This Is How Alibaba Is Using Big Data to Fight Fakes

Alibaba has long been haunted by a controversial reputation as the go-to marketplace for counterfeits and fakes. However, they are now under increasing pressure to clean up.

In their latest anti-counterfeiting initiative, the China’s e-commerce behemoth is drawing upon the big data technologies in the monitoring, tracking, and detection of counterfeit goods and manufacturers offline.

Operation Cloud Sword (云剑行动), led by Zhejiang law enforcement, used information provided by Alibaba to stop 417 counterfeit production groups, including 332 suspects and counterfeit good valued at 1.43 billion RMB (205 million USD).

As a continuation of the initiative, the operation will be further extended to Shanghai, Anhui, Jiangxi and Jiangsu, the regions where China’s counterfeiting goods and manufacturing ran rampant, to form the “Cloud Sword Alliance.”

alibaba-anti-cf
Counterfeit Facilities Location (source: Alibaba)
How Big Data is Applied in Anti-Counterfeiting

The figures sure appear impressive. But for average customers and tech fans, we may easily wonder how the technology is actually applied. Here’s some insight.

Real Time Scanning and Detection Models Powered by Big Data  

Alibaba’s AI bot continuously scans entire platforms such as Taobao and 1688 (an online market for wholesalers) to detect counterfeit goods. With its 600 data analytics models, the bot analyzes in real-time merchants and product listings using hundreds of millions of data points such as product specification, customer reviews, and user reviews. All the new listings that enter the system each day have to go through the bot’s scanning system, taking about 30 milliseconds from start to finish for each product listing.

The model also analyses relational data associated with the user behavior, merchandise, payment, and logistics (receiver/return addresses) to detect anomalies, determine suspected counterfeit goods, and high-risk merchants for timely interception and disposition.

Optical Character Recognition

OCR (optical character recognition) technology processes text analysis and detects textual information on product images. For example, some watch counterfeiters may put RMB 100,000 in the price field, but they put something like price ranging from RMB 1,000 to 7,800 on the product image which tells it is potentially a fake product, according to information provided by Alibaba.

“For example, some watch counterfeiters may put RMB 100,000 in the price field, but they put something like price ranging from RMB 1,000 to 7,800 on the product image which tells it is potentially a fake product,” said the company.

Textual analysis capabilities are used at a higher-level to analyze syntax and semantics rather than only compare keywords. The image recognition algorithm enables the company to identify the information related to counterfeit goods-related, in particular, irregularities with brand logos and trademarks.

Machine Learning

Alibaba’s detection technology improves itself constantly through self-learning. In addition, all the data gleaned from offline investigations will be adopted by the system to enhance its counterfeit detection and tracking capabilities, creating a virtuous cycle.

Anti-Counterfeiting: A More Pressing Issue For Alibaba Than Ever

While Alibaba is tuning up its global expansion strategy after the record-breaking IPO, anti-counterfeiting is becoming the most pressing issue that affects the confidence and trust of Alibaba’s customers and investors. Chairman Jack Ma has set a goal of getting over half the company’s revenue from overseas. But there’s still a long way for the company to go: that the figure was just 4% when Ma set this goal in 2015.

In addition to expanding businesses, Alibaba is also struggling to rebuild its image as a trusted and responsible international company. Shortly after becoming a member of The International Anticounterfeiting Coalition under the “intermediary” category, Alibaba’s membership was suspended amid a backlash from brands that include Tiffany, Michael Kors, and Gucci. Similarly, the American Apparel & Footwear Association called to re-list Alibaba as a notorious market.

To solve the problem, the company founded the Platform Governance Department in 2015 to fight against fakes and IP infringement. With an annual investment of nearly 1 billion RMB and the joint effort of 2,000 staff, Alibaba’s big data system was able to remove 120 million suspicious items from the platform, according to data released by the company in 2015.

At the same time, the company is firing at other disingenuous practices that used to be prevalent on the platform such as click farming.

“With our big data analytics and technology, we have the ability and we have the will to track down counterfeits once they are detected online, We won’t stop until we bring them to justice with the help of the authorities,” says Jessie Zheng, Chief Platform Governance Officer of Alibaba Group.

Anti-counterfeiting & OEMs

Despite the impressive figures of anti-counterfeiting endeavors, the scale of fakes is still enormous in China.

Earlier this year, at the Alibaba investor day, Ma told that audience that “[f]ake products are of better quality and better price than the real names. They are exactly the same factories, exactly the same raw materials but they do not use the names.”

Later, in a signed article for the Wall Street Journal, he pointed out several reasons why counterfeit goods are so ubiquitous, shedding some light on how to solve the problem. Small Chinese businesses, which used to serve as OEMs for overseas partners, are finding it hard to survive when export demand from Western markets is shrinking. The rise of Internet, with its lower communication and distribution barriers, has become the most accessible channel for the OEMs to sell their extra product, according to Ma.

Upgrading the whole industry perhaps is one way to fix the problem, but this would be long-term crusade and there’s no quick fix. For Alibaba, anti-counterfeiting will be their top priority for now, according to Ma.

Image Credit: Shutterstock