Perform Text Embedding

This handler (com.rierino.handler.langchain4j.EmbeddingEventHandler) provides ability to convert text into embedding for advanced search functionality

Handler Parameters

Parameter	Definition	Example	Default
model.store.system	Name of the system to get model details from	model_fs	A default huggingface repo
model.path	File path for the model to use	/model.onnx	A default ONNX model
tokenizer.path	File path for the tokenizer to use	/tokenizer.json	A default tokenizer
model.local.dir	Local directory to store downloaded model files	/app/models	System temp dir

This handler requires the following dependency added to deployment contents:

implementation (group:'com.rierino.custom', name: 'langchain4j', version:"${rierinoVersion}")

Actions

EmbedText

Transforms text into an embedding. Event metadata fields applicable for this action are as follows:

Field	Definition	Example	Default
inputElement	Json path for the input in request event payload	input	-
outputElement	Json path for the output in response event payload	output	-

With event metadata parameters as:

Parameter	Definition	Example	Default
inputPattern	JMESpath pattern to apply on input element for getting text and metadata fields	{text: "", metadata: ""}	-
extraAction	Optional extra action to perform after generating embedding ("add" to local index, "search" in local index)	add	-
maxResults	Maximum results to return if "search" extra action is used	5	-
minScore	Minimum similarity score required if "search" extra action is used	0.5	-

Input element can be a text value, or an object in {chat, message} format, where "chat" field includes ID of an ongoing chat for an assistant type use case with memory.