Aspectiva – Artificial Intelligence For Product Reviews
Artificial intelligence (AI) is a smoking hot technology at the moment with all kinds of money being poured into startups that use deep learning and/or machine learning to do things that traditional algorithms can’t. That last part is really the key as to why investments are being made in AI. Nobody cares that you are using AI unless your AI technology can do things that traditional programming approaches can’t. One example of this would be when an AI program beat a human in a game of “Go” and it was remarked that the “AI program did things humans never would have thought of”.
When a technology like AI is attracting a lot of funding, it’s very tempting for companies to just liberally scatter mentions of “AI” across their websites in hopes of attracting investors. We recently wrote an article about 9 applications of artificial intelligence that we didn’t think made any sense. For example, do you really think a dating site powered by AI is going to add any value beyond what a traditional matching algorithm can?
Another area we thought didn’t make sense to use AI for was that of product recommendations. Companies like Amazon know exactly what products to show you that you are most likely to buy. Using AI to figure out what products you are most likely to buy probably isn’t going to have enough incremental value that would justify the investment. Consequently, we made the claim that we wouldn’t see “AI used for product recommendations“. A company called Aspectiva wrote us to say that in fact there is an opportunity for AI to be used in product recommendations. While their technology is primarily used for aggregating product reviews as opposed to product recommendations, it is well worth taking a closer look at.
Founded in 2013, Aspectiva has taken in $2.8 million to develop a cloud-based service that turns user-generated content (like product reviews on Amazon) into actionable insights and effective recommendations. In more basic terms, Aspectiva takes unstructured free-form content and tries to quantify it. So how does the “artificial intelligence” piece come into play? It comes in the form of “natural language processing” or NLP which is the ability for an algorithm to understand what humans are trying to say. It’s easy enough to write a software program that can scrape product reviews from a website, and then chuck them into a database, but it’s far more difficult to try and understand what the reviews are actually saying or more importantly, understand customer sentiment across an entire collection of product reviews.
Making sense of product reviews is the first application for Aspectiva’s technology and we have to say it’s quite interesting since we can actually see it live and play with it. There are 3 use cases for this technology:
- If you have a relatively new eCommerce site and therefore have no product reviews, you can implement Aspectiva’s technology with just a few lines of code and suddenly have all kinds of reviews.
- If you are a company with a lot of reviews, you can make sense of them all with Aspectiva’s technology and even use the data to make product recommendations.
- You want to search for products by the way you intend to use them. Aspectiva provides a good example of this by showing how someone looking for a guinea pig house might find that storage cubes work perfectly for that as seen below:
In order to understand how the technology applies to these use cases, let’s take a look at a live usage of the Aspectiva technology for a kitchen mixer seen below:
As you can see, above there are listed various product dimensions or “aspects”. While some like “price” and “color” could be used across all products, others like “bowl” and “mixer” are unique to this particular product. When you drill into an “aspect”, you can then see various adjectives used to describe it. In the above example, you can see that while the mixer is powerful, it is also noisy. We like the graphical user interface (GUI) but would have liked it much more if it showed us how many people used each adjective. For example, we would like the ability to hover above each adjective and then a number would appear showing the number of data points. We also thought it would be nice to be able to click the sample one-liners that were listed so we could read the original review behind them.
The product has some rough edges but what they are doing is very slick. The biggest chip we had on our shoulder before digging into Aspectiva was trying to determine if they are actually using artificial intelligence. Sure, they claim to be using “natural language processing” and “machine learning” but are they really? It certainly looks like are using both. It would be very difficult to aggregate free-form product reviews like that and turn them into structured data without using AI. While there are ways to aggregate free-form content without using artificial intelligence, the results aren’t the same. For example, you could just simply look for commonly used words and then highlight these as a form of navigation. Trip Advisor does this as seen in the below example:
That’s just not the same as what Aspectiva is doing.
If you’d like to play around with their technology, go to Shop.com and take a look at some of their products:
- American Weigh Scales – http://www.shop.com/-
- Food processor –http://www.shop.com/725409685-o+.xhtml
- Air purifier – http://www.shop.com/-
- Elliptical trainer – http://www.shop.com/-
While we played around with the technology live on Shop.com, the Company also has a number of other websites as clients and even claims to be working with Walmart. We’ll be excited to see what forms this technology takes going forward, and in the meantime, we’re going to sit down and eat a big plate of crow for saying that you couldn’t use AI for product recommendations.
Update 02/27/2019: Aspectiva was acquired by Walmart after raising around $4 million in total disclosed funding. Terms of the deal were not disclosed.
Pure-play disruptive tech stocks are not only hard to find, but investing in them is risky business. That's why we created “The Nanalyze Disruptive Tech Portfolio Report,” which lists 20 disruptive tech stocks we love so much we’ve invested in them ourselves. Find out which tech stocks we love, like, and avoid in this special report, now available for all Nanalyze Premium annual subscribers.