Glossary
This glossary defines key terms used across MiaRec Conversation Analytics documentation.
AI Task
A configurable unit of analysis that applies AI to a conversation’s text and writes results into one or more Custom Fields. An AI Task typically includes a prompt, output mapping, and optional filters.
Attribute (Output Attribute)
A named output produced by an AI Task (often a JSON key). Attributes are mapped to Custom Fields so results can be stored and used in dashboards and search.
Conversation
A single customer interaction record analyzed by MiaRec. Depending on deployment, this can be a voice call, chat, email thread, or ticket thread.
Conversation Analytics
MiaRec capabilities that turn conversations into structured insights for dashboards, search, and review.
Custom Field
A typed field used to store structured data about a conversation (number, date, dropdown, text). AI insights are stored in Custom Fields so they can be filtered, aggregated, and reported.
Dashboard
A visual report built from conversation data and insight fields (trends, averages, distributions, drilldowns).
Drilldown
The workflow of clicking a dashboard element (e.g., a bucket) to view the underlying conversations that make up that metric segment.
Explanation
A short, human-readable rationale provided alongside an extracted value to help reviewers understand why the AI produced that result.
Filter (Task Filter)
Eligibility criteria that determine which conversations an AI Task should run on (e.g., direction, duration, channel, team).
Insight
An AI-powered result derived from a conversation (e.g., summary, sentiment, topic, CSAT score). Insights are typically stored in Custom Fields.
Metric
A numeric or categorical measurement tracked over time (e.g., CSAT, churn risk category). Metrics often drive dashboards.
Named Entity
A specific item identified in a conversation, such as dates, times, names, monetary amounts, product names, or company names. Named Entity Recognition (NER) is an AI task type that extracts these items for storage and reporting.
Prompt
Instructions provided to the AI model that define what to extract and how to format the output.
Scorecard (Auto QA)
A structured rubric used for automated QA, typically organized into sections and questions with scoring rules.
Transcript
Text produced from a voice call recording. The transcript is the primary input for AI analysis on calls.
Thread
The message history for text-based conversations (chat/email/tickets) when enabled. Threads can be analyzed similarly to transcripts.
Threshold
A rule that maps numeric values into labeled buckets (often with colors) for easier interpretation and dashboarding (e.g., CSAT 1–5 into satisfaction categories).
Topics
A multi-label classification of conversation subjects. Topics help categorize what conversations are about (e.g., "Billing", "Technical Support", "Account Changes") and are useful for identifying trends and routing patterns.