An Intelligent AI-Powered Meeting Analytics Extension with Automated Calendar Synchronization, Productivity Scoring, Visual Insight Modeling, and Custom Task Export Framework
Srimathi R , Monica K , Rithanya S, Vidhurshavarshini S
Computer Science and Business System, Rajalakshmi Institute of Technology
Computer Science and Business System, Rajalakshmi Institute of Technology
Computer Science and Business System, Rajalakshmi Institute of Technology
Computer Science and Business System, Rajalakshmi Institute of TechnologyAbstract - This paper presents a production-oriented intelligent meeting analytics extension that transforms conversational data into structured, actionable outputs within a unified Chrome-based system. Unlike conventional meeting assistants limited to transcription, the proposed approach introduces a semantic execution layer that extracts action items, aligns them with deadlines, and synchronizes them with calendar platforms. The methodology integrates streaming speech understanding, role-aware dialogue segmentation, transformer-based hybrid extraction with constrained decoding, and temporal conflict optimization. Additionally, a dynamic productivity scoring model evaluates engagement quality, task follow-through, and discussion efficiency. A visualization module enables longitudinal analysis of organizational patterns such as speaker contribution imbalance and workload drift. The system also supports interoperable task export across multiple platforms using a canonical task ontology. Experimental evaluation demonstrates strong performance, achieving a macro F1 score of 0.89 for task extraction, 0.93 accuracy in intent-to-calendar mapping, low latency (p95 of 1.8 seconds), and high correlation (r = 0.81) with human productivity assessments. These results highlight the effectiveness of treating meeting intelligence as a closed-loop productivity system.
Key Words: meeting analytics, action item extraction, productivity scoring, calendar synchronization, conversational AI, workflow interoperability
