No open-access location is known for this record yet.
No open-access location is known for this record yet.
This research is focused on using the deep learning technique to improve patent quality classification performance. The patent quality detection is a major issue in the patent analysis field, and it can provide high information of decision making for business and industry. Patent quality can be measured by the predicting system for innovation protection, market value, technical ability, leading position and so on. Therefore, in this research, we propose a powerful feature extraction approach to extract hidden features from the raw text of a patent document using deep recurrent neural network (DRNN) model. The texts such as title and abstract are providing the rich information about a patent status and potential in future. Our proposed patent quality classification system is based on a gradient-based neural network with four functions including word embeddings, DRNN, fully connected neural network, and cross entropy loss computation. In the experimental results, our proposed patent quality classifier has obtained the best classification performance, compared with other classification models.
Loading…
Stored only when both papers are in the corpus — 6 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.