Topic extraction is very often a crucial step in Natural Language Processing, It results in lists of entities, concepts, phrases, numbers that often are the input for further processes, used for Competitive Intelligence, Social Media Analysis and Search and Content Recommendation. The items that are extracted are very diverse. Often they are also densely connected. These characteristics are in the sweet spot of graph databases, and more precisely in graph databases that use a property graph model, because these offer the possibility to add other that results from post-processing, like similarity coefficients, sentiment scores and user-ratings.
The video below illustrates how easy it is to integrate MeaningCloud’s topic extraction service with Graphileon, and how you can create a graph with a document linked to an hierarchy of concepts and entities that is automatically extended when new concepts or entities are found.