Software Developer in Montreal, QC, Canada

Abstract:
The advent of natural language processing (NLP) and deep learning allows us to achieve tasks that sounded impossible about 10 years ago. One of those tasks is genre classification for large bodies of text. Movies, books, novels, and various other texts, more often than not, belong to one or more genres. The purpose of this research is to classify those texts into their genres while also calculating the weighted presence of this genre in the aforementioned texts. Movies, in particular, are classified into genres, mostly for marketing purposes, and with no indication on which genre is the most autocratic.

In this thesis, we research various methods of classifying movies based on their script using deep neural networks. The results were obtained by training convolutional neural networks (ConvNet or CNN) and hierarchical attention networks (HAN) and compare their performance to the de-facto architectures for NLP, namely recurrent neural networks (RNN) and attention models.

The model was able to achieve 92% test accuracy and correctly labeled movies like “Godfather” to drama + action, “Dumb and Dumber” to comedy, and “Lincoln” to politics + action.

With thousands of thermostats reporting at a fast pace, we needed a solution that can handle such amount of requests and still be near real-time. The project included introducing a very light-weight microservice whose job is solely storing data into Kafka and writing a series of consumers, each with a very specific job that rendered the old monolithic method obsolete and increased the request processing time by almost 40%.

With the client constantly needing to change the structure of a certain entry in the system to match their needs, using a classical CRM database was not sufficient. In Propspace, we used a plethora of technologies to guarantee that the system would satisfy all the client's needs while performing at lightening fast speed.