LinkedIn has shared information and details on the origins of LinkedIn Post Search, a tab that heavily dictates content consumption on the platform, and enables relevant content discovery for users.
LinkedIn Post search saw organic growth of a 35% year-over-year increase in user engagement in 2020. The insights on the development would help businesses and users on the platform to learn how the LinkedIn Search engine works and how these learnings can help provide a better understanding of this functionality on the platform.
LinkedIn Post Search intends to serve results that are relevant to a member’s query, from a wide variety of verticals such as jobs they may be interested in, people they may want to connect with, or posts that are trending within their industry.
The search midtier at LinkedIn follows the federated search pattern, it fans out calls to various search backends and blends the results together. The Post search stack, however, is different, as it was designed to be compatible with the Feed ecosystem at LinkedIn. For Post search, the fanout and blending logic depended on Feed services, including feed-mixer and interest-discovery.
Apart from the different service architectures, Post search also uses an intermediate language called Interest Query Language (IQL) to translate a user query into index-specific queries used to serve search results.
Due to the complex architecture, increasing the development and experimentation velocity proved to be difficult, as any search relevance or feature improvements required changes to multiple points throughout the stack. It was challenging to address many of the unique needs of Post search, such as balancing multiple aspects of relevance, ensuring diversity of results, and supporting other product requirements.
Simplifying system architecture
The platform set out to simplify the system architecture to improve productivity and facilitate faster relevance iterations. To achieve this, they decided to decouple the Feed and Post search services in two phases. The first phase removed the feed-mixer from the call stack and moved fanout and blending into the search federator. The second phase removed interest-discovery. This enabled getting rid of all the cruft built up over the years and simplified the stack by removing additional layers of data manipulation.
Improving Post search relevance
As LinkedIn thought about ways to improve the relevance of results from Post search, they realized that the user’s perceived relevance of results is a delicate balance of several orthogonal aspects, such as:
- Query-document relevance
- Query-agnostic document quality (i.e., static rank)
- Personalization and user intent understanding
- Post engagement
- Post freshness/recency (especially for trending topics)
In addition to those aspects, the platform wanted to easily implement other requirements from their product partners to satisfy searcher expectations (e.g., ensuring diversity of results, promoting serendipitous discovery, etc.). To meet these goals for post relevance, they implemented a new ML-powered system; the high-level architecture is shown in Figure.
As a single, unified model did not scale well for the platform's needs, they invested in modelling the First Pass Ranker (FPR) as a multi-aspect model, wherein each aspect is optimized through an independent ML model. Combining the scores from all these aspect models in a separate layer to determine the final score for ranking. This approach enables them to:
- Have separation of concerns for each aspect
- Decouple modelling iterations for each aspect
- Add more explainability to our ranking
- Control the importance of each aspect based on product requirements
To iterate quickly on a multi-layered, complex ML stack, testing and validation was a foundational piece. They built a suite of internal tools to assess the quality of new candidate models and quantify how they differed from the current production model. This enabled them to have a principled approach to testing relevance changes and ensured they did not regress on the core functionality/user experience.
- Pre-ramp evaluation: This tool helps compute and plot descriptive statistics for the results of a collection of user queries to test and validate intended effects. This helps LinkedIn to understand its models better before ramping any change to members and to catch any unintended side effects early on.
- Validating table stakes: The Build Verification Tool (BVT) helps generate model reliability tests to assert if a specific expected document is correctly surfaced for certain queries and members.
- Human evaluation: To have a better understanding of basic query-document relevance, the platform invested heavily in crowdsourcing human ratings for the search results. Leveraging this crowdsourced data to evaluate the performance of offline ranking models with respect to search quality, to ensure it meets the basic quality bar.
These changes to the system architecture have helped us unlock several wins, such as:
- Developer productivity: The platform reduced the developer's effort to add new features from 6 weeks to 2 weeks. Additional productivity wins include faster developer onboarding time and lower maintenance costs.
- Leverage: Because search federation was already integrated with newer machine learning technologies, this migration has also empowered relevant engineers to iterate faster and run more experiments to improve the Post search results.
- End-to-end optimization: With the extra layers removed, the search federation now has access to all the Post-related metadata from the index. This data is being used to improve the ranking of posts when blended with other types of results, help reduce duplication, and increase the diversity of content.
- Engaging and personalized results: Pertinent results, which are highly relevant to the user’s search query, have led to an aggregate click-through rate improvement of over 10%. Increased distribution of posts from within the searcher’s social network, their geographic location, and their preferred language has led to a 20% increase in messaging within the searcher’s network.
- Superior navigational capabilities: Better navigation allows members to search for posts from a specific author, for posts that match quoted queries, for recently viewed posts, and more, leading to an increase in user satisfaction, reflected by a 20% increase in positive feedback rate.
Limitations and future work
- Further simplify: Removing IQL from the stack will help remove two layers of query translation from the flow, making it much easier to add new, useful filters for Post search. To do this, they plan to merge all of the Post search backends and translate directly from the member query to the Galene query.
- Semantic understanding: they will be investing in Natural Language Processing (NLP) capabilities to understand the deeper semantic meaning of queries.
- Detect trending content: To quickly detect trending, newsy, or viral content on our platform and surface fresh results for queries on trending topics, they plan to use a real-time platform for computing engagement features to reduce the feedback loop from hours to minutes.
- Promoting creators: Results are ranked today mainly by using viewer-side utility functions such as the likelihood of a click, user action originating from search, etc. To support the creators, they will evolve this ranking, along with their experimentation and testing stack to also optimize for creator-side utilities, such as content creation or distribution for emerging creators.
- Multimedia understanding: Expanding the document understanding capabilities beyond text data to include handling multimedia content such as images, short-form videos, and audio, is another opportunity for future investment. As web content becomes increasingly diverse and newer content formats become popular, it will be important to ensure these content types are easily discoverable on Post search.
In large complex systems, the existing state can be suboptimal due to incremental solutions to problems over time. By stepping back and taking a high-level view, it is possible to identify several areas of improvement. It takes an open mind and support from engineering, product, and leadership to entertain these improvements.