json-extraction

JSON Extraction Skill

This skill covers how to efficiently navigate and extract information from the enterprise data JSON files.

Data Structure

The product data files (e.g., /root/DATA/products/Product.json) typically have the following structure:

• slack: A list of channel messages. Each message has a User, text, and timestamp.
• documents: A list of document metadata, including author, date, type, and content.
• meeting_transcripts: Transcripts of meetings.
• urls: Shared URLs and their descriptions.

Techniques

Extracting Document Authors and Reviewers

1. Search the documents list for a specific type (e.g., "Market Research Report").
2. Identify the author field.
3. Search slack messages around the document's date to find discussions and feedback.
4. Users who provide specific suggestions or critiques in the Slack threads are considered reviewers.

Identifying Competitor Insights

1. Search slack messages and meeting_transcripts for keywords like "competitor", "strengths", "weaknesses", or specific competitor names (e.g., "Salesforce", "HubSpot").
2. Map the userId from those messages to identify the team members.

Finding Shared URLs

1. Search the slack messages for "http" or "demo".
2. Cross-reference with the urls list in the JSON if available.