Pluribo: Instant Summaries of Amazon Reviews
Pluribo is a Firefox plugin that displays short summaries of product reviews on Amazon.com. Pluribo scans through reviews customers on Amazon have left and automatically creates a one sentence summary that is somewhat akin to a Zagat review. While Zagat uses human editors to compile its reviews, though, Pluribo is fully automated. Right now, Pluribo only works for the electronics section of Amazon's store, but the developers are planning to expand this to the rest of Amazon's offerings soon.
Here is a typical summary that Pluribo created for a SanDisk MP3 player: "This has been on the market for a while. Although there were objections to the software, users are happy with the low price, product support, and battery. If you don't care about the software, it has potential."
Besides summarizing the reviews for a given product, Pluribo also compares those reviews to other products in the same category to see where the reviews for this item were different. Pluribo will also give more weight to reviews that were considered 'helpful' by other Amazon customers. Pluribo keeps most of this information in the background. However, when hovering over a keyword in the summary, a small pop-up will display a list of relevant phrases Amazon's customers used to describe the item, as well as some more of the statistical data Pluribo used to compute its summary.
In testing out Pluribo, it consistently displayed accurate summaries of the actual user comments - a testament to how well the developers have tuned their algorithm to at least this limited range of product categories.
Pluribo's overall execution is quite seamless and Pluribo does not slow down the load times on Amazon, as it only gets to work after the page is fully displayed.
However, Pluribo seems quite restricted when it comes to what items it will display reviews for and for which it will just display a 'coming soon' message. Right now, it only works well for MP3 players, GPS navigation systems, and digital cameras. It also seems to work best for products that have been reviewed at least 30 times.
It would be nice to see Pluribo start pulling in reviews from other sources besides Amazon's own customers. The fact that it only works as a Firefox plugin is also going to limit its appeal to technically savvy users for the time being.
Overall, Pluribo is a fun and (when it's working) useful plugin - though for the time being, its a bit too limited to be of real help. It's real potential is only going to be realized once the developers get out a version that works across all of Amazon's offerings and maybe even expands beyond Amazon to include other online stores.
Comments
Subscribe to comments for this post OR Subscribe to comments for all ReadWriteWeb posts
No wonder people nowadays have ADD. How hard can it be to scan a few reviews? And if this becomes a hit, are people going to write their reviews SEO-style?
Seriously, who needs this product? Who is the buyer? Sales sites? The end-user? Product vendors? I can imagine a number of business models, none of which can be executed on, they just turn out that way.
Summarisation software products is very usefull for the legal profession. Lawyers can summarise a huge number of documents which allows them to narrow down the list of documents they need to start with. A defense team for a murder case could have scanned of thousands of documents or even more (I have heard some local court cases that this high number of legal documents that is common) to find out where to start.
This just narrows down the option. It doesnt mean that the other documents are going to be discarded, but it is important that the defense starts on the most important ones, to build the case and then they have the option to read the rest of the documents which might not be relevant at all to the case. There is time to be saved here.
I've just pointed out to my neighbor who is a commercial lawyer about a free Java applet on the net (source codes available) for single document summerization and he loves using this tool.
This technology will get better and better as natural language is being included with other numeric-based algorithms for summarisation.
It would be useful for publishers to summarise a whole book, for example to be used in promotion. There are many useful applications to be explored.