Step 2: Navigate to our web application using the link: Kinetik Solutions. This will lead you to our Question Answering Recommendation Engine’s main interface.
Using the Recommendation Engine Interface
Upon accessing the site, an intuitive dashboard will greet you.
Enter your query in the provided field to begin.
Utilizing the Control Panel
Before submitting your query, refine your search parameters using the Control Panel on the left.
You can:
Utilize the 23 labels and keywords provided.
Define your question’s scope and context.
Select the desired level of detail for responses.
Important: Follow the data format guidelines in our documentation for seamless integration.
Query Submission and Obtaining Results
After setting parameters and uploading necessary data, click the ‘Ask’ button to submit your question.
The engine will process your query and display answers, data points, and insights in the ‘Answers’ section.
Creating and Exploring Scenarios
Adjust the interactive elements in the ‘Answers’ section to experiment with various data scenarios.
Observe how different inputs affect the engine’s responses.
Web Scraping and Database Generation
If you wish to scrape URLs and generate your own database, ensure you have access to ZenRows API and an AWS IAM user with S3 bucket access.
WebScraping.ipynb: Automates the scraping process. It accesses URLs from your S3 bucket, converts them to text files, extracts URLs from these files, and then scrapes these URLs for information.
RecommendationEngine.ipynb: Processes user input, performs semantic analysis to identify keywords, and extracts relevant data from the database. It then feeds this data to ClaudeV2 for response generation.
ClaudeV2Summary.ipynb: Inputs scrapped text files from the S3 bucket into ClaudeV2 to generate a database.
Installation Requirements:
Run the following commands to install necessary packages:
!pip install boto3
!pip install boto3[crt]
!pip install awscrt
Final Steps
Familiarize yourself with each Jupyter notebook and its specific role in the process.
Test the entire system to ensure that each component is functioning as expected.
Refer to the provided documentation for detailed information on each step and tool.
Setup Video
Code Access
You could access the front-end code from GitHub repository. Since our frontend uses OpenAI GPT 3.5 based on access key. It is impossible to make GitHub repository public because OpenAI will automatically disable the access key. Therefore, we have the repository private but the client has full access to all our code and this is done with his consent. To be more specific, the code about chatbot with its unit test is under the relative path /kinetik-simulation/src/Chatbot and /kinetik-simulation/src/Tests. Under the Chatbot folder, you could see a markdown called Basic_Structure_for_the_Chatbot.md which contains a basic structure introduction about our frontend and backend.
If you would like to simply view or download single without cloning the whole repository, you could try do this over our Google Drive Shared Folder. Under the shared root folder, there are bunch of model & ML file that we have successfully implemented and the failure ones due to various reason. So, we decide to put all of them there for reference. Under the folder “Chatbot”, you could view all frontend JavaScript about Chatbot