In implementing the technology, Shopee believes that it is important to be considerate of your business’ needs
Anisa Menur A. Maulani , 03 Apr, 2019
For More Information : e27.co
Data science has become one of the hottest trends in the Southeast Asian tech community nowadays, that many businesses have recklessly jumped into the bandwagon for fear that they might miss out on the opportunities it can provide.
Dr. Pan Yaozhang, Head of Data Science at Southeast Asian e-commerce giant Shopee, reminds us of the need to be considerate in implementing the technology.
“The use of technology needs to meet the needs of your business,” she says.
“Not every area might requires the use of data science in the beginning. Even with Shopee, we didn’t have that since the very beginning. We started one year after the launch of our app, so that we have enough data to work with,” the scientist continues.
In principal, for data science to be able to drive an e-commerce business, Dr. Pan stresses on the importance of combining deep business understanding with the technology itself. Businesses need to have an understanding of where they are as a company and the strategy that they intend to use at that particular stage.
That way, the use of data science will be aligned with the company’s strategy in driving key business metrics.
Dr. Pan is leading the data science team at Shopee, where she had built the team from scratch and has grown it to a more than 50 person strong team. She joined the company in 2015
Before Shopee, she worked for Grab, Greenwave Systems, and was a research scientist at the Institute for Infocomm Research, A*STAR.
In an Academy interview with e27, Dr. Pan shares how the company is implementing the use data science in order to answer some of the greatest challenges in running an e-commerce business in Southeast Asia.
What they do
She begins by giving examples of the three use cases of data science in the company:
- To optimise key business processes, such as detecting bad listings or potential fraudsters
- To reduce costs, such as by using in-house translation engine to translate product information
- To improve customer experience and optimise services, such as giving personalised recommendation of products
Apart from these examples, Shopee is also currently working on their long-term data-focussed goal: Developing an E-Commerce Knowledge Graph.
For context, the Knowledge Graph is Google’s systematic way of putting facts, people and places together, to create interconnected search results that are more accurate and relevant. Using Natural Language Processing technology, the Knowledge Graph will encode relationships based on taxonomy and characterise the nature of each linkage, responding to queries even before if answers were not explicitly discussed beforehand.
Started by Google and popularised in China, the Knowledge Graph is described as being “quite massive” in scale.
Shopee’s implementation of the technology aims to better understand changing consumer habits and patterns in Southeast Asia.
Dr. Pan says that the company is currently in the “very initial stage” of implementing this technology.
“Usually, building up a knowledge graph includes several stages. The first stage is the knowledge attraction, the second stage is the knowledge storage, and the third stage is the knowledge processing and calculation. We are currently at the stage of knowledge attraction, so we use all tangible technologies including NLP, predictive analysis, and machine learning model to try to extract the most valuable information from the abstract data,” Dr. Pan explains, adding the abstract data include product images and information.
“The purpose is to attract the key information from the abstract data and to store the information in a structured way which is the graph database,” she concludes.
Once completed, the Knowledge Graph will enable the Shopee platform to run faster and further with greater efficiency, benefiting the whole ecosystem – buyers, sellers and the business.
The company says that it will be one of the first e-commerce platforms to launch the system in Southeast Asia and Taiwan.
The process would be even more complex compared to its implementation in US and Chinese markets, as e-commerce platforms in Southeast Asia are dealing with unique challenges, such as the use of multiple languages even within a single market.