topicmodel-extractor

A repository for the "Combining DBpedia and Topic Modeling" 2016 GSoC project. It consists of three modules - spotlight-connector, topic-modeller and inference-service:

• Spotlight-connector is a collection of utilities for querying SPARQL endpoints and using DBpedia Spotlight to annotate texts. Incoming responses from the Spotlight server are parsed and can be used as Java objects.
• Topic-modeller is the core of this project. It allows for mining topics (using LDA and hLDA algorithms) from DBpedia abstracts and Wikipedia corpus.
• Inference-service is a REST service that predicts the topic distribution for arbitrary texts using previously mined topic model.

For more information about these components, please have a look in their folders.