Dataset Owners

Choose how to provide your data, impacting its searchability:
- Utilize a vector database.
- Opt for a simple text file.
For efficient semantic search and minimal latency, process user prompts directly from apps, performing embedding and search on your server. This option allows you to choose the embedding/retrieval models you use and is adapted for large datasets with a high degree of content overlap across documents.
Alternatively, traditional keyword search remains an option.
Either way, it is important to note that it is up to you to manage the retrieval of data from your dataset and to manage the search algorithms that will select the chunks to be passed back to the app/user. It is recommended to make sure your search algorithms are fast and accurate and to choose a data chunk length adapted for RAG.
Establish APIs to grant apps or models access to your dataset.
Adhere to KIP's API standard to ensure compatibility within the KIP Ecosystem.
Consider alternative formats or implement key-based controls for restricted access within your organization.

Tokenize your dataset using the ERC-721 or ERC-3525 standard.
The token store crucial information about your dataset and access URLs, enabling discovery and connection to apps or models in the KIP Ecosystem.

Determine your profit margin and set the overall price-per-query.
Users will be charged this amount each time they access an app or model using your dataset service.

Conduct searches within your dataset based on received queries, extracting the most relevant results, and return responses to specified endpoints.
The system automatically deducts the appropriate price-per-query from the user's wallet for each query answered and accumulate your revenue in the KIP service contract.
You can claim and withdraw your earnings at any time.

Last updated 6 months ago