#Japan #Tokyo #Uridoki #AI Bidding Search #Price Detection

Uridoki's Innovative Upgrade to AI Abnormal Price Detection

In an era increasingly driven by technology, Uridoki Inc., based in Shinjuku, Tokyo, has set a new standard in the world of C2B purchasing platforms. The company, known for its comprehensive service at Uridoki, announced substantial enhancements to its AI-based abnormal price detection system, the "Uridoki AI Bidding Search." This upgrade promises to reduce the risk of encountering abnormal pricing discrepancies by approximately 73% within just one month.

Background of the Upgrade

Since its inception in May 2025, the initial version of Uridoki's AI system leveraged extensive valuation data to detect and signal irregular valuation prices automatically. It provided real-time notifications to the operational team, allowing for quick evaluations and adjustments. However, as the platform dealt frequently with high-value items such as gold, luxury handbags, and watches—where valuation can fluctuate significantly—there was a pressing need to enhance the detection accuracy further.

Key Improvements

The recent upgrades focused on four main areas:

1. Data Organization and Preparation: Previous valuation data has been systematically organized, considering various factors including product names, categories, conditions, pricing, and valuation timestamps. This structured approach enabled the easier construction of predictive models.

2. Refined Data Classification and Analysis Elements: By refining how data is classified and utilized, particularly by utilizing category-specific statistics and textual data, the inference accuracy has been substantially improved.

3. Concurrent Learning Using Two AI Models: The system now integrates two distinct algorithms to build learning and prediction models. This dual model approach enhances the analysis, allowing for nuanced detection that categorizes findings as "caution" when only one model raises an anomaly flag, and "high anomaly probability" when both algorithms identify an issue.

4. Optimal Threshold Setting: The system can now autonomously adjust thresholds based on an anomaly score, minimizing both false positives and negatives, thus enhancing the overall reliability of the detection system.

These improvements culminate in a considerable leap in the precision of identifying abnormal valuation prices, which Uridoki aims to use for enhancing service reliability and customer satisfaction.

Operational Flow and Strengths of the Detection System

The upgraded "Uridoki AI Bidding Search" employs a multi-step process that includes:

• - Pre-processing Valuation Data: The AI organizes new valuation information and creates analytical factors for efficient data classification.
• - Predictive Modeling: Each of the two models, operating independently, calculates anomaly scores. If a score exceeds a set threshold, that model designates the input as "anomalous."
• - Comprehensive Judgement: The results from the two models are integrated to deliver a final verdict:

- Both models flagging an anomaly indicates high confidence in the abnormality.
- One model signaling caution prompts further inspection.
- Both returning normal indicates appropriate valuation prices.

• - Output Results: The final output includes anomaly determinations, probability values, confidence levels, and an agreement status between the models.

Advantages of the Enhanced System

• - High Accuracy: The double-checking nature of the two models ensures a reliable detection mechanism.
• - Comprehensive Assessment: The analysis accounts for various factors beyond pricing, such as category details and condition.
• - Gradual Assessment Levels: The output categorizes anomaly severity into four gradations, supporting human intervention effectively.
• - Continuous Improvement: Through accumulated learning data, the models become progressively accurate with each use.

Future Directions for AI Development

With the upgrade dramatically enhancing detection precision and operational efficiency, Uridoki is now focusing on developing AI models that will facilitate automatic analysis and classification of user-uploaded product images. This initiative aims to simplify the listing process for sellers and align with the platform's core objective of facilitating effective transactions between sellers and buyers.

Conclusion

Uridoki continues to innovate in the realm of valuation technologies, leveraging AI's capabilities for more accurate, efficient, and user-friendly service experiences. As they advance toward further developments in their AI functionality, Uridoki remains committed to reducing the risks associated with abnormal pricing, fostering a secure environment for their users to engage in transactions confidently. The successful upgrade of the "Uridoki AI Bidding Search" underscores the company’s commitment to excellence in the C2B marketplace.