Backend Development 10 min read

Design and Practice of the tanGo Search Presentation Framework

The article presents Baidu’s Aladdin vertical search product and introduces the tanGo framework, which abstracts search pipelines into resources, cards, and scenes, enabling configuration‑driven, graph‑based resource scheduling for single results, demand clusters, and groups, while measuring scale, efficiency, and user satisfaction.

Baidu Geek Talk
Baidu Geek Talk
Baidu Geek Talk
Design and Practice of the tanGo Search Presentation Framework

This article introduces Baidu Search’s vertical product “Aladdin” and explains how scene‑oriented search results are organized into single results, demand clusters, and demand groups. It then presents the tanGo business framework, which was created to support these complex, immersive search experiences.

Requirement Analysis – The article classifies search outputs into four levels: single result, demand cluster (multiple single results for the same query), demand group (a collection of clusters for large‑scale events such as the Gaokao or the Olympics), and the overall result page that may contain several groups.

Business Abstraction – tanGo abstracts the search pipeline into three core entities: resources (retrieved summaries), cards (the smallest display unit that assembles resources into front‑end templates), and scenes (logic that identifies sub‑scenes from a query and drives the retrieval of appropriate demand groups).

Framework Design – The design emphasizes protocol conversion (HTTP, nshead, protobuf/JSON), configuration‑driven processing, componentization, and graph‑based resource scheduling. The framework is layered into request‑level, card‑level, and resource‑level processing components, each configurable and reusable.

Configuration‑Driven Design – All retrieval steps can be described in a single configuration file, reducing learning cost and operational overhead. The configuration supports a three‑layer topology (strategy → card → resource) and enables hot‑loading for dynamic updates.

Resource Scheduling – A lightweight DAG execution engine is introduced to orchestrate parallel and sequential tasks, handle exceptions and timeouts, and provide a simple syntax for workflow definition.

Metrics and Evaluation – The framework measures scale (application coverage), efficiency (project creation cost, delivery cycle, code‑line savings), and user satisfaction (NPS). These metrics guide continuous improvement.

Conclusion and Outlook – The article summarizes the tanGo framework’s current capabilities and outlines future work, including tighter integration with code‑hosting services, standardized build pipelines, and CI/CD support to cover the entire development lifecycle.

backendarchitectureDAGConfigurationframeworkSearchtanGo
Baidu Geek Talk
Written by

Baidu Geek Talk

Follow us to discover more Baidu tech insights.

0 followers
Reader feedback

How this landed with the community

login Sign in to like

Rate this article

Was this worth your time?

Sign in to rate
Discussion

0 Comments

Thoughtful readers leave field notes, pushback, and hard-won operational detail here.