DataWeave, a Competitive Intelligence as a Service provider for retailers and consumer brands, announced recently the launch of its Counterfeit Products Detection solution, which enables consumer brands to discover and curb the presence of counterfeit products on e-commerce websites using AI-powered image and text analytics.
With open marketplace models increasing in commonality, so too has counterfeiting on these sites.
Karthik Bettadapura, Co-founder and CEO of DataWeave, discussed the Counterfeit Solution in detail for our uninformed subscribers.
“As more and more brands sell over the online channel today, they feel the need to monitor and manage their presence on eCommerce websites, as a large portion of their customer base now interacts with the brand almost exclusively on these websites. Most leading online retailers like Amazon and eBay have adopted an open marketplace model, which means that third-party or unknown merchants too list a brand’s products and fulfil all aspects of order processing and delivery. Unfortunately, several of these online merchants list and sell counterfeit products masquerading as originals on open marketplaces. This adversely affects the original manufacturer’s brand value and the consumers’ trust on the brand.
While listing and selling fake products is the most common form of counterfeiting, there are other forms as well. Unauthorized white-labeling occurs in some product verticals, which is when the original brand’s name of a product is replaced with another when listed online, while all other catalog elements, such as the product description and images remain the same. Sometimes, online merchants even steal images from original product catalogs and include them in other product listings, which misleads shoppers during the purchase process.
Such acts of counterfeiting are quite prevalent mainly due to the ease of replicating an online product catalog on eCommerce websites. Our Counterfeit Products Detection solution is built using our powerful, AI-based image and text analytics platform, which detects possible counterfeits at a high level of accuracy by capturing minute discrepancies and differences in the catalog images and text, categorizing the merchant as authorized or unauthorized, and analyzing product reviews. This way, brands can use our technology to work with online retailers to remove counterfeit products and blacklist the defaulting merchants.
We offer this solution as part of our Brand Governance product suite, which in addition to helping detect counterfeit products, enables brands to monitor and minimize minimum advertised price (MAP) violations and unauthorized merchants, as well as optimize the brand’s representation on online catalogs.”
Throughout the years, it has been common knowledge to be careful on the websites with open marketplace features. Will DataWeave’s AI-powered new software eliminate the counterfeit merchants from the real ones so that humans do not have to filter out content? Not so fast, says Bettadapura.
“AI has progressed rapidly in the space of Computer Vision, which is typically used in retail for object identification, finding similar products, tagging product attributes, etc. However, solving these problems today requires labeled training data (the machine initially needs to be taught a few hundreds or thousands of times that a T-shirt has a round neck before it "understands"). So far, deep learning has been successful primarily for such supervised learning tasks. Now, there is great potential in unsupervised representation learning, which does not require labeled training data sets. While some interesting work has been done in this space already, such as developing autoencoders and Deep Convolutional Generative Adversarial Networks (DCGANs), this is an active and exciting area of research for AI technologists to branch further into.
Other areas of application are even more complex. While there has been a lot of progress in text classification, summarization, and even translation, we have a long way to go before machines can come close to human-like behavior. For instance, detecting sarcasm in a product review by a shopper is a very difficult problem to solve. To be fair, many humans too do not recognize sarcasm, and so teaching it to machines is a unique challenge!”
While DataWeave’s new Counterfeit system is a great starting point in terms of eliminating fraudulent merchants, AI still has a long way to go before it can detect the nuances that are associated with humans.
The next few years will surely be interesting to see where brands and vendors choose to go in terms of adopting to AI or trying to stay away.