the beginning of a home-grown review system, to supplement the crawler results.
Finally Summize shows you the price next to each item, on Amazon. At the moment, the service is only affiliated with Amazon and offers Ajax popups to buy the item from Amazon. From the way the popup is designed, we can deduce that other sites will be integrated in the future.

Comparison shopping is an essential part of shopping and product research. Summize uses heatmaps to compare products. Again, the 'a picture is worth a thousand words' wisdom proves true. Seeing the heatmaps side by side gives the consumer a quick verdict on what other consumers collectively voted for:

Summize is impressive. Its power is in delivering a lot of information, using a relatively simple interface. And , like the Metacritic movie review aggregator which we profiled earlier this year, Summize brings together reviews and normalizes ratings to output essentially a single rating number.
The presentation is not only innovative, but useful. One thing that could hurt the adoption somewhat is the fact that there is a substantial percentage of the population with various forms of color blindness. But to address this issue, Summize could develop a version that uses grayscale.
Overall Summize, which is still under the radar, looks promising. Please take a look at it and let us know what you think.
TrackBack URL for this entry: http://www.readwriteweb.com/cgi-bin/mt/mt-tb.cgi/2054
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
This is a great post, Alex -- thank you. The whole product review space seems to be heating up (pardon the pun!). Do you know if anyone's done a comparison of some of the newer, independent review sites? One I know of is Wize.com, which has 1,000,000+ reviews online.
regards,
Graeme
Graeme,
We also looked at metacritic recently. Retrevo is doing it too, but from a different angle.
Alex
Really, really impressive !!!
Main drawback: the relevancy of the search results is not really good.
I think www.cubalaya.com is about to launch a similar app.
will be interesting to follow up how users respond to it.
B.
Pretty cool - re: relevancy, they are only in beta, so I'm sure their coverage is not particularly strong right now in every area...but its a great start!
This is great stuff! Thanks for the post. Don't think I've ever seen anything quite like it.
In my opinion, it's quite nice, but I find it hard to draw a quick (split second) conclusion from the colors. I prever the way wize.com does tackle this problem: convert the data into one number.
The use of a single number hides the data, you don't know if it everyone is happy or sad. With this visualization you can quickly scan the strip and understand all the reviews. All reviews sites with numbers or stars hid the details.