The knowledge-as-a-service framework and business model is shaping how we think about our position as a source of trusted knowledge in a new era. In a previous post, I shared some things we have learned and guiding principles for our AI/ML product strategy. Today, I will share some of the key components of this strategy to provide more context on how this strategy influences each initiative. Some of these areas are in active development, while other components are on the horizon. But most importantly, all of these areas require a healthy and engaged community where every interaction provides value and encourages contributions back to the corpus of knowledge.
Improving content loops and the knowledge lifecycle
The content creation process should be easy, effective, and enjoyable. We continue to invest in ways that assist our community members with the process of building and evolving knowledge.
- Staging Ground is an area we’ve invested in to improve question success for new askers. Earlier this year we rolled out the core experience—a space where new askers get help from experienced users to improve their questions before posting to the broader community. As a result, we’ve seen question quality improve. Now, we are looking at ways AI/ML can help speed up the process for askers in writing their drafts so they can get answers faster, learn community norms, and take the burden away from human reviewers without removing them from the loop.
- Answers are the other half of the equation in creating quality Q&A artifacts. We’ve been in discovery mode identifying opportunities that encourage knowledge sharing from experts, reduce the number of unanswered questions, and maintain answer quality through community-led curation and verification.
Support for Mods and Curators
This piece of work is all about finding ways to reduce the burden on the most overworked users. AI/ML can perform sentiment analysis to step into toxic comments early, flag content, mark potentially out of date information, assist with content review, and more. There have been some examples of this in the past, such as the unfriendly robot, but the platform primarily relies on manual effort or community-built tools to moderate and curate content. While there is complexity and nuance to the work of moderating highly-technical content, an activity that requires humans in the loop, many tasks are ripe for AI and ML assistance.
Personalization using ML
Most users are looking for answers to help them get unstuck. Some are here to pay it forward and share their knowledge. In either scenario, we want to help get the right content in front of the right user at the right time. A personalized homepage, matching subject matter experts with relevant content, and surfacing out-of-date content are three ways in which personalization can help improve content discovery and the content lifecycle.
These are broad strokes and certainly don’t cover all of the areas that we could work on. But we believe they are necessary components to evolve in the era of knowledge-as-a-service.
Lastly, I’ll close with some updates over the last quarter and what we have planned over the next quarter.
Staging Ground updates – a space for new askers
Since its launch in June, we’ve been focused on user feedback and iterating to improve the user experience and efficacy. Reviewer recognition and engagement has been a focus to ensure reviewers feel appreciated for their effort in helping new askers. A new stats module was released showcasing reviewer impact. Coming up next are new Staging Ground badges and exploring what it would mean to grant reputation to reviewers for their contributions.
For askers, we are evaluating whether a question assistant can be effective to improve their drafts. This would allow askers to improve question drafts earlier in the process, while saving time for reviewers.
User activation – improving new user success
With the intent of introducing newer users to the platform to help them accomplish their goals, we ran a series of experiments to understand how tags are used today and improve the discoverability and usage of them. Next up is a test that will use tag preferences to customize the homepage experience.
Community data protection – improving data protection and accessibility.
As part of data protection and accessibility, we brought hosting of the quarterly data dump in-house. Stack Exchange has a long history of publishing Q&A data which has been used for a variety of applications such as building community tools and academic research. The data dump for each network site is now accessible for community members through the user profile.
Community asks sprint – giving back to the community
Our community-focused product teams completed the second quarterly sprint where we focus on community requests, quality of life improvements, and long standing bugs. Now that we’ve completed a couple of cycles, we will be doing a cross-team retrospective to gauge what has gone well and what can be improved going forward.
As always, we’re excited to share findings as we go and value your input along the way. If you’d like to provide your input, we invite you to join the conversation on meta.stackexchange.com, meta.stackoverflow.com, or opt into our user research list.