Altmetric Attention Digest

The Altmetric Attention Digest is an AI-powered feature designed to move beyond simple attention metrics and provide users with a clear, concise, and narrative summary of a research output's attention data.

It uses a powerful Large Language Model (Gemini 2.5 Flash) to synthesize information from various attention sources—including news, policy documents, social media, and citations—to create an on-demand, high-quality and easily interpretable report on the research's influence, while also enabling more advanced prompt refinement and deeper analytical outputs.

Attention Digests will be available for research outputs with a minimum of 5 mentions (excluding mentions from discontinued sources).

The Digest addresses the challenge of making sense of complex attention data, enabling users to:

• Quickly Grasp Impact: Instantly understand the significance of a research output's attention, saving time on manual analysis.

• Identify Key Events: Highlight specific connected events or trigger points that led to notable surges in attention (e.g., a policy change or a press release).

• Focus on Authority: Prioritize and synthesize information from key indicators of impact, such as policy documents, clinical guidelines, and prominent news media.

• Facilitate Communication: Generate ready-to-share narratives for stakeholders, making it easy to convey the reach and influence of research for grant reports, tenure reviews, or internal strategy briefings.

• Streamline Reporting: Provide a comprehensive, high-level summary that connects the quantitative metrics with the qualitative context.

• Incorporate Sentiment Context: Integrate insights from Altmetric’s Sentiment Analysis to provide additional context on how research is being discussed, complementing the narrative with signals about tone and perception.

• Improve Consistency Over Time: Highlight both peak activity periods and notable mentions from quieter periods, including viral or high-impact posts, to ensure a more balanced and representative view of how attention evolves over time.

The Digest is distinct from Sentiment Analysis. While Sentiment Analysis categorizes the tone (positive/negative), the Digest provides a narrative overview of what is being said, where, and when, focusing on building a coherent story of impact.

The Digest is an on-demand feature accessible from the Details Page of any research output.

1. Trigger: On the Details Page, Altmetric Explorer users will see a clear "Show Attention Digest" Call-to-Action (CTA) button, typically located near the main attention score breakdown.

2. On-Demand Process: When the CTA is clicked, the system initiates the process to generate the Digest. This process is resource-intensive and is limited to once per 24 hours per research output. If a Digest was generated within the last 24 hours by any user, subsequent users will be shown the cached version.

   • The system will only create a brand new Digest if the underlying attention data (eligible mentions) for the research output has changed since the last Digest was generated. This means you may see a Digest with a timestamp older than 24 hours if no new attention data has been recorded, and this Digest will be the most up-to-date summary of the existing data.

3. Display: The generated Digest appears in a pop-out drawer overlay on the Details Page.

The generated Digest follows a clear, hierarchical structure:

1. High-Level Narrative Summary: This initial paragraph provides a concise, overall summary of the research's impact, synthesizing the most notable events and attention highlights.

2. Source-Specific Insights: This section provides more detailed information, breaking down the impact by source type (e.g., News outlets, Policy documents). It includes:

   • Synthesized findings regarding attention from that source.

   • Specific examples or links back to notable mentions to enable verification.

   • A structured approach to identifying and prioritising “notable mentions”, based on factors such as reach, engagement, source authority, and contextual relevance, ensuring the most impactful examples are surfaced.

   • Visibility of mentions across different time periods, including both spikes in attention and significant activity during quieter periods, to provide a more comprehensive and temporally balanced analysis.

3. Transparency and Footprint: The Digest is designed to incorporate a robust footprint where notable mentions have clear, accessible links back to their original sources. This enables users to verify the AI-generated information independently.

When generating a Digest, you may see an option to provide feedback. This feedback helps us refine the AI model for more accurate content.