Skip to main content

Search

This document explains how search functionality is integrated into the Next Best Message, Search & Autocomplete feature. The search process is a critical component that enhances the user experience by providing relevant response recommendations.

Overview​

During the training pipeline, all found responses are stored in an Elastic index. This index serves as a centralized repository for efficient search and retrieval of responses. The stored responses can be accessed in two primary ways:

  1. Direct Search by the User: Users can directly query the Elastic index (via Backend) to find specific responses.
  2. Autocomplete Component: The Autocomplete component leverages the Elastic index to include search results as part of the response recommendations.

Key Features​

  • Efficient Storage: The Elastic index is optimized for storing and retrieving large volumes of response data.
  • Real-Time Search: The index supports real-time search capabilities, ensuring that users receive instant results.
  • Integration with Autocomplete: The search results are seamlessly integrated into the Autocomplete component, enhancing the relevance and diversity of recommendations.

Process Summary​

  1. Training Pipeline:

    • During the training pipeline, responses are extracted and processed.
    • The processed responses are stored in the Elastic index.

  2. Direct Search:

    • Users can perform direct searches on the Elastic index using specific queries.
    • The search results are returned in real-time, providing users with immediate access to relevant responses.

  3. Autocomplete Integration:

    • The Autocomplete component queries the Elastic index to retrieve relevant search results.
    • The retrieved results are included as part of the response recommendations, ensuring that users receive comprehensive suggestions.

By leveraging Elastic, the system ensures that response recommendations are both accurate and contextually relevant, improving the overall user experience.