ParallelDots Research & Development

Making a contextual recommendation engine using Python and Deep Learning at ParallelDots

When building a contextual news recommendation engine, traditional topic models do not scale. In this talk, we describe our approach of making a news recommendation engine using Deep Learning algorithms. We describe various Deep Learning algorithms and recursive neural networks which we use to make sentence representations. Then we talk about the search algorithm which can fetch contextually related results in near realtime. We also talk about the web services we wrote in Go to scale up the model.