ETD

• ConvNets
• Deep learning
• IPC
• Patent classification

Machine learning is crucial for providing intelligent features to text analytics platforms. In this respect, deep learning techniques have gained increasing interest in the natural language processing community in the latter years. Nevertheless, consolidated statistical models are still robust, cheaper, faster and easy to apply than deep learning models. In this thesis, the problem of analyzing textual descriptions of patents is approached by a mixture of deep learning and statistical models for word embeddings and text classification, embedding them into the Mergeflow AG analytics platform