A General Solution for Improving Daily Business Issue Diagnosis Efficiency

In daily development, various feedbacks from users, partners, operations, product, and testing are often forwarded through multiple people before developers finally handle them. The main pain points are the many hand‑offs and long investigation chains.

The proposed solution aims to boost the efficiency of everyday business problem diagnosis and has been applied in the Xianyu platform.

Key Features :

One‑click retrieval of all related business data : Input different dimensions of parameters and obtain consistent results.

Business data is easy to understand and usable by anyone : Explain data attributes for non‑technical users.

Business data can be diagnosed : When data is abnormal, the system reports the anomalous information.

Overall Approach

The problem is divided into two categories: (1) abnormal cases where business data is inconsistent, and (2) normal queries where data is needed for inspection.

Three main steps are implemented:

Provide a generic data aggregation query that calls multiple system services to gather all related data.

Perform semantic conversion of business data, adding human‑readable explanations and virtual attributes.

Abstract business rules to detect anomalies using a rule engine.

Data Panorama

GraphQL is chosen as the aggregation layer because it can invoke various backend services (order service, valuation service, payment service, etc.) with a single query, reducing integration cost. An example GraphQL query for the Xianyu recycling business aggregates data from six systems and a distributed cache.

Business Semantic Conversion

Data returned by GraphQL is re‑grouped and annotated: fields are renamed, enumerated status values are translated, and virtual attributes (e.g., “high‑price order”) are computed. JSONPath is used to unify heterogeneous data structures.

if(actualValue != expectedValue){ print abnormalResult }

The rule engine QLExpress executes business rules expressed in a Java‑like script. Example rule:

if(order.isCreditOrder == "1") return order.idleCreditPayAmount > 0; else return true;

All rules are evaluated, and their results are merged with the aggregated data to produce a diagnostic view.

Execution Flow

The entire process—from data aggregation, semantic conversion, rule execution, to result presentation—is configurable and can be quickly adopted by other business lines.

Future Extensions

Planned improvements include integration with log‑search tools, automated handling of abnormal data, sensitive data protection for customer‑facing scenarios, and real‑time monitoring and alerting.