Icon

4. Topic_​Detection-Embeddings+Neo4j

Topic Detection (Using LLM Embeddings) Get embeddings associated with given topicsfrom openai gpt embeddings Get embeddings for emails Select Plot Colours Now we can classify / search forsimilar embeddings to ourclassifier labels Provide a comma separated list ofterms/topics. We will classify the emails accordingto this list of topics. (You can change the 'Default Value'in the node settings). Neo4j Connection(Add your own credentials in the settings here) BONUS: Generating Graph in Neo4j Based on TopicsWe take the output from 'Topic Detection' so that we can include the topic-labels in the graph. We also use the 'LLM Prompter' to perform named entity recognition (NER) and relation extraction (RE).You can inspect the specific prompt in the 'String Manipulation' node.In summary, we supply the LLM with the email text, and ask it to extract the entities and their relations, where the relation is a verbal phrase (however this can be updated in future to capture other relations).This is given back to us as a structured JSON format (which I specified to the LLM how to give the response). We then parse out the JSON, and insert the data into the Neo4j database. Extracting the 'To' and 'From' fields from theheaders. This will be added to the next versionof the PST node instead. Node 3Topic SelectionNode 23Node 760Node 761Node 767Node 768Node 769Node 777Node 789Node 818Node 870Node 871Node 872Node 874Node 875Node 876Node 877Node 878Node 888Node 898Node 933Node 935Node 937Node 939Create NER promptNode 946Node 947Node 948Node 949Node 950Node 951Node 952Node 954Node 955OPENAI_API_KEY(Set 'Password' in node settings,can leave 'Username' blank)Node 986 Local File BrowserConfiguration String Widget Variable toTable Row ConstantValue Column ConstantValue Column Cell Splitter Unpivot Column Filter Column Filter String Manipulation Missing Value Similarity Search Split CollectionColumn Split CollectionColumn Column Renamer Column Filter GroupBy Column Filter Column Filter Table View(JavaScript) MISSING CounterGeneration MISSING Neo4jConnection MISSING Neo4jWriter MISSINGPython Script Column Renamer String Manipulation OpenAIAuthenticator OpenAI LLMConnector LLM Prompter Column Filter JSON Path String to JSON Ungroup JSON Path MISSING Neo4jWriter OpenAI Embeddings OpenAI Embeddings CredentialsConfiguration PST Reader Topic Detection (Using LLM Embeddings) Get embeddings associated with given topicsfrom openai gpt embeddings Get embeddings for emails Select Plot Colours Now we can classify / search forsimilar embeddings to ourclassifier labels Provide a comma separated list ofterms/topics. We will classify the emails accordingto this list of topics. (You can change the 'Default Value'in the node settings). Neo4j Connection(Add your own credentials in the settings here) BONUS: Generating Graph in Neo4j Based on TopicsWe take the output from 'Topic Detection' so that we can include the topic-labels in the graph. We also use the 'LLM Prompter' to perform named entity recognition (NER) and relation extraction (RE).You can inspect the specific prompt in the 'String Manipulation' node.In summary, we supply the LLM with the email text, and ask it to extract the entities and their relations, where the relation is a verbal phrase (however this can be updated in future to capture other relations).This is given back to us as a structured JSON format (which I specified to the LLM how to give the response). We then parse out the JSON, and insert the data into the Neo4j database. Extracting the 'To' and 'From' fields from theheaders. This will be added to the next versionof the PST node instead. Node 3Topic SelectionNode 23Node 760Node 761Node 767Node 768Node 769Node 777Node 789Node 818Node 870Node 871Node 872Node 874Node 875Node 876Node 877Node 878Node 888Node 898Node 933Node 935Node 937Node 939Create NER promptNode 946Node 947Node 948Node 949Node 950Node 951Node 952Node 954Node 955OPENAI_API_KEY(Set 'Password' in node settings,can leave 'Username' blank)Node 986 Local File BrowserConfiguration String Widget Variable toTable Row ConstantValue Column ConstantValue Column Cell Splitter Unpivot Column Filter Column Filter String Manipulation Missing Value Similarity Search Split CollectionColumn Split CollectionColumn Column Renamer Column Filter GroupBy Column Filter Column Filter Table View(JavaScript) MISSING CounterGeneration MISSING Neo4jConnection MISSING Neo4jWriter MISSINGPython Script Column Renamer String Manipulation OpenAIAuthenticator OpenAI LLMConnector LLM Prompter Column Filter JSON Path String to JSON Ungroup JSON Path MISSING Neo4jWriter OpenAI Embeddings OpenAI Embeddings CredentialsConfiguration PST Reader

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: jacobcoles
Created at: 2024-05-27
On NodePit since: 2024-05-29
Last update: 2024-11-23
Created with KNIME version: v5.2.4

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!