python - suggestions for resolving an ambiguity in a system - Stack Overflow

admin管理员组

文章数量:1122846

I have a huge dataset containing information about painters, divers, drivers, electricians, chefs, gardeners, and many more, including the different types of work they can do and other related conditions. Based on this data, there are corresponding prices for each type of worker, such as painters, drivers, electricians, etc.

I am using a Large Language Model (LLM) to calculate the charge for a worker based on the user's input, which specifies details such as the type of work they've done, its complexity, timing, and other relevant factors. The LLM will use these input parameters to return the total charge for that particular worker.

For example, if a user enters "a diver dives at 100m," the LLM will calculate the total charge for that diver, as the system recognizes that the diver has performed a diving task at a depth of 100m.

However, if the user provides ambiguous input, such as the word "diver" alone, it becomes impossible to calculate and return the charge for that diver because there is insufficient information. I want my system to clarify this ambiguity with the user in a friendly way to ensure accurate charge calculation.

Can you suggest how I can implement this clarification step in a user-friendly manner, so that the ambiguity is resolved and the user does not get frustrated? I would love detailed step wise response.

Note: As I mentioned earlier, the dataset is huge and includes many different types of workers, meaning various different types of ambiguities can arise. The solution should work for all these cases.

Update:
Challenges I am facing:

1. Handling large datasets: The dataset is enormous and is in human/natural language, and sending this entire data with every request to the LLM is impractical. LLMs have token limits and such an approach would be cost-prohibitive. I need a cost-effective and efficient solution.
2. Interactive communication flow: This requires a back-and-forth interaction between the server and the user. The user provides input, and the server (assisted by the LLM) responds, addressing ambiguities in the input.

Example of user-system interaction for clarity:
User: "operator"
System: "I found multiple entries with 'operator'. Options are:
JCB operator, Excavator operator, Crane operator, Truck operator."
User: "Crane operator"
System: "Work location can affect charges. Select a location: Delhi, Mumbai, Hyderabad, Pune."
User: "Hyderabad"
System: "The charge for a Crane operator in Hyderabad is 100 Rs (for example)."

This flow includes only location, but there can be different parameters for different workers. I want to achieve this type of interactive flow, possibly enhancing it for a better user experience. Suggestions needed:
How can I achieve this solution?
What technologies or approaches would best suit this use case?
If any such solution has been built before, please mention it for reference.

本文标签： pythonsuggestions for resolving an ambiguity in a systemStack Overflow