How to use docminer?

Purpose of DocMiner

DocMiner is a tool designed to streamline the management of chatbot response documents. It simplifies the process of organizing, uploading, and maintaining decision tables for chatbot interactions. This feature ensures that clients and customer success managers (CSMs) can efficiently manage response templates and embeddings, enabling accurate and contextually relevant responses from the chatbot.

How to Use DocMiner

Uploading a CSV

DocMiner supports uploading structured CSV files to integrate chatbot decision tables. Follow these steps to upload and manage your documents:

1. Prepare Your CSV File Ensure the CSV adheres to the following column structure:
   • motive_name: Main category of user inquiry.
   • sub_motive: Specific subcategory for refined classification.
   • type: Type of response (e.g., solution, qualification).
   • expected_answer: The exact text response the chatbot should use. This should be in markdown format for optimal behaviour of the chatbot.
   • question_examples: (Optional) Example of a user question leading to this motive.
   • folder_id: ID of the folder in Spidy where this document will reside.
2. Upload via DocMiner

   • Navigate to the DocMiner interface in edition > sources

     

   • Click on the “Add source” button.

     

   • Click on files and drag and drop your prepared file.

   • Ensure the file respects the column structure described above.

3. Testing: Wait for the end of the processing to test your chatbot on the testing page.
   • The uploaded CSV is processed into a structured document for the chatbot.
   • Each row in the CSV corresponds to a motive or sub-motive with associated metadata (e.g., Folder ID).

CSV Format

Below is an example of the expected CSV format:

Please note: each pattern must have a maximum of 10 to 15K characters (otherwise create another pattern) to ensure proper operation and recognition of embeddings.