How Figure Eight Helps eBay Improve Product Categorization with Machine Learning

September 12, 2019 0 comments

Figure Eight is an essential Human-in-the-Loop machine learning platform. Its platform transforms text, image, audio, and video data into customized high-quality training data for a wide range of use cases, including autonomous vehicles, consumer product identification, natural language processing, search relevance, intelligent chatbots, and more.

 

Background of the Company

Founded in 1995, eBay Inc. is an American multinational global technology corporation. The company manages eBay.com, an online marketplace that connects millions of buyers and sellers worldwide facilitating consumer-to-consumer and business-to-consumer sales. eBay is a multibillion-dollar business with operations in different countries.

 

Business Challenge: Cost-Effective Use of Human Brainpower

eBay presents millions of product that originate from users across the globe. Therefore, it becomes difficult to ensure a constant change in the online catalog and its optimal management using computer algorithms alone. At a certain point, human brainpower becomes essential for accuracy in the categorization of products and the creation of product taxonomies from the user’s standpoint.

Additionally, the company also requires an efficient method of finding a product’s Global Trade Item Number (GTIN) which is a unique 12- to 14-digit identifier often missing from posted product descriptions. The task of finding it is necessary yet complicated by the lack of a central repository of GTINs.

Although eBay tried to deploy offshore teams at various low-cost destinations to finish such processes, the conventional outsourcing model introduced several challenges with respect to cost, scalability, and accuracy.

 

Solution: Micro-tasking Along with Automated Quality Control

To curb the issues of eBay, Figure Eight’s platform took huge amounts of former’s product information and broke it down into small packages (micro-tasks). These tasks were completed online by thousands of individuals collaborating all around the world. Such collaborative effort at massive scale ensured the huge data task was accurately done in record time and at a reduced cost.

eBay availed the benefits of Figure Eight’s workflow management and accuracy check of the data returned to enhance its crowdsourcing process. Recently, the e-commerce company has completed more than 15 different types of high-value projects through Figure Eight.

Both the companies collaborated to design an ML workflow in order to improve product categorization algorithm. The collaboration was able to achieve job faster, at a lower cost and best-possible accuracy through crowdsourcing.

Additionally, Figure Eight got the opportunity to create a geographic filter to ensure native English speakers who make up eBay’s core customer base and hence are more familiar with eBay product taxonomy completed the project.

The collaborative effort of eBay and Figure Eight also developed an exhaustive search process for product GTINs. The latter’s contributors searched for GTINs for a specific product through a variety of reliable channels and then the company compared the contributors’ responses, ensuring that each piece of data was verified and accurate.

 

Results: Accuracy of An Iterative Process

1. The large online contributor pool of the provider company completed product categorization 5X faster than a conventional outsourcing team with great accuracy.
2. GTIN collection was performed 10X faster with best-possible accuracy.
3. Additionally, Figure Eight increased the number of judgments collected on products that are difficult to categorize or for which it is difficult to find GTINs. This implies that, in case of disagreement, the company collected judgments until a reliable answer was found.
4. Figure Eight facilitated eBay with the quality and accuracy of an in-house team that too at a lower cost and in a highly scalable manner.

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.

Your data will be safe!Your e-mail address will not be published. Also other data will not be shared with third person.