Personal Identification

The “Personal_Identification” tracker detects moments in a conversation where a participant identifies a person—this could be stating a name, a role (like “project manager” or “team lead”), or any reference that personally pinpoints an individual. Understanding when and how people are identified provides insights into conversation dynamics, relationships, and who’s involved in decision-making or problem-solving.

Who Benefits:

• Project Managers & Team Leads: Quickly see who is mentioned, helping to map responsibilities and follow-ups.
• Sales & CX Teams: Identify key stakeholders and decision-makers mentioned in calls to inform outreach or follow-up strategies.
• Business Managers & Strategists: Understand role references and person mentions to better analyze organizational structures or customer hierarchies.
• Compliance Officers: Confirm that identifying information shared follows privacy or compliance guidelines.

Value Proposition:

• Contextual Clarity: Identifying person references aids in mapping conversation roles, relationships, and responsibilities.
• Efficiency: Structured output easily integrates into databases, dashboards, and workflows, streamlining analysis.
• Continuous Improvement: Recognizing key individuals or roles helps refine targeting, training, or compliance checks.

Data Dictionary

Objective:

Detect phrases that personally identify someone in the conversation, such as mentioning a name, a job title linked to a specific individual, or other identifying attributes. The tracker returns occurrences of matched phrases, along with timestamps and confidence.

Proposed JSON Schema:

{
  "trackers": [
    {
      "name": "Personal_Identification",
      "matches": [
        {
          "text": "<string>",
          "timestamp": "<ISO 8601 or relative time>",
          "confidence": <number> // 0.0 to 1.0
        }
      ]
    }
  ]
}

Data Dictionary:

If No Matches Found:
If no personal identifications are detected, return an empty matches array for "Personal_Identification" or omit the tracker. If no trackers match, return:

{
  "trackers": []
}

Determining Confidence & Thresholds

How the Personal_Identification Tracker Works:

• Seed the tracker with phrases or patterns that indicate person identification (e.g., using proper names, titles, or known roles).
• The system interprets these as signifiers of personal identification, possibly linking them to context (e.g., "Alice, our lead engineer").
• Assign confidence based on how closely the detected phrase matches the intent of identifying a person.

Confidence & Calibration:

• Initial Testing: Validate against calls where known individuals or roles are mentioned.
• Refine Phrases: Add or remove seed phrases or patterns to improve accuracy.
• Thresholding: Downstream consumers can filter out low-confidence matches if needed.
• Stakeholder Feedback: Involve team leads, compliance officers, or analysts to refine what constitutes meaningful person identification.

Prompt Construction & Instructions

Role Specification & Reiteration:

• Present the system as a “highly experienced assistant” identifying personal mentions.
• Reiterate instructions: output only JSON, follow the schema, no extra commentary.

No Hallucination:

• Only detect person references if clearly stated.
• If uncertain, assign lower confidence or omit the match.

Strict Formatting:

• Return only JSON with trackers.
• If none found, return empty trackers.

Example Prompt for Implementation

System Message (Role: System):
"You are a highly experienced assistant that identifies 'Personal_Identification' mentions in a conversation. You have:

• A transcript
• A 'Personal_Identification' tracker defined by seed phrases or patterns indicating a named individual or role.

Your task:

1. Detect phrases that mention or identify a person.
2. Return text, timestamp, and confidence for each match.
3. Return only JSON, no extra commentary.

Instructions & Variables (example):

{
  "tracker": {
    "name": "Personal_Identification",
    "phrases": [
      "our project manager",
      "John from accounting",
      "Alice, who leads engineering",
      "CEO Mark Johnson"
    ]
  }
}

User Message (Role: User):
"Analyze the following transcript and detect 'Personal_Identification' mentions:

[TRANSCRIPT_JSON][TRANSCRIPT_JSON]"

Expected Output Format:

{
  "trackers": [
    {
      "name": "Personal_Identification",
      "matches": [
        {
          "text": "John from accounting said we can proceed.",
          "timestamp": "2024-07-10T09:20:00Z",
          "confidence": 0.9
        }
      ]
    }
  ]
}