Over the last few months, the development team responsible for the documentation hub has been working to deliver a new search experience for technical documentation, while building towards a coherent search strategy that will encompass all of Arm’s web platforms.
Our search strategy is based on the Coveo relevance maturity model, which maps out a journey from siloed (single source) search and federated search (many sources of data), to self-learning, predictive recommendations.
As we commit more content and data from Arm’s web platforms into Coveo, we can benefit from categorized and modeled insights on users’ behavior patterns and apply those patterns across all our websites, bringing improvements such as:
To underpin the search strategy, there needs to be alignment in the data we store with content.
For technical documentation surfacing through the documentation hub, we have started that journey, and will continue building on that standard for all our content.
With alignment in searchable fields between systems and content, we are able to provide a consistent search experience across our web estate, eventually enabling one place to search all of Arm’s content, with use case specific ‘search hubs’ to help people find content scoped to the context of the page they are on.
We have released a new homepage on the documentation hub.
The primary features we are introducing are:
Focusing on search results and data to help users decide which result to choose, the new search interface is customized to fit Arm’s developer and engineer audiences.
We have worked closely with Arm’s user experience and design teams to ensure the approach is minimal, focuses on the provision of relevant data points, and with high information density.
The biggest advantage the new interface brings is providing analytics data into Coveo Cloud on user search behavior and choices, which are used to train machine learning (ML) models in Coveo Cloud.
The introduction of ML models allow us to take advantage of Coveo’s advanced features such as:
In short, automatic relevancy tuning learns what users want, and delivers it.
In more detail, ART analyses user behavior patterns, search visit actions, query formulations and reformulations, and clicked results, to understand which clicked results and content lead to successful outcomes, such as self-service success. It automatically adjusts future search results, so that the best-performing content always rises to the top.
When we go live with these new features, we begin gathering data on user’s search behavior, which in turn will be used to train the Coveo ML models. We expect to see suggestions and recommendations coming through within a day or two, which the developer and content teams are ready to validate and apply to Docs Hub production site.
This allows us to quickly iterate and improve the search experience, taking advantage of the key capabilities and features.
The search filtering options have been expanded in this release, and now allow filtering by audience type, product, and content type. Content type filtering for technical content is a new feature made possible by alignment of metadata in technical documentation content. Currently, the broadest range of content types being published are:
Further content types are supported, such as release notes and different flavors of errata, and will become available as content teams adopt the centralized documentation experience.
The product categories facet is a new, lightweight way of focusing a search on a product and includes its own scoped search to find documentation related to popular Arm products.
These new facets also benefit from the machine learning capability’s Dynamic Navigation Experience, which will reorder the filter groups, and items within a filter group, based on learning from successful user journeys.
Over the next few weeks, we deploy a new navigation for developer.arm.com and community.arm.com, aligned to documentation hub. Looking forwards into the next quarter, there will be further enhancements to search in documentation and the developer.arm.com website, as we align the features and experience between the different web platforms.
If you would like to provide feedback on any of your Arm web experiences, just hit the yellow feedback button to the right of any page.
Visit the documentation hub