Using a Graph Database for Deep Learning Text Classification

Graphify is an unmanaged Neo4j extension that uses natural language text classification to train deep learning models. It provides a plug-and-play solution for extracting features from text data using a graph database, allowing for the generation of vector space models and cosine similarity-based classification results. The Graphify system consists of three main endpoints: classify unlabeled text, get similar classes, and training. The classify endpoint takes in an article of text with provided labels and returns a sorted list of matches based on cosine similarity. The get similar classes endpoint takes a class name as input and returns the most similar classes ordered by their similarity scores. The training endpoint is used to train the model using repetition-based learning, where features are learned through repetition of similar phrases in text data. Graphify provides an easy-to-use API for developers to implement this technology in their applications.

No tracked trend matches for this post yet.