Linq.AI.OpenAI 1.0.1

Linq.AI.OpenAI

This library adds Linq extension methods using OpenAI structured outputs. (heaviy inspired by stevenic's agentm-js library)

Architecture

For each element in a collection an openAI API call is made to evaluate and return the result. These are done in parallel on background threads.

ChatClient

To use this you will need to instantiate a ChatClient like this:

new ChatClient(model: "gpt-4o-2024-08-06", "<OpenAIKey>");

Extensions

These extensions all take a ChatClient() as parameter and use OpenAI API to resolve.

Examples

.Classify( )

This allows you to classify each item using LLM.

enum Genres { Rock, Pop, Electronica, Country, Classical };
var classifiedItems = items.ClassifyAI<Genres>(chatClient)).ToList();

.Where()/.Remove()

This lets you filter a collection using natural language

var breadboxItems = items.Where(chatClient, "item would fit in a bread box")).ToList()
var bigItems = items.Remove(chatClient, "item would fit in a bread box")).ToList()

.Select()

This lets you transform the source into another structure using natural language.

Object transformation

You can use it to transform an object from one format to another by simply giving the types. It will use LLM to do the transformation.

var targetItems = items.SelectAI<SourceItem,TargetItem>(chatClient"))

String transformation

var markdownItems = items.Select(chatClient, "transform each item into markdown like this:\n# {{TITLE}}\n{{AUTHOR}}\n{{Description}}")

.Summarize()

Generate text summary for each object.

var summaries= items.Summarize(chatClient)

You can control the summarization with a hint

var summaries = items.Summarize(chatClient, "genreate a 3 paragraph summary");

.Answer()

This operator let's you ask a question for each item in a collection.

var answers = items.Answer(chatClient, "What is total cost of the item as a float?").Select(answer => Convert.ToFloat(answer));

Chaining operations

Linq.AI operations and normal linq operations can be mixed together.

var populations = items.SelectAI<SourceT, TargetT>(chatClient)
	.Summarize(chatClient)
	.Answer(chatClient, "What is the population?")
	.Select(answer => Convert.ToInt32(answer));