Created multiple labels automatically for documents containing at least 50,000 rows each. Grouped them into top 50 labels identified through NLP processes and Topic Modeling.
-
Preprocess the unstructured text. Follow the steps in the kg_pipeline.ipynb. Please run pip install -r requirements.txt to install all the dependencies.
-
Create Semantic Triples (Subject, Verb, Object) for each row of the data.
-
Create a TypeDB database once you've installed TypeDB on your system and started the server.
-
Create a SCHEMA for the database. Look at kgSchema.tql for example.
-
Use Python/Java/NodeJS as a client to access TypeDB API. In this case Python has been used.
-
Connect to the typedb server and transfer the data from python to typedb.
-
Install TypeDB Studio and connect to the server and database.
-
Use match $x isa thing; to see the snapshot of the entire knowledge graph. Or, open Visualize_KG.tql using TypeDB Studio.