AI‑Powered Credit Scoring Fuels the Shared Wardrobe Economy: The Sesame Credit Case

Recently, CCTV‑9 broadcast a self‑produced documentary series “The Big Data Era”, the first domestic documentary on the big‑data industry, which vividly explains how big‑data technology impacts government governance, public services, data security, industrial transformation, and future life. In episode 4, it tells the story of how Sesame Credit helps the sharing economy and promotes the construction of a commercial credit system. We excerpt it here and briefly introduce the technology behind Sesame Credit.

Cao Xueying is a fashion model whose work requires her to wear different outfits in various occasions, leading to high clothing expenses. Many women feel they lack a piece of clothing; Cao feels the same.

By chance, Cao discovered a shared‑wardrobe consumption model, costing less than 500 yuan per month for 30 outfits, including many affordable‑luxury brands, most without deposit. Over a year, she saved a lot on clothing purchases, thanks to the credit system.

Sesame Credit’s technical director Mao Renxin is one of the builders of this credit evaluation system. He continuously uses algorithms and model innovation to create a non‑financial commercial credit evaluation system.

“We actually have many innovations in technology; through this credit evaluation system we compute user data across dimensions and output a user’s compliance portrait,” Mao said.

The commercial credit evaluation system scores a person based on identity features, credit history, fulfillment ability, etc., using machine learning, time‑series modeling, deep learning, and network analysis to accurately assess user compliance in various commercial scenarios. The scoring rules must be objective and fair, prompting Mao’s team to constantly improve prediction precision, enabling merchants to expand business while controlling risk.

As the credit platform improves, more netizens join the Chinese shopping festival, creating a new industry chain. Workers transport recycled clothes to cleaning workshops, sort, wash, high‑temperature iron, ozone disinfect, seal, and store them in a large clothing warehouse.

When the shared‑wardrobe system receives a consumer order, staff select the appropriate style from hundreds of thousands of garments in the warehouse; the computer system refines sorting on the production line, packs, and ships the items.

With this credit evaluation system, the shared‑wardrobe model gains popularity among young people; as the sharing economy spreads, more users can showcase compliance, accumulate credit, and the system continues to be optimized, stimulating market vitality.

The credit assessment involves machine‑learning methods from shallow to deep, using a unified modeling framework for linear, temporal, and structural data behind the Sesame Score’s DeepCredit technology. Traditional scorecard techniques provide interpretable components but rely heavily on feature engineering. Recent deep‑learning advances have boosted time‑series mining, graph analysis, NLP, and image recognition, enabling more accurate credit portraits.

Through research on time‑series mining, Sesame Credit explores deep learning in credit assessment, modeling user compliance history over time. Traditional financial compliance mining depends on developer experience and limited features, failing to meet commercial performance. In modeling, recurrent neural networks (RNN) and variants handle sequence data well; the Sesame Score team first applied multi‑layer RNNs to billions of users, with architecture shown below. By incorporating Stacking, Embedding, Wide&Deep optimizations, the algorithm learns compliance habits such as punctuality and frequent adherence, achieving over 40 % performance improvement, bringing more qualified users and lower capital costs.

Currently, Sesame Credit is deeply integrated into various Alipay products and offers services to partners. It will invest more to optimize technology and accelerate the commercial credit system.

Click the lower‑left “Read Original” to visit Ant Financial’s official website for more information.

- END -