Human - Robot interaction


Conversational ChatBot with Enhanced Context

Szymon Fonau, CTO @ OctoShrew


Conventional chatbots struggle to maintain conversation context and provide accurate responses. Storing a larger context than what the neural network accepts as input size typically requires vector databases. Those struggle from problems like a lack of global context (if a question cannot be answered by an independent paragraph, they typically fail). This resents a significant challenge which is at the time of writing this article yet unsolved and our innovation currently exceeds the state of the art as per the time of writing.

OctoShrew developed an advanced chatbot that surpasses traditional conversational capabilities. By leveraging smart tagging and a proprietary hierarchical search system, the chatbot adeptly retains and retrieves contextual information. This case study explores the solution's impact on accurate question-answering, overcoming limitations of standard vector databases.

Solutions Provided by OctoShrew

OctoShrew's chatbot revolutionizes conversation understanding through the following solutions:

  1. Smart Tagging and Vector Database Storage:

    • The chatbot intelligently tags and stores relevant messages in a vector database, ensuring contextual retention for future conversations.
  2. Hierarchical Search System:

    • A proprietary approach enables the chatbot to narrow searches from chapters to individual pages and paragraphs, surpassing limitations of traditional databases.
  3. Automated Tagging:

    • Automated tagging algorithms efficiently identify and assign appropriate tags, enhancing retrieval accuracy.

Results and Benefits

OctoShrew's solution delivers exceptional results and benefits:

  1. Enhanced Contextual Understanding:

    • The chatbot retrieves a much larger context, surpassing the limitations of standard vector databases, leading to a deeper understanding of conversations.
  2. Increased Answer Accuracy:

    • By leveraging smart tagging and hierarchical search, the chatbot achieves a remarkable accuracy rate of 95%, significantly improving response quality.
  3. Improved User Experience:

    • Users enjoy precise and relevant responses, leading to a satisfying interaction with the chatbot.

Implementation and Integration

OctoShrew seamlessly integrates the chatbot within existing chatbot systems such as Rasa, allowing for much more accurate corpus-based question answering as well as long-term retention of conversations.

Future Possibilities and Expansion

Future enhancements may include advanced natural language processing techniques and integration with additional data sources like images and diagrams. The implications of this advanced contextual understanding extend beyond the current implementation.


OctoShrew's advanced chatbot elevates conversational capabilities by leveraging smart tagging, hierarchical search, and an expanded context retrieval system. The solution's exceptional accuracy rate of 95% enhances user experience, paving the way for future advancements and broader applications in various industries.


Related blogs